Are we looking at a future where home computers are replaced by thin clients and all the power lies in subscription services?
‘You don't need storage space, use our cloud subscription’
‘You don’t need processing power, stream your games through our subscription service.’
Game publishers have already publicly floated the idea of not selling their games but charging per hour. Imagine how that impact Call of Duty or GTA.
Physical media could easily be killed off. Does my iPhone need 1TB of storage or will they shrink that and force everything through iCloud?
How long before car ownership is replaced with autonomous vehicle car pools? Grocery stores closed to visitors, all shopping done online and delivered to your door by drone.
I don't know how everyone arrives at that conclusion when the cost of the subscription services is also going up (as evidenced by the very article we're talking about). People who are renting are feeling this immediately, whereas people who bought their computers can wait the price hikes out for a couple years before they really need an upgrade.
Subscriptions have a "boiling frog" phenomenon where a marginal price increase isn't noticable to most people. Our payment rails are so effective many people don't even read their credit card statements, they just have vampires draining their accounts monthly.
Starting with a low subscription price also has the effect of atrophying people's ability to self-serve. The alternative to a subscription is usually capital-intensive - if you want to cancel Netflix you need to have a DVD collection. If you want to cancel your thin client you have to build a PC. Most modern consumers live on a knife edge where $20/month isn't perceptible but $1000 is a major expense.
The classic VC-backed model is to subsidize the subscription until people become complacent, and then increase the price once they're dependent. People who self-host are nutjobs because the cloud alternative is "cheaper and better" until it stops being cheaper.
My bank has an option to send me a notification every time I'm charged for something. I've noticed several bills that were higher than they should have been "due to a technical error". I'm certain some companies rely on people not checking and randomly add "errors".
Notably there's no way (known to me) that you can have direct debits sent as requests that aren't automatically paid. I think that would put consumers on an equal footing with businesses though, which is obviously bad for the economy.
It's normally an option in my experience. I have mine set for charges over $100. I don't want a notification every time I buy gas (I do check my statements every month though).
What is the harm in being notified when you buy gas? It doesn’t hurt anything, and I DO want to be notified if someone else buys gas on my card!
The discussion started as a way to avoid forgetting to cancel subscriptions or to catch subscription price increases; if you are setting your limit to $100, you aren’t going to be seeing charges for almost all your subscriptions.
I have my minimum set to $0, so I see all the charges. Helpful reminder when I see a $8 charge for something I forgot to cancel.
Alert fatigue. Most people, if they get an alert for every single purchase they make, will learn to ignore the alerts as they are useless 99% of the time. Then when an alert comes through that would be useful, they won't see that either.
Anyone who has had the misfortune to work on monitoring systems knows the very fine line you have to walk when choosing what alerts to send. Too few, or too many, and the system becomes useless.
As I said, I have my alert set to $0 and it really hasn’t caused fatigue. For one thing, when it is something i just purchased, the alert is basically just a confirmation that the purchase went through. I close it immediately and move on.
If I get an alert and I didn’t buy anything, it makes me think about it. Often times it just reminds me of a subscription I have, and I take the moment the think if I still need it or not. If I start feeling like I am getting a lot of that kind of alert, I need to reevaluate the number of subscriptions I have.
If I get an alert and I don’t immediately recognize the source (the alert will say the amount and who it is charged to), it certainly makes me pause and try to figure out what it is, and that has not been “alert fatigued” away from me even after 10+ years of these alerts.
Basically, if I get an alert when I didn’t literally JUST make a purchase, it is worth looking into.
I dont think it causes alert fatigue; I am not getting a bunch of false alerts throughout my day, because I shouldn’t be having random charges appear if I am not actively buying something.
> The alternative to a subscription is usually capital-intensive - if you want to cancel Netflix you need to have a DVD collection.
I did Apple Music and Amazon Music. The experience of losing “my” streaming library twice totally turned me off these kinds of services. Instead I do Pandora, and just buy music when I (rarely) find something I totally love and want to listen to on repeat. The inability to build a library in the streaming service that I incorrectly think of as “mine” is a big feature, keeps my mental model aligned with reality.
I do wish these services would have an easier method to import/export playlists and collections. But that would make it easier to leave, so its not going to happen.
This is something I’ve been seeing for a while. As a teen that kept his 300 dollar paycheck in cash that money would last a very long time. Now I make a good 6 figures and was seeing my accounts spending way more than I should. It wasn’t big purchases it was 50 dollars here 200 hundred there. A subscription here and there. By the end of the month I would wrack 8k in spending.
Going line by line I learned how much I neglected these transactions being the source of my problem. Could I afford it? Yes. But saving and investing is a better vehicle for retirement early than these minor dopamine hits
> if you want to cancel Netflix you need to have a DVD collection
You don't need a whole DVD collection to cancel Netflix, even ignoring piracy. Go to a cheaper streaming service, pick a free/ad supported one, go grab media from the library, etc. Grab a Blu-Ray from the discount bin at the store once in a while, and your collection will grow.
No, I do own some (actually it was more in the VHS days so tapes) and I just found that I never really watched them again. So I stopped buying movies. I'm the same with books. Once I read it, I've read it. I would rarely read a novel twice. I know what's going to happen, so what's the point? Reference books are different of course.
You're not really thinking this through enough. The exact same logic you used can be applied to music: once you've listened to the album once, you know how it will go, so what's the point of listening again? Presumably you do get something out of listening to music again (since you said you do listen to it more than once), so whatever that "something" is... you can infer that others get similar value out of rereading books/rewatching movies, even if you personally don't.
For myself, the answer is "because the story is still enjoyable even if I know how it will end". And often enough, on a second reading/viewing I will discover nuances to the work I didn't the first time. Some works are so well made that even having enjoyed it 10+ times, I can discover something new about it! So yes, the pleasure of experiencing the story the first time can only be had once. But that is by no means the only pleasure to be had.
> The exact same logic you used can be applied to music: once you've listened to the album once, you know how it will go, so what's the point of listening again?
Most music doesn't have the same kind of narrative and strong plot that stories like novels and movies do, this is a massive difference. And even if it does, it doesn't usually take a half hour or more to do such a change. That's a pretty big difference about the types of art.
I've bought a ton of movies in the past. The vast majority I've sold second hand or thrown away because I just didn't care to watch again and I didn't feel like storing something I'd never use forever.
Same goes for a lot of other media. Some amount of it I'll want to keep but most is practically disposable to me. Even most videogames.
Sure but modern cloud subscriptions have a lot of service layers you otherwise won't pay for so effectively you may be buying the hardware yearly that's a lot different than renting a media collection that would be assembled over a lifetime for the price of one new item a month.
> Subscriptions have a "boiling frog" phenomenon where a marginal price increase isn't noticable to most people.
This is so apt and well stated. It echos my sentiment, but I hadn't thought to use the boiling frog metaphor. My own organs are definitely feeling a bit toastier lately.
Difference is that if subscription goes up from $10 to $15, that doesn't seem to bad.
But if you want to purchase a new computer, and the price goes from $1000 to $1500, then that's a pretty big deal. (Though in reality, the price of said computer would probably go up even more, minimum double. RAM prices are already up 6-8 fold from summer)
Where I live, a pair of Kingston FURY Beast Black RGB DDR5 6000MHz 32GB (2x16GB) has literally gone up from what is equivalent to $125 this summer, to currently selling for what is equivalent to $850.
I think looking at the same exact product from the same retailer is not really the full story. Personally I would accept looking at the same exact spec ram across retailers in your region.
Maybe its still a lot more for you, but in the US it's not as bad as I see people say.
Realistically people normally buy whatever ram is the cheapest for the specs they want at the time of purchase, so that's the realistic cost increase IMO.
Wouldn't historical data also be inflated by the gold plated Monster branded RAM sticks too though? Making the now to then comparison, well, comparable.
The MRSP for those GPUs is already inflated. There's a reason Nvidia is going to start making more RTX 3060 GPUs. Because people (and system builders) can't afford 40XX and 50XX GPUs.
Difference is subscriptions need to support IT staff, data centers, and profit margins. A computer under your desk at home has none of those support costs and it gets price competition from used parts which subscriptions don't have.
Cloud (storage, compute, whatever) has so far consistently been more expensive than local compute over even short timeframes (storage especially, I can buy a portable 2TB drive for the equivalent of one year of the entry level 2TB dropbox plan). These shortage spikes don't seem likely to change that? Especially since the ones feeling the most pressure to pay these inflated prices are the cloud providers that are causing the demand spike in the first place. Just like with previous demand spikes, as a consumer you have alternatives such as used or waiting it out. And in the meantime you can laugh at all your geforce now buddies who just got slapped with usage restrictions and overage fees.
Subscription is still worth it for most people though. Sure it costs more, but your 2TB plan isn't a single harddrive, it is likely across several harddrives with RAID ensuring that when (not if!) they fail no data is lost, plus remote backups. When something breaks the subscription fixes that for no extra charge.
If you know how to admin a computer and have time for it, then doing it yourself is cheaper. However make sure you are comparing the real costs - not just the 2TB, but the backup system (that is tested to work), and all your time.
That said, subscriptions have all too often failed reasonable privacy standards. This is an important part of the cost that is rarely accounted for.
I’m not even sure it does cost more. I could have a geforcenow subscription for like 8 years before it’s more expensive than building a similar spec gaming rig.
Depends on the service, and timeframes. For geforcenow, you also need to consider the upgrade cycle - how often would you need to upgrade to play a newer game? I'm not sure but probably at least once within that 8 years. Buying a new car, or almost new car, and driving it until it falls apart is a better financial option than leasing. But if you want a new car every year or two, leasing is more affordable - for that scenario. Also it depends on usage. My brother in law probably plays a video game once every other month. At that point, on demand pricing (or borrowing for me) is much better than purchase or consistent subscription. You need to run the numbers.
Depends on how much you play. geforcenow is limited to 100 hours a month, with additional hours sold at a 200% premium. This dramatically changes the economics ( https://www.techpowerup.com/344359/nvidia-puts-100-hour-mont... has a handy chart for this )
I'm not sure what the value of shaming people's hobbies is. 3 hours a day is easy if it's your primary hobby, and likely double/triple that on weekends.
> Sure it costs more, but your 2TB plan isn't a single harddrive, it is likely across several harddrives with RAID ensuring that when (not if!) they fail no data is lost, plus remote backups. When something breaks the subscription fixes that for no extra charge.
Well yes, of course. And for cloud compute you get that same uptime expectation. Which if you need it is wonderful (and for something like data arguably critical for almost everyone). But if we're just talking something like a video game console? Ehhh, not so much. So no, you don't include the backup system cost just because cloud has it. You only include that cost if you want it.
But if you finance the computer (not hard to get 0% financing on consumer electronics), the price goes from $41 a month to $62 a month. It’s the same difference.
The mental model of subscriptions and financing are totally different. If I'm paying a subscription I might cancel next month, and that's a sort of freedom. If I'm financing a piece of hardware I don't want to stop paying, I want that hardware, so that's a commitment.
The difference is that with financing you're stuck with it (and your credit rating drops, at least in the EU here). You're not stuck with a subscription. If your income changes and you can't afford it anymore then you can cancel your subscription.
In the US if you don't have any debt, that is bad for your credit rating. Perversely, the more debt you have, the easier it is to get more credit, at least up to a point.
Oh sure, my original comment’s point was just to allude to the point that costs are going up for all methods of compute, so that fact alone shouldn’t influence your buy versus rent versus finance decision too much.
This idea that there’s a conspiracy to take personal computing away from the masses seems far fetched to me.
Or much longer. The computers I use most on a daily basis are over 10 years old, and still perfectly adequate for what I do. Put a non-bloated OS on them and many older computers are more than powerful enough.
There must be a breaking point. We’ve reached ours last year, when price went up again and my grandfathered plan wasn’t accepted anymore. So I talked to the missus and we cancelled our Netflix.
It had been my account for, what, a decade? A decade of not owning anything because it was affordable and convenient. Then shows started disappearing, prices went up, we could no longer use the account at her place (when we lived separately), etc. And, sadly, I’m done with them.
I think most people will eventually reach a breaking point. My sister also cancelled, which I always assumed would never happen.
> I don't know how everyone arrives at that conclusion when the cost of the subscription services is also going up
Of course they will go up, that's the whole idea. The big providers stock on hardware, front-run the hardware market, starve it for products while causing the prices to rise sharply and at that point their services are cheaper because they are selling you the hardware they bought at low prices, the one they bought in bulk, under cheap long term contracts and, in many cases, kept dark for some time.
Result - at the time of high hardware prices in retail, the cloud prices are lower, the latter increase later to make more profits, and the game can continue with the cloud providers always one step ahead of retail in a game of hoarding and scalping.
Most recently, scalping was big during the GPU shortages caused by crypto-mining. Scalpers would buy GPUs in bulk then sell them back to the starved market for a hefty margin.
Cloud providers buying up hardware at scale is basically the same, the only difference is they sell you back the services provided by the hardware, not the actual gear.
That's one common reason for renting, not the only one.
I've rented trailers and various tools before too, not because I couldn't afford to buy them, but because I knew I wouldn't need them after the fact and wouldn't know what to do with them after.
Is this true? I'm trying to think of a solid example and I'm drawing blanks.
Apartments aren't really comparable to houses. They're relatively small units which are part of a larger building. The better comparison would be to condominiums, but good luck even finding a reasonably priced condo in most parts of the US. I'd guess supply is low because there's a housing shortage and it's more profitable to rent out a unit as an apartment than to sell it as a condo.
It seems to me that most people rent because 1) they only need the thing temporarily or 2) there are no reasonable alternatives for sale.
Exactly, if you can’t afford the high upfront cost that you can stretch out over a longer period of time, you’re stuck paying more over the long term as the subscriptions get more expensive.
Because the World Economic Forum, where our political and corporate leaders meet and groom each other, point-blank advertised "you will own nothing and be happy."
For Christmas I got an alarm clock that basically doesn’t function without a $50/year subscription. For an alarm clock (Hatch.co).
Consumers need to get better at understanding TCO when buying things. Or maybe the government should be slapping those “annual cost” stickers like they do on washing machines to understand how much electricity they use.
I almost fell into this trap when buying a 'smart ring' for my spouse this year. Oura wanted $349 for the silver (!) and a subscription fee on top of that.
RingConn was cheaper, the build quality seems great, and there's no subscription fee. That made the decision a no-brainer.
I’m working on an app as part of my unemployment trajectory. It was going to be a one-time purchase thing, but almost everyone on the business advisory team thinks it’s a bad idea. I’m running out of excuses/arguments as to why I think the app should not be subscription-based, seems like bad business to do so.
An immense amount of abuse is being done to "consumers" because most people don't have a mental model for how expensive subscriptions are and end up counting them as nearly free, because they are not even tomorrow's problem, but next month's problem. Entire industries run on that principle, like the burgeoning buy now pay later space (which promise no interest but I promise you they're not making money on a whole bunch of "0% interest" loans so you do the math). Meanwhile companies love the recurring revenue. It's a perfect storm of corporate greed meeting a consumer blindspot.
Even if they "knew" they may well have not been accounting for it properly. I've been annualizing all my subscription fees for a long time now and dealing with the resulting number, but that's still an unpopular approach. Subscription fees are bleeding more people than ever dry.
Are most people really this naive? I might be "the exception", and have had this discussion with friend _who are accountants no less_ and they all consider the monthly price "cheaper" than when presented as a yearly price.
Should we start a "subscription review day": make a list of all your subscriptions, change them to yearly amounts and ask yourselves "am I getting that much value out of it"?
“I've been annualizing all my subscription fees for a long time now and dealing with the resulting number, but that's still an unpopular approach.”
This is my trick. I simply take the monthly price and multiply am by ten to quickly get a crude imperfect annual cost (adding two more months if I wanted to be exact)
Then I’ll look and go gee is this thing actually worth $150 or whatever the value is and ask “or more” assuming I wouldn’t cancel?
The answer is usually no. I’m slowly teaching this trick to my elementary school daughter.
> Are we looking at a future where home computers are replaced by thin clients and all the power lies in subscription services?
The supply chains for high-end chips are brittle enough that it's a very real possibility we end up with a severe supply crunch such that neither clouds nor individual users can access new chips at anything approaching reasonable prices.
TSMC owns 60% of the foundry market. So if China decides to invade Taiwan, that would likely mean ~60% of CPU and GPU manufacturing capacity permanently destroyed at once. That would be "iPhones are no longer for sale this year, and PCs now cost $5000 if you're lucky enough to get ahold of one" kind of territory.
> TSMC owns 60% of the foundry market. So if China decides to invade Taiwan, that would likely mean ~60% of CPU and GPU manufacturing capacity permanently destroyed at once.
While it would certainly be devastating, do note that TSMC has fabs in places that aren't Taiwan. So their entire production wouldn't immediately go offline, and presumably China would still want to keep selling those products and would have an interest in avoiding destroying those factories.
If China suddenly decides it doesn't want to export electronics, though, then we're all super fucked. After all, what percentage of those TSMC chips flow through China to get mounted onto PCBs or need major supporting components from one of the "Foxconn Cities" in China?
> presumably China would still want to keep selling those products and would have an interest in avoiding destroying those factories
There are rumours from seemingly credible sources that Taiwan has the TSMC factories (at least the ones located in Taiwan) rigged with explosives that they intend to trigger in case of invasion by China (as a disincentive against China invading). So China may well not have any say in the matter.
Presumably at least in part because China's just as dependent on TSMC as everyone else (at least for the time being). So it's a form of Mutally Assured Destruction, kind of like nuclear weapons. If they actually have to be used, then everyone's in for a bad time, but seeing as nobody wants that, it acts as a disincentive.
This makes no sense. China doesn't want Taiwan because of TSMC, it wanted Taiwan long before TSMC was a major player. The only effect of destroying factories would be to make Taiwan poorer, while China would still get what it wanted.
You're right that this isn't why China want's Taiwan. But the point is that it would also make China poorer. In fact, it would be highly likely to cause a global recession of a magnitude that could threaten the Chinese government due to pressure from it's own citizens.
It actually wouldn't. It would hit many US companies that rely on TSMC's latest nodes like Nvidia and Apple. And TSMC also has its Nanjing fab in China. What this would do is strengthen Samsung and make China even more determined to accelerate SMIC or Huawei's efforts to build a TSMC equivalent. Apple would be more impacted than Huawei, because Huawei isn't using TSMC at all. So in practice, this would hurt the US the most, since it relies on Taiwan based fabs for leading edge nodes.
You're arguing about whether it makes sense for China to invade Taiwan. I agree it doesn't right now, but that's not the topic here.
The topic is the claim that Taiwan has explosives around TSMC factories, which is ridiculous. So what exactly is your point?
PS. Rozwiń? Nie o tym jest dyskusja "czy Chiny to zrobią", tylko o tym czy Taiwan zrobił to co powtarzane tutaj bezsensowne plotki mówią. To są dwie zupełnie różne sprawy, mimo że mogą się wydawać tożsame.
If Taiwan is being invaded, the annexation is happening. There's no longer any reason to disincentivize annexation. Destroying the fabs is about denying China a major prize.
Destroying the fabs would hurt the West a lot more than China, which is rapidly playing catch up (while US and EU are not).
The other glaring flaw in this pop-geopolitics narrative is that China already has enormous economic leverage over the West, even without the chip supply chain.
> Destroying the fabs would hurt the West a lot more than China, which is rapidly playing catch up (while US and EU are not)
Is that true? My understanding is that Intel while somewhat behind TSMC, is (along with Samsung) still broadly keeping pace. Whereas SMIC while rapidly improving is still playing catch-up.
I doubt it’s something we could know without it happening.
US has intel and some other options, but it would be a colossal issue and adjustment.
China has its well funded, fast progressing Chinese chiplets, but it would be a colossal issue and adjustment.
All we can tea leaf is this: which party has a better history of making large fast industrial adjustments, and which economy is more reliant on cutting edge chips? I think china wins on both personally, so I would give them the edge, gun to head. But it’s an extremely messy process for either.
Did Hong Kong destroy its financial sector to deny China a "major prize"? If someone were going to invade and occupy your country, would you destroy your huge source of revenue so they couldn't claim it as a "major prize"? And then what? Stay poor?
I feel like people who repeat this view (something they read somewhere) haven't really analyzed it in a social, economic, historical, and geopolitical context. Because if you do, there's zero logic to it, given the consequences for the 23 million people who would still be living on the island afterward.
No one believes it, so it won't strengthen your negotiating position. It's an unconfirmed rumour of unknown origin, and nobody is taking it seriously. And you're missing the historical context, which makes TSMC irrelevant to China's claims.
Those are much better reasons. (Though I don't think the historical context cancels out a risk of losing TSMC. It just means the motivation wouldn't drop as much.)
> presumably China would still want to keep selling those products and would have an interest in avoiding destroying those factories
It has been hinted by people who might know something that Taiwan has rigged their factories to explode if China invades to ensure China can't get a hold of those factories. I'm not sure if it is true, but it wouldn't be hard to do (the hard part is ensuring the explosives don't go off for other reasons)
ASML can also disable much of the equipment remotely, from Europe. So even if the buildings aren't actually bombed (they likely would be though), someone presses a button a few thousand miles away and most of it gets bricked anyway.
I agree, I've often wondered if this is more of a dead man's switch where a signal gets transmitted every X minutes or something and if the system doesn't get a signal in some much-longer-but-not-weeks timeframe everything goes kaput.
Not to mention the number of random individuals, with enough access, who might want to sabotage them in those circumstances. And fuck knows what the Trump administration decides to bomb. And the general fog of war. And how delicate everything is.
It's a one-way journey, unless we can adapt quickly enough to a drastic reduction in the general availability of compute power. IMO, our reliance on bloated tech is an existential risk, and reverting to some reasonable baseline needs addressing as urgently as any other current crisis.
I have been waiting for quite some time for some sort of reckoning with our glut for compute resources, but for years I had optimistically assumed this would be due to physical constraints rather than artificial socioeconomic ones. Now is the most advantageous time to be a retrocomputing enthusiast as the definition of "retrocomputing" may seek to expand to engulf the whole category of home computing.
Destroyed? It's far more likely that China will find a diplomatic solution to prevent sabotage of the foundries. Supply capacity will collapse for the US, but only because it will blanket ban imports of TSMC-made products, i.e. a self-imposed supply shock.
I sometimes wonder if this is the real reason for the AI bubble. Major cloud providers are trying to front-load their computing purchases in case geopolitical trouble makes it impossible to get computers. At cloud scale, that's billions of dollars in new investment. So they kick off a new computing paradigm, hype it up, and now their cost of capital for this new investment is significantly lower because they can sell it as getting a piece of future revenues. Basically just a way to make shareholders pay for geopolitical hedging.
> Game publishers have already publicly floated the idea of not selling their games but charging per hour. Imagine how that impact Call of Duty or GTA.
MMORPGs have had monthly subscription fees for a long time.
For a lot of games if they charged by the hour would probably see less revenue...people buy tons of games and then barely ever play them.
You assume there's only one type of player. Some players fall into a category I lovingly call Madden Guy. These people will play some other games, but they will play one game *a lot*. Call of Duty, Arc Raiders, Destiny, Battlefield, Fortnite, these are the type of games who attract these kind of players. If a game has 600 purchasable items, a seasonal battlepass-type thing, multiple female rappers skins, and a publisher-financed pro scene, it's probably one of those games.
Those games 100% already have game modes you pay by the hour. They will have special modes you access with currency and you need to keep paying to keep playing. Those modes are usually special, with increased and unique drops.
Ah, but that's the genius of this circuit around the tech cycle. You need a "thick client" to access those subscription services, as running the (web) interface requires shameful amounts of RAM and CPU resources.
I think in the mid-term, many of these AI companies are going to run out of cash.
Users are more and more going to be able to run models locally so it's a race to nowhere (all you need to have a very good model, is to have a Mac with 128 GB of memory, but at 16 GB you already have something usable but not so nice)
Microsoft sells Windows 365 which is essentially cloud VMs with RDP to give to workers instead of employees running things locally.
Cloud gaming continues to grow. Nvidia GeForce Now, Xbox Cloud Gaming, PlayStation Plus, and a number of smaller companies sell remotely rendered gaming services for subscriptions.
Nobody wants to run browsers in the cloud. Too expensive. We really are getting the worst of all worlds with this, where we have to do all the expensive computations and all our subscriptions do is flip a flag in some database somewhere so that we're allowed to run a few milliseconds/month of computation on their servers while we're providing vast CPU resources to the rendering, and if the company is even half clever, WASM execution that actually costs money.
Exceptions for services that actually cost some non-trivial money per consumer, but there's a lot of crap like an alarm clock or your smart watch's subscription for fitness tracking or other completely trivial bullshit charging $10/month out there.
Valve changed its Steam Store's ToS, so officially it means Steam users no longer own the content they purchased in it, they merely bought a subscription to rent it for a while, until the game publisher decides to revoke the online content at any time they choose.
TakeTwo Interactive (2K Games, the makers of popular games series Borderlands) changed its ToS to allow it to spy on and capture anything on the user's machine, including browser behaviors (websites visited, bookmarks saved, etc.), payment information (credit card details, etc.), programs installed & usage, etc.
Gaming industry has moved on to become evil, by default. Most AAA games these days cannot even be played without an internet connection. And most AAA games demand intrusive DRMs, and same games demand the dangerous kernel-level DRMs like Denuvo which cannot be monitored by major antimalware.
That makes more sense though. Since one can't easily take a home with them (it's fixed to the ground), renting a dwelling provides flexibility and liquidity for life changing events.
Many of these people are in essentially double financialized situations, where they have a loan on the mobile home and they pay monthly pad fees to the trailer park.
At least some portion of utilities and tax charges pay for ongoing maintenance and investment to provide expected quality of life. Similar to how rent for a home eventually pays for a new roof or other repairs.
The profit margin component of rent is probably what most are referring to in this discussion, but presumably tax and (government owned) utilities don’t have that.
This context is about a profit that gets distributed to shareholders, which obviously does not benefit the payer. Whatever "profit" a government owned utility earns is presumably, eventually re-invested into the utility eventually benefiting the payer.
| Are we looking at a future where home computers are replaced by thin clients and | all the power lies in subscription services?
arguably, we are already there. The majority of people don't use a computer anymore, they use their phone. The phone itself is just an interface to the cloud. If I lost my phone today, and got a new one, the only inconvenience would be the time it takes me to set up the new phone, and the cost of the new phone. I wouldn't lose anything because everything is on the cloud
Its a market signal; we’re going to see a massive boom in investment (mostly from China, but still) in capacity expansion. Eventually the prices will stabilize.
China just needs to produce 3080s/3090 cards (which they are capable of doing given enough investment) and they can supply the gaming market for 95% of its needs
The problem I've seen described by some is that the industry is resistant to expand capacity right now, because a partial pop of the AI bubble is expected aaany time now for months, but it keeps not happening. But if it does happen and they have too much capacity, the whole GPU and RAM industry would not be able to recoup that investment and would collapse.
Western (+allied) firms should 100% be worried but they are in a difficult spot: risk losing out on market share or risk overcapacity. The Chinese have the full might of Federal/State Governments behind them which may allow them to survive overcapacity, not so for the western firms.
People confuse themselves with the bubble-metaphor. If an AI bubble exists and pops (we need not discuss either) the already existing and on-the-way-demand will not just disappear. Millions of todays users will not just decide that they don't want to use claude code or chatgpt anymore.
Instead, an increasing number of people are going to want AI stuff from here on out, forever, because it's proven to be good enough in the eyes of hundreds of millions and that will create continuous hardware demand (at least because of hardware churn, but also because there are a lot of people in the world who currently don't have great access to this technology yet).
I don't know how much optimization will drive down hardware per token, but given that most people would rather wait like 5 seconds instead of 15 minutes for answers to their coding problems, I think it's safe to assume that hardware is going to be in demand for a long time, even if, for whatever wild reason, absolutely nothing happens on top of what has already happened.
The "bubble popping" mostly means that investment will drastically fall, investors will start demanding profit, and costs will soar. This will cause a lot of tools currently built on top of LLMs to become too expensive. Free tools will likely become rare.
There's a significant number of users that will not pay for AI. There's likely also a significant number of users that will not accept higher subscription costs no matter how much they use AI tools today.
When this happens, the market will go back to "normal". Yes, there will still be a higher demand for computer parts than before ChatGPT was released, but the demand will still go down drastically from current levels. So only a moderate increase in production capacity will be needed.
AI is already easily profitable without further optimization. If at any point investors decided that this is the best the models are going to get, because it's not worth further investment then we will run inference on the existing hardware, forever. What will not happen:
- The models going away. There is no future where people will start doing more coding without AI.
- Everyone running all AI on their existing notebook or phone. We have absolutely no indication that the best models are getting smaller and cheaper to run. In fact GPUs are getting bigger.
This might hurt OpenAI, depending on how good the best available open models are at that point, but this will in no way diminish the continued increased demand for hardware.
> When this happens
I think all of this is highly unlikely, and would put a "If". But we will see!
> Millions of todays users will not just decide that they don't want to use claude code or chatgpt anymore
Won’t they? For a great number of people, LLM’s are in the “nice to have” basket. Execs and hucksters foam at the mouth over them, other people find utility but the vast majority are not upending their life in service of them.
I suspect if ChatGPT evaporated tomorrow, the chronically dependent would struggle, most people would shrug and go on with their lives, and any actual use cases would probably spin up a local model and go back to whatever they’re doing.
I’m not denying hardware demand will evaporate, it definitely won’t, but interrupt the cycle and “ehh, good enough” will probably go a very long way for a very large percentage of the userbase.
I am not sure I understand. I agree: If AI were to disappear tomorrow, people would adjust (as they did when AI, or the iPhone or the internet appeared). That's what people do.
But now there is user demand. Who or what would take away AI? What is the scenario?
Lots of companies right now have slapped AI features on their products without extra cost. Some are offering it for free (e.g. search engines). If LLM costs significantly increase, I would expect free AI features to disappear or become extra paid features (already started to happen in some SASS), while any free LLMs become simpler and cheaper.
This is making the scenario more complicated, because now we also have to consider products without product market fit, which is an entirely different issue. I don't know how to think about that.
But that's exactly why this "bubble" is so hard to predict, isn't it? The dot com bubble was simple. All companies were jumping the gun to make websites. Most were not useful, but the Internet was useful so it survived. LLMs are useful, but how much of its current use is actually valuable? If 95% is useful while 5% is memes, we're fine. But if 95% is useless, the industry will collapse hard. Who can tell what is useful when LLM companies sell the service to everyone?
Stabilize at an incredibly high price. Once they realize they can make more money selling high and producing less. I hate to say it, but the savings does not get passed on to the customer.
> You don't need storage space, use our cloud subscription
This is here already. A long time ago, maybe even before covid, I asked a table of iPhone-owning friends who pays Apple a monthly sub for storage, and every hand went up.
I know you mention home computers, but most of my friends don't have one. Their iPhone is their computer.
As sibling says, the question is too vague. I also pay for icloud, but I store all my things locally on my devices. The point is effortless sync and basic backup (protect pictures from phone theft).
I also pay Hetzner for a storage box or whatever it's called, where I regularly send backups of my stuff with restic. One of the sources is a local fat ZFS NAS, which I can access from everywhere via wireguard.
Yet, the only reason I'm contemplating buying a new iPhone is to get larger local storage. I'm also biting my fingers for not having pulled the trigger on a larger SSD for my main machine two months ago.
Every solution has different use cases, and I think no single one is perfect. I get the best value from using a mix.
You're someone who is comfortable in a terminal desktop environment on Linux. What you or I do for storage is irrelevant to the wider consumer market.
Apple's hardware prices mean that millions of people buy the smallest on offer and pay Apple monthly instead. It's a deliberate play for recurring services revenue.
My point is this: would they buy a phone that had virtually-zero free storage and rely solely on iCloud? Probably.
Some of them had 64GB models, in my view they are already doing that!
I don't know, do these small-storage-size models actually have that big a market share compared to the others?
They also sell 1 TB iPhones and I think on the latest generation the minimum storage has been increased. If nobody bought them they wouldn't sell them (see the lack of newer "mini" models).
I always thought that these models with tiny storage and tiny ram (for laptops) where just so that they could hook you with a low "starting from" price.
My point wasn't that "nobody falls into apple's trap", I'm sure plenty do. Rather, unless you're sure your audience is representative of the "wider consumer market", just asking them if they pay for icloud and they say yes, it doesn't prove much.
The smallest one is the cheapest, it's basic economics to assume that it sells the best. I would guess 60%+ buy the smallest, and I don't even need to look at data. You're welcome to disprove me.
The popularity of the storage subscription is basically why Apple is a $4tn company.
I don't know that I buy your basic economics theory. Starter prices are a thing to get people in the door and upsell them later. According to [0], the iphone 17 pro has a larger share than the regular 17, although it's more expensive.
And, again, paying for cloud storage doesn't automatically mean that they do it because their local storage is too small for their needs. Apple pushed iCloud as a solution to safely back up your stuff, and I doubt nobody bought into this angle.
I very much doubt that. I just checked the apple store page for an iphone 17. They try to sell me the larger storage model, they try to sell me apple care, but I don't see anything related to icloud. If I look at the pro, they try to sell me the max.
The difference between a 512 GB and a 256 GB non-pro model is 250 Euros. The 200 GB icloud subscription (which, again, they don't talk about when buying an iphone) costs 2.99 Euros a month. Break even is in seven years. I bet many phones don't actually last that long. If you look at a 2TB plan (which doesn't have an equivalent phone) the break-even is 2 years.
It makes no sense to try to sell the cheaper iPhone in the hope that I'll buy some icloud storage, since they actually leave money on the table. Looking at pro models, the difference between the base 256 + 2TB icloud and 2 TB model has an even longer break-even period!
So, basically, it looks actually cheaper to get a smaller phone + icloud than a bigger one.
> Services are on track to make up a quarter of Apple’s revenue but as much as 50 per cent of its profit, said JPMorgan analyst Samik Chatterjee, reflecting the “stickiness” of products such as recurring payments for iCloud storage.
Your rational take makes sense but the market disagrees. Apple cloud storage is very, very poplar.
I think we're talking past each other. I never argued icloud wasn't popular or people didn't buy it. I disagree with the reason why it's popular and with the fact that apple would push cheaper devices in the hopes of enticing people to buy icloud.
I posit many people use it for syncing / sharing stuff and easy "backup", and not mainly as a way to increase storage for entry-level devices.
You tell me how to automatically sync photos across iOS+mac devices using non-Apple services. I'll wait. (hint: it's a monopoly, you can't use other services). Yeah, I know you can do a manual backup to e.g. Google Photos. It's not the same as seamless bidirectional sync.
Will it actually do it in the background on iOS though? It's been years since I had an iPhone, but basically you had to keep the phone awake to keep the sync moving for any application that wasn't Apple's.
I think GP's point was rather the "optimized storage" schtick of apple devices, where they "unload" whatever you have locally on iCloud and use that as "regular" storage. Syncing, even though it also actually stores, is a fairly different use case. Could you implement that some other way? Sure. But I still think it's fair to point it's a different use case than "regular storage".
> Grocery stores closed to visitors, all shopping done online and delivered to your door
In the UK at least, and I'm sure in a lot of other places, a solid proportion of groceries are now delivered to the door. But, that doesn't mean that supermarkets have closed; if anything, they seem to be busier than ever.
Instead, we have a hybrid market where convenience for the consumer is the ruling factor. The same is going to be true for most of the other situations you mention.
In parts of the US, even low-crime areas, a significant amount of the items at grocery stores are locked up in glass cases. If you want them you have to track down an employee and beg for access (and in some stores they won't let you carry the items to the register). That part of the store might as well be closed to visitors, replaced by vending machines.
Hah. In my Safeway, the ice cream and half the frozen aisles have a lock on every door. I can’t imagine how much inconvenience that causes everybody. The employees openly say it’s ridiculous and you regularly find a queue in each aisle waiting to be individually served by an employee with a key unlocking and re locking each door they want something from.
From a resource allocation perspective, this doesn't feel particularly undesirable, at least in terms of certain assets like vehicles. The current system of ownership is quite wasteful. I own a high end GPU that I use maybe 4 hours a week for gaming.
Not a bad idea, though 'too recent' I think would be the only benchmark - for now, until they get faster at training. OpenAI has thrown away most of their safety team so that shouldn't be a reason anymore to delay model releases "for safety checks".
What would be too obscure or useless, when every new model boasts increasing parameter count (as if having more parameters would make the models better after a certain threshold)?
> From a resource allocation perspective, this doesn't feel particularly undesirable, at least in terms of certain assets like vehicles. The current system of ownership is quite wasteful.
People are by and large not that dumb. If they consistently choose a more expensive alternative, it means the cheaper alternative is missing something that matters.
> I own a high end GPU that I use maybe 4 hours a week for gaming.
Why? Why currently prevents you from being more efficient and streaming your display from the cloud?
I agree. However, I don't see the savings from the reduction of waste primarily going to consumers and everyone at large but being kept and collected by the owning class.
Can we please not kid ourselves with thoughts about how this being good from certain perspectives, when the development is _clearly_ bad for consumers?
I do agree with the negatives, but at the same time, I do see some upside. I live in a cycling city, and need to rent a car maybe once a year. why then should I bother myself with the annoyances of vehicle ownership?
Let me be clear here: I do not own a car and I live in a city that doesn't require car ownership.
There is a difference between choosing not to own something because it is personally more efficient or reasonable to do so, and being priced out of owning something. I don't own a car because I don't need it, I rent because I cannot afford a home.
Done charitably, I think the mainframe model of shared compute does meet most person's needs where they don't need to care about latency. It would allow us to take advantage of economies of scale. The problem, imo, is that no one has an incentive to do this as a service, so it would turn into rent-seeking.
Sure, a shared model does make sense in many ways. We could share within a family, neighbour cooperatives, and similar scales. With the users co-owning the means of processing.
But the current model is that we all rent from organisations that use their position of power to restrict and dictate what we can do with those machines.
But I do care about latency . . . and I want things to still work when the wifi is dodgy. I already find things like Office 360 deeply frustrating (only use it for work).
You do, but most people don't. So not enough people will complain that it will make any difference. And the people who don't complain will just keep forking out money because they're addicted.
Property prices are a big concern nowadays indeed, especially in high-demand areas. Complete (rich) strangers can come and outbid you.
In the most capitalist places (rich areas without rent control), you can rent a place for years trying to save money to buy in the area and see the rent grow fast enough that you can't buy and even have to leave as a renter.
Capitalism seems to work well for transportable things though, including cars. A house isn't transportable and it also tends to be something quite unique, which makes it incompatible with production in series. Even if you are somehow authorized and able to buy a cheap home, you still have the issue of the terrain, which can be more expensive than the home.
That being said I'm sure that there are people living on cheap (per square meter) terrain and happy about it, but that requires the ability to make the best of it, work on it or find work close to it.
I think the standard argument is that capitalism inevitably develops into oligarchy. Even the most basic setup of property rights protected by a legal system, presupposes a central authority of some kind who makes the ultimate decisions. Where there is law, there are courts, executive and legislative functions. More complex structures like joint stock companies introduce their own dynamics but also depend on a central authority to enforce contract law. Everything is downstream of security.
This is not the case, has been corrected a million times by millions of people, and, frankly, if it was about anything else would not be considered appropriate on HN as a comment.
We can do better than play out the same conversations also happening in middle school cafeterias. It helps everyone and could even reinforce your opposing views on the matter. You do your entire ideological position a disservice just doing this old hat!
Trust me, losing freedom you most probably consider as basic as air we breathe since your birth will stop such petty worries.
And no system would give you property just because it would be nice, neither did communism (I know since I grew up in it and saw its destruction of everything good first hand - it had to be bought for non-trivial money with good old mortgages, and only regime-aligned people could).
> neither did communism (I know since I grew up in it
Impossible. Communism is a work of science fiction, much like Star Trek which is a more modern adaptation of the same idea. Like Star Trek, the concept is dependent on post-scarcity, which we've never seen, and isn't likely to ever happen. Perhaps you mean you grew up under rule of the Communist Party?
> it had to be bought for non-trivial money with good old mortgages, and only regime-aligned people could
The defining features of communism are no class, no state, and no money. It imagines these will no longer be relevant in a post-scarcity world.
Well maybe but what examples out of real world do you mean?
Nobody is justifying anything here btw, I don't get why people hyperfocus on imperfections and claim whole thing is useless without understanding underlying reasons and thus options for fixes. Or providing long term working & proven alternatives.
> and is not capitalism and communism in opposition.
Hard to say. The prevailing assumption, and basis for the Communist Party (on paper, at least), is that capitalists will try to block reaching a state of post-scarcity — the necessary precondition for communism. This is why they are sometimes considered to be at odds with each other.
They don't have to be. And thus far they don't seem to be. Capitalism, and especially American capitalism, has done far more to getting us closer to post-scarcity than anything else, with US-centric agriculture innovation being the shining example. We're almost there in that particular area.
But we're not there yet and things can quickly turn. It is apparent in that agriculture progress that the capitalists remain deathly afraid of losing control (see the tales of Monsanto, John Deere, etc.), which is exactly the foundation on which the assumption is built.
>American capitalism, has done far more to getting us closer to post-scarcity than anything else, with US-centric agriculture innovation being the shining example
Uh, the American food industry, like nearly every first world food industry, is super state run. We stopped letting capitalism run farms because regular famine was awful.
Have you seen how much we pay per bushel of corn? Our beef is not cheaper because of capitalism. It's cheaper from enormous state subsidies that are designed to ensure we grow shitloads of certain crops regardless of economic or market factors.
But it stopped all the crazy boom-bust cycles of farming that kept ruining farms, harming farmland, and starving Americans.
Even food stamps is largely about giving farmers more state money for growing things that aren't strictly profitable.
> Uh, the American food industry, like nearly every first world food industry, is super state run.
The topic is capitalism. It only speaks to ownership. If you want to talk about who is running the show, you'd need to change the subject to command/market economies. But there is absolutely no reason to change the subject. So, getting us back on track: What agriculture-related capital do you think these first-world states own, exactly? The Canadian government used to own some grain elevators, but even that was sold off to private interests many years ago.
> Have you seen how much we pay per bushel of corn?
How could I not? What a curious question.
> It's cheaper from enormous state subsidies that are designed to ensure we grow shitloads of certain crops regardless of economic or market factors.
I have no idea what you are trying to say here. Post-scarcity, by very definition, is approached through technical innovation. There is a case to be made that subsidies have helped compel people to develop that technology, I guess, but subsidies and capitalism are in no way at odds with each other anyway. This seems to have absolutely no applicability to the conversation at hand.
Billions are already "priced out" of owning property, why do you think you joining them would have any broader impact?
EDIT: I don't know why I'm being downvoted. Billions of human beings have absolutely nothing to their names, and they have no power to change things. Because they own nothing. That's a basic fact of our globalized capitalist economy.
I guess the broader impact is cause the people on here actually do have assets and can donate, vote, call, and influence.
So if the upper middle and lower upper classes are being hollowed out...
Well... what's that leave? Just the super rich, and the rest...
And considering the way the super rich are acting, I suppose they're just fine with that. The morally bankrupt sacks of shit that most of them seem to be, or become...
I'm not sure if you're doing this intentionally or not, but it is incredibly funny to see people on HN doing what is effectively manufacturing consent for people to not actually own anything. We've gone from home computing is the future to no one should own a computer because it's wasteful.
I'm not sure where this sentiment even comes from but if the economy only consists of renters and landlords then we don't even have the thinnest veneer of capitalism anymore. We're just Feudalism 2.0.
not quite: its about accumulating capital yes, but also with private ownership (as opposed to collective ownership) and the use of said capital to gain more capital (investment returns put back into more investment)
> renting stuff is more profitable than selling stuff
maybe, maybe not, but it is usually more stable and predictable
Nope, you can have capitalism where the vast majority of the population is legally or economically barred from owning anything. The only private ownership required in capitalism is that of the means of production, which says absolutely nothing about the rest.
phones have 128GB storage and are vastly more powerful than the workstation i did my grad thesis on. Now that electron has exhausted the last major avenue for application bloat, i don't see why thin would mean anything.
> Are we looking at a future where home computers are replaced by thin clients and all the power lies in subscription services?
Always have been. Ever since the SaaS revolution of the early 2000s high-growth software businesses have been motivated to chase subscription revenue over one-time sales because you get a better multiple.
From an economic perspective The Market would like the average person to spend all their money on rents, and the only option is how you allocate your spending to different rentals. Transportation, housing, food, entertainment (which is most of computing) are just different fiefs to be carved up by industry monopolists.
Arguably a lot of non-techie personal computing has moved there.
A lot of people are phone+tablet only, no desktop or laptop. As a result they are already living in a thin-ish client with fat server for storage/app/etc from their ecosystem of choice.
Thin in form factor only! Phones and tablets represent the cutting edge of what humanity can mass-manufacture, and prices and difficulties are only going up.
“The personal computer was designed as a standalone device. There was no Internet around 1981 when the PC was invented. There weren't a lot of local area networks and corporations and schools and government agencies [online] back in 1981. The world has changed — there are networks everywhere; around the world and offices and schools and major governments and institutions. So why not have computer networks that are similar to television networks or telephone networks?
A television network is enormously complicated; it's got satellites and microwave relay stations and cable headends and recording studios, and you have this huge professionally-managed network accessed by a very low cost and simple appliance: the television.
Anyone can learn to use a television. 97% percent of American households have televisions. 94% of American households have telephones. They can have very simple appliance attached to enormously complex professionally-managed network. Why shouldn't the computer network be just the same?”
While businesses hate being a dumb pipe and love vendor lock in, lots of customers choose dependence on big tech. Each retail business that only has a Facebook page to save the cost of hiring a web developer reinforces this dependence.
I think NVidia already decided they have all the power when the decided to add 100h limit to their GForce Now service, with that limit getting effective for legacy plans right now. It would be bad if people like they service so much that they avoid buying overpriced hardware :D
More likely people will retain their current devices for longer and wait this whole thing out - there's less reason to upgrade than there was a decade ago.
Processing power increases have noticeably plateaued, with e.g. Nvidia GPUs steadily increasing in TDP to make up for there not being enough gains from updated process nodes. The RTX 5080 is rated at 67% higher TDP than the RTX 2080, but you don't see such an increase throughout most of the 2010s, so before the latter was released.
> More likely people will retain their current devices for longer and wait this whole thing out
Wait it out? The rich have realised that they can buy the entire stock of computing power and deprive the common folk of it, renting it to them instead.
In practical terms, they have infinite resources. Definitely enough to long outlast our lifetimes and the next generation and the next. We can’t “wait out” until they run out of money.
Something or someone will eventually fill the void. Chinese are trying to manufacture GPUs, they are very far from what Nvidia can offer, but they'll slowly get there. As for RAM - the new factories are being built and will hopefully become operational in 2028.
They never run out of "their" money either. Can keep borrowing until eye-watering values are breached, making out like bandits and then run to the govt for a bailout when the bubble bursts.
Computers have never been more powerful. There is less and less of a reason to do annual upgrades. I don’t follow this logic at all. The average consumer does not even need a home computer anymore. Sure if you are upgrading you gaming PC or something similar than your going to feel a bit more of a squeeze but even then it’s not a wild increase in price compared to the hours spent using.
Sorry, China is a unitary communist state. Fascism and communism are the opposite. Both have state control but the fascist control is from the corporations.
The reason China has better chips is because they have a form of socialism
Don't be a propagandist regurgitating propagandists ideas, it does nothing for intelligent conversation.
Subscriptions have always been, and will always be, more economical and efficient than ownership.
Unless you're using a resource 100% of the time, that resource is partially wasted. A GPU can be much cheaper for you if others are allowed to use it when you're asleep.
I don't think this will ever work for gaming, as gaming has strict bandwidth and latency requirements. This means you need to colocate gaming datacenters with the people who actually use them, and people in a given timezone usually play at similar times.
LLMs are a completely different beast. The user experiences of using an LLM next door and using an LLM from across the world are basically indistinguishable. This means you can have a single datacenter with really high GPU utilization during the day.
> Physical media could easily be killed off. Does my iPhone need 1TB of storage or will they shrink that and force everything through iCloud?
For the mobile space, I think we are ripe for a memory inversion of sorts, where going forward a phone has 1TB or more of ram but very little non-volatile storage. The RAM would need to be fast enough to run inferences locally, whereas the storage is does the bare minimum to boot the device on the rare occasions it requires a reboot. All user-specific artifacts would be stored in the cloud.
This seems a likely future for phones and other wearables going forward.
> Are we looking at a future where home computers are replaced by thin clients
FWIW these kinds of claims have been a cyclical thing constantly throughout my nearly 3 decades long career. We're always on the verge of thin clients replacing desktops, and then we're suddenly not and things calm down for a couple of years before, oops back again!
Every time there's arguments about saving on cost of hardware, etc. but the reality never seems to line up with the dream.
Subscriptions usually add up to the cost of buying in less than an year.
Companies hope on people being none-the-wiser, but live mostly of people that subscribe for short durations or need the internet component of the services. So, no, prices going up will make subscriptions less popular, not more.
This is different from expensive goods pools (rental cars, houses, airplanes, etc) because there's actual competition on those.
> How long before car ownership is replaced with autonomous vehicle car pools?
That is very much a future I look forward to living in. Not requiring to own a car but sharing it efficiently with folks in the neighborhood, would save quite some parking space for unused vehicles in front of homes, and centralize maintenance to the companies operating the vehicle fleet.
All the more reason to self host some part of your life at least. If there’s a market, someone will figure out how to service it, but if everyone gives up totally then all we’ll have is massive, cold, immovable corporations holding all our data.
I don't think that laying bricks and operating a crane is on the same level(or universe) of complexity as making and operating high-end semiconductor machines.
Companies are always slow to expand production capacity. It takes time to react to unforeseen shifts in the market - but once it does happen, the scale can be quite huge.
The free market as described is imaginary. Large companies due to their size hold a lot of power, and can use a mix of regulatory capture and monopolies to make better ideas financially infeasible to bring to light.
This assumes everyone’s sitting on a fast fiber connection which is hardly the case. Even metropolitan areas in most cities don’t have universal fiber connection.
Also check the agenda 2030. The European digital wallet. The cyberscore.
All those elements together makes a digital Europe (and more) where there's no cash anymore, where your website has to be compliant to be working with EU's ID and payments systems.
It's going to be all centralised and about subscriptions, no matter if it's about your electricity bill or your groceries.
> Are we looking at a future where home computers are replaced by thin clients and all the power lies in subscription services?
I've been hearing this since 2010s when Microsoft introduced UEFI iirc and I've heard it for a while. I honestly thought in my teens in the mid to late 2010s that by now I'd have a few Petabyte hard drives. Shame.
I've said it many times, everyone wants AI but nobody wants to foot the real bill for AI. The cost is too high. Once the bubble bursts, whoever is left standing might charge reasonable prices that are profitable, until it becomes more cost effective.
I think what'll happen here is that these computing price increases will be what finally makes certain ML startups un-economical. That's what will precipitate the AI bubble burst. The whole thing runs on investor capital, so it keeps going until some investors lose their capital. Once the bubble bursts, you're going to get some really cheap GPUs, RAM, and hard drives (as well as cloud computing prices), as many of the more marginal data centers go out of business and liquidate their hardware.
It's going to be a rough couple years for hobbyist computer aficionados though. In the near future I'd try to wait this one out and get a job at an AI startup instead.
Assuming you are a capitalist that is in it to maximize shareholder value then yes, that is the direction you will push the world in. Why sell me a car once if you can charge me a rent forever?
I think a better or at least adjacent question might be: do consumers want to pay full price for an object that isn’t subsidized by services? Do we actually want physical objects?
Even the things that aren’t technically subscription feel like they are. I have a Kenmore fridge I bought in 2020. The extended warranty just ran out and the thing died. I called a tech. $400 to replace a series of motors. I looked into doing it myself and it’s outside of my time or ability. I have a basement beer fridge that is admittedly less efficient but it’s still kicking and it’s from the mid 80s. I realized I’m effectively ON a subscription plan for a fridge. $900-$1,200 for five years.
How much is a smartphone that lasts (do they even) and is NOT subsidized by cloud services? I have a 128gb iPhone and though I barely use any apps I’m constantly maxing my space because I take a lot of photos.
I hate to sound like a graduate student writing a thesis on capitalism but like water flow, it just feels like companies will always default to maximum profits. Didn’t instant pot and tupper ware just go out of business because they made a product everyone needed but only once? There’s no long term profit growth in any model where we’re not sucking off the teet of some company.
Also, even if/when the AI bubble pops (whether it exists is irrelevant to my point), all the companies who have been hoarding the hardware can naturally progress into selling this new "stream everything through a subscription" service, further delaying (possibly forever if it's successful) price normalization.
> Are we looking at a future where home computers are replaced by thin clients and all the power lies in subscription services?
No we don't. In Eastern Block during communism everything was either expensive or unavailable to the point that picking up broken TV on the side of the road and dismantling it for parts was making complete sense and you could actually learn something.
Today it is much easier. You can visit eBay or AliExpress and buy old server crap from 2016 - i.e. Xeon E3 + X99 board + 32GB DDR4 RAM = 150EUR or just buying older laptops, computers from eBay. That's where consumer market is heading.
The fact that hardware is going to be computationally constrained will finally force software developers to stop wasting resources. Having application which is able to run on old Windows 10 with i3 and 4GB of RAM without being sluggish will become competing advantage.
This should be a crime and they should be forced to either keep supporting Win10 (at least security updates) and/or let you install Win11 on older hardware without TPM 2.0 modules (which is a total fucking lie that it's necessary, it's purely for monetary gain so that they can track you with a unique ID for ads).
That does not matter at all. When it comes to computation people will just use what's at hand. Windows 7 are still around, Windows 10 are going to be here for a long time.
New physical media could be killed off, but we already have a ton of old retro physical media, more than a person could ever consume in their lifetime. I wouldn’t worry about it. There are more old games and movies out there than you’ll ever be able to watch or play.
Next stage will be significant broadband/mobile price rises.
This is happening in the UK - 24 month contracts with annual 10-18% rises built in are the norm now
(The regulator worked in cahoots with the companies. A big storm was created during the high-interest rate period inventing/overstating a problem that people can't predict their bills because the rises were based on official inflation figures, which naturally varied. So the companies went: ok sure we can do fixed price rises...(At a nice high rate). This was sold as a 'win' because the rises are now 'predictable')
Plus the cheaper broadband (ADSL) is being phased out and replaced with fibre which for many people is overkill and everyone has to pay the price for that the upgrade whether you need it or not.
Three and Vodafone just merged which will mean price rises in mobile data too due to reduced competition
Had to spend ~$200/ea on refurbished EXOS for work (14TB) when I got refurbished 14TB ultra stars for $90/ea a year ago. It’s getting ridiculous. The ram I put in my gaming rig I built earlier this year was $100, now it’s $500. I want to say “it can’t stay this high forever” but who knows?
This was always the plan, it's why they keep building datacenters when they know it's a bubble, they just plan to do GeForce NOW on steroids, this time with PCs costing $10k so nobody can afford it.
And as we've seen, once the market has been captured, they will start enshittifying everything to squeeze as much profit as they can and we'll have no alternatives.
Eggs and hard currencies like gold are also going up as denominated in USD.
It’s not so much that the things are increasing in value as much as the part of large scale inflation that the federal government can’t hide by cooking the CPI.
It already has and I’ve seen the worst of it in less than 24 hours.
I signed up for a Windows Cloud PC trial, got settled into the Windows instance, then the next morning Microsoft terminated the trial for zero reason and wiped out everything in that instance.
It's got another name: inflation. Western economies are crumbling under gigantic public debt, representing 130% of the GDP in many countries and they're doubling down on public spending.
The only way out is either defaulting on the debt (like Greece partially did) or debasing the currency.
Monkeys have been left at the helm since decades and they only know how to do one thing: spending taxpayers dollars and endebtting countries ever more.
This is great news. It means the industry is expanding a lot and we'll be getting better consumer hardware at the end of the day. Innovation has always dropped down from the enterprise space to the consumer, as far back as the first electronic calculators and microcomputers.
Keeping in mind the consumer space will see minimal trickle down from used datacenter electronics in ~3-5 years from this boom.
The GPUs are generally rack-scale integrated units rather than PCIe. The bulk of the GPU RAM is HBM, so not very scavenge-able for consumer GPU mods. Power consumption of the blackwell GPUs in most solutions like the DGX B200 isn't really viable for home use even if you had the space and hookups for a fraction of the original 10ru system. The hard drives and SSDs will be likely be shredded on site and never re-sold as used. RAM will be registered ECC, only suitable for server-class motherboards.
I'm pretty sure that those racks will be usable for something, even if it's not direct-to-consumer. Startup businesses, academic/research use, smaller-scale HPC etc. will all be creating demand for the stuff long after it stops being useful for cutting-edge AI workloads.
That's fallacious: enterprise stuff doesn't always make sense at consumer level and doesn't always "drop down" for that reason. We've yet to see whether that particular stuff makes consumer sense.
Body: The change had been telegraphed: AWS's pricing page noted (and bizarrely, still does) that "current prices are scheduled to be updated in January, 2026," though the company neglected to mention which direction.
These do not seem entirely consistent?
.
> This comes about seven months after AWS trumpeted "up to 45% price reductions" for GPU instances - though that announcement covered On-Demand and Savings Plans rather than Capacity Blocks. Funny how that works.
Assuming I found the right pricing page(s), this new increased price is still lower than those other prices that were lowered.
AWS sends out targeted notifications via email and via alerts in the console if you are a current customer of a service that is changing somehow. It’s pretty normal that they don’t send out a press release and blog post every time something changes - if they did it would be a mass overload of such releases.
And if you are spending enough, a rep will probably personally reach out to you. I've never personally been blindsided by price changes, they seem actually one of the better vendors about this.
You, hopefully. I mean if you're an individual then I can see it but at work we do a monthly billing review to catch things like this. I hope you're not treating AWS entirely like a black box money pit and actually look to see where it goes.
> But the plans were on display…”
“On display? I eventually had to go down to the cellar to find them.”
“That’s the display department.”
“With a flashlight.”
“Ah, well, the lights had probably gone.”
“So had the stairs.”
“But look, you found the notice, didn’t you?”
“Yes,” said Arthur, “yes I did. It was on display in the bottom of a locked filing cabinet stuck in a disused lavatory with a sign on the door saying ‘Beware of the Leopard.
I would be very surprised if that’s true for anyone using enough GPU capacity to be reserving it in bulk. There’s so much time and attention to cost optimization and that’s going to be one of the top drivers for anyone doing AI.
The “hopes you weren't paying attention” part of the headline seems needlessly inflammatory. All I see is demand rising into a supply-limited market (GPUs and RAM). That doesn’t seem nefarious, just high-school economics.
It’s The Register. I love them but sometimes they go a bit too far with their inflammatory remarks in order to get clicks. In this case it seems like it worked based on the number of HN comments about it.
GPU and RAM supply doesn't change over the course of one weekend usually though. So there would have been no reason not to wait until Monday or do it before Saturday.
I don't think that the day of the week matters that much. But the correct way to rise prices is clearly:
- announce the new prices and the day from which on they are valid.
- ensure that your customers have time to react based on your announcement.
Based on the article it seems like the only information before the change was "some day in january we will change prices" which would not be enough for me as a customer...
They've got the analytics. The correct answer is the time of week the most customers are paying attention to their pricing and services, which probably falls within the standard 9-5 work week.
Give customers time to react and plan, so announced at least a couple weeks ahead of time.
I always found that idea funny. For most people the most important thing over the course is working days and time off days. When do they have to get up, when do they have to work, when do they need to get going early, to get to some work location, an office maybe, and when do they not have to work. Saying that "the week starts on Sunday" makes me immediately think: "Oh, you start working on Sunday?". In most people's minds in my bubble the work starts on Monday, which makes it the more natural start of the week, at least I would argue. To me personally starting the week on Sunday seems removed from most people's job and work reality.
In many calendar applications one can actually configure things like that. Often the option is directly named something like "start week on Monday/Sunday".
We also shouldn't neglect the fact, that other cultures other customs. Maybe the week starts on Sunday in your location, maybe it doesn't in my location.
True, but that’s a different issue. Further, while it can go on for a long time, as those of us who lived through the dot com bubble remember, the AI bubble will pop, because all bubbles must pop. Look for peak-hype, then sell your nvidia into that and prepare to go short.
I think a lot of businesses don’t actually need cloud AI at all. Once workloads stabilize, cloud is mostly a convenience tax. Most business use cases (docs, forecasting, monitoring, support, control systems) don’t need frontier models or hyperscale elasticity. Efficient models running locally are already “good enough”. Continuous inference + data gravity + latency/privacy constraints make owned edge hardware economically and operationally sensible again.
I've been prototyping using LLMs for some borderline use cases, and the cost isn't really the concern, it's the reliability. Using less than the most frontier model seems irresponsible if it could mean the difference between 99.95% reliability and 99% reliability, and that's the threshold where you should've hired a human to do it because you lost more money on that 0.95% error rate than you saved on salaries. (I don't actually have any use cases where this kind of calculation makes sense, but in principle I think it applies to most uses of LLMs, even if you can't quantify the harm.)
Problem is that the frontier models are nowhere near 99% reliable. Orchestration and good system design is how you get reliability. Yes, the frontier models still are going to be better by default than open source models. But the LLM is still only a component in a broader system. What's seeming to be actually necessary for any high-usage worthwhile use case is making your model task specific (via fine-tuning / post-training / RL). I build these systems for enterprises. The frontier models are not enough.
First off, you’re ignoring error bars. On average, frontier models might be 99.95% accurate. But for many work streams, there are surely tail cases where a series of questions only produce 99% accuracy (or even less), even in the frontier model case.
The challenge that businesses face is how to integrate these fallible models into reliable and repeatable business processes. That doesn’t sound so different than software engineering of yesteryear.
I suspect that as AI hype continues to level-off, business leaders will come to their senses and realize that it’s more marginally productive to spend on integration practices than squeaking out minor gains on frontier models.
Philip K. Dick and Frank Herbert saw the fundamental invariants of a capitalistic society much more accurately than we can, even with all of our claimed progress.
A few months ago, there was a lot of news lambasting tech companies for extending the depreciation lifespan of GPUs from ~3 years to ~5 years. Do these price hikes suggest a longer lifespan is probably the right way to see how long these GPUs will be valuable?
Not a finance guy, so fully prepared to be wrong here. But my interpretation is that an increase in price corresponds to a shorter lifespan. i.e. less time to make money so we need to charge more to get the same return over less time
It could also be a supply/demand issue, generally price increases are caused by either 1. demand increasing, or 2. supply decreasing.
In this case we can interpret a shorter lifespan as decreased supply, but it can also be because the demand for GPU compute has gone up. I think in this case we're seeing a bit of both, but it's hard to tell without more data.
We could also consider the supply / demand elasticity changing, f.x since demand has become more price inelastic it could result in a higher price.
The thing historically about GPUs has not been the actual lifespan of the hardware (at least half of the hardware will probably work fine for 10 or more years) the problem is that work/watt is dropping for newer hardware, so there's a point where even if you had an equivalent quantity of 10-year-old GPUs, powering them for some period costs $40k and you can buy a single brand-new GPU that costs $40k but only costs $20K to power for the same period which is less than a few years.
I don't think we're seeing any decrease in supply though, ignoring 2020 I'm pretty sure the number of GPUs manufactured has been steadily increasing. It might be the case that projected manufacturing was higher than what actually happened, which is not the same thing as a decrease in supply, but companies like Amazon will talk about it like it is, and from the standpoint of their pricing it essentially is.
> the problem is that work/watt is dropping for newer hardware, so there's a point where even if you had an equivalent quantity of 10-year-old GPUs, powering them for some period costs $40k
Sell the old-gen GPU's to on-prem users (including home consumers) who are going to run them a small % of the time (so power use is more or less negligible to them compared to acquisition cost), problem solved.
The same math applies for on-prem/home users. If you actually have some workload where it makes sense to get a free GPU that costs $40/hour to power because you only need it for a few hours a month, it's probably cheaper to rent a more efficient GPU from someone who can power it at a lower cost.
Oh my reasoning was coming at this from a different angle: H200s were released in November of 2023, so they're over 2 years old at this point while still being valuable
It mainly depends on how much NVIDIA is overselling the improvements.
With adding RL functions, separating prefill and decode chips, nvfp4 and lots of other architectural changes efficiency of the most valuable tasks goes up as long as the algorithms don't change significantly.
The people who think that high end GPUs don't last for ~5 years (really it's 6) do not know what they are talking about. 6 years is likely too low. If the cooling is good and they don't fail early, most of these GPUs could still keep going past 6 years.
I'm convinced that anything with more than 80gb of VRAM will be worth it for closer to 10 years at this point.
Is there a graph view that charts all GPU prices on one graph?
If not I think the landing page should be just that with checkbox filters for all GPUs on the left that you can easily toggle all on/off to show/hide their line on the graph.
I was not expecting that the prices are going down. Makes sense as the hardware gets older but I always assumed the prices must be inflated given how much competition there is to make new datacenters
Yes i was surprised too. I think it's mostly newer models pushing older ones down. I think there's also a lot of competitive pressure in this market. And the GPU shortage is not really a thing anymore.
Not sure, but historically, AWS as far as I know has never raised prices on specific instance type usage like this. It makes sense that this would be the first attempt since it’s for apparently guaranteed capacity (vs the normal model of “if we’re out of capacity, too bad for you”).
That said, the real disturbing part of this is not so much the raising of the price for an extremely high-demand resource, but the utter lack of communication about it.
I keep asking our product team what the plan is when our API/Infra costs eventually skyrocket and our AI features become unsustainable, but there's absolutely zero preparedness.
We throw in all these cute little AI features to fill out marketing bullet points because they're basically free, but if they had real cost we're going to have no choice but to take them away.
Recently while interviewing for a SWE position, I was talking to the company's VP of engineering and expressed my concern over these LLM tools precisely because of the economics of it, because right now most if not all of these tools are being subsidized. Also, the environmental impact.
I was talking mainly about code generation tools, which can be completely shut down today without affecting production. Not even considering LLMs that are implemented in user facing features or customer service right now.
The guy's response: "oh but these things will surely get cheaper. Every technology gets cheaper with time. And about the environment, yeah... unfortunately there's not much we can do."
That's the level of forethought for a VP of engineering where I live.
That’s the level of forethought encouraged by investors and the CEO in all likelihood. VPs are just doing their jobs. Even if they might agree with you, that sort of sentiment is a quick way to get you on the CEO / board’s bad side. The incentives are perfectly aligned to keep shoving air into this bubble for the time being.
Don't worry your users will rejoice when you rip them out. I know when Atlassian gets hit by this and has to disable their AI features their products will be a little bit less unresponsive. They'll still suck, but they'll suck a bit less
But you gotta have that AI strategy or your good numbers won't go up. You can't be left behind in this business of business. I hear everyone is already moving on to post-agentic AI, whatever that means.
What is the real AI cost at which point this is an issue for a corporation you think? I contract at a place that gives $1,000/month budget per SWE (5 Max accounts) and I am sure if the price was 10x that no one would even blink.
It is basically cost-benefit analysis just like with any other cost (and there is always option to revert to local?)
Yeah, and I doubt that this is the last price hike. But I don't think this actually has a whole lot to do with ram prices. The thing is, AI cloud providers have generally been running at a massive loss, if you do the napkin maths, the procurement cost of these machines is equivalent to approximately a year of rent at these rates, for normal servers at AWS it is approximately a month.
I think a lot of usage will move to the cheap open weight Chinese models once there is an incentive to do that. Everything below the highest end frontier models are becoming commoditized and I suspect the commodity segment will pass the "good enough" bar for most applications.
As someone who can't stand these tools, I'm ecstatic. Y'all have been hooked by the early venture-backed "scale up at all cost" period. You're gonna get squeezed so hard once investors start demanding returns on their investments -- and this whole craze is, hopefully, coming to an end.
Maybe don't hold your breath on that. Even if the price for AI tools would double, or tripple...that's still a very small part of the actual cost of an employee.
Sure, but employers will want a return on investment too. And productivity studies which aren't backed by these AI companies haven't exactly been promising. With how stingy some companies can be with the basics such as getting a laptop with enough RAM and drive space, a lot of the spending on AI tools for employees is clearly driven more by hype and the false promise of huge productivity gains than the normal expense approval process.
And plenty of people get hooked on these tools through using them for free or almost-free in their spare time. Those people will balk at huge price increases.
... and almost no company of any size above startup counts cost like that.
But I agree no point holding breath, whether somebody jumps on wagon or not won't change if price per query doubles, either its this massive productivity increase where costs of llms are a rounding error in overall costs or it isn't.
You're likely going to stay seething for a while (even if bubble pops). the product market fit of these tools is so strong that companies paying tens of thousands per employee for them isn't really that big of a deal
That's what people widely claimed about Uber: it was toast once the investor subsidies stopped. Now it's quite profitable.
People will pay more. Claude Opus 4.5 is worth more than $20 per month, as is Gemini 3 Pro. These services keep getting better. Another three years of improvement, why shouldn't that command $30 or $40 instead?
$20 is ~$10 in the year 2000 per the BLS inflation calculator (or $1.25 when priced in gold). Nobody would have thought that was expensive for such utility. These are inexpensive tools at present.
At its worst, Uber had a net margin of -~60%. The AI labs are all running at least negative triple digit net margins, some running negative quadruple digit net margins. This is why AGI has been "forecasted" to death by the labs, because investors need the promise of infinite automation to stomach the losses.
Anyway, in this instance, what you received for $20 in 2025 will run you somewhere in the range of $60-$90 in 2027/2028. In the interim, you will likely see that $30-$40 of service gets you what cost $20 in 2025. The most likely avenue for this will be reduction in subscription user limits, and for API customers premiumization through substitution. The latter being a situation where what would be the next Claude Sonnet model is now sold as Claude Opus, for example.
The only way the math works for the consumer is if the user base has become dependent on the service instead of remaining in a conventional cost/benefit relationship.
I think you're ignoring that most users of AI currently aren't paying anything, nor would they. I believe the value of a Facebook user was $70 per year in 2023, for the US and Canada. Assuming that the AI companies could make twice that from ads, that's still only $10 - $12 per month, and even less in the rest of the world. Obviously there's going to be some business users as well, so they can cover some of the cost, but would also be responsible for a larger portion of the running cost.
The question should be how many free users can the AI companies convert.
The cost of an Uber has also gone way up, and they basically have a monopoly in many areas.
The cost of free users is much lower because they are served lite models, hit quota limits quickly, and can't soak up tokens by using agents. The main capacity usage is from agent loops, which is universally behind a paid tier.
To me this all seems like downstream from massive capital expenses on GPU purchases, driving up the demand for memory, etc. It begs the question of how much of the GPU capacity that has been sold/delivered actually being utilized? Are we paying across the board for a bunch of inventory sitting idle?
What's the most cost-effective way to run open source models using cloud infrastructure? AWS? Digital Ocean? salad.com? Lambda.ai? Vast.ai? Use case is infrequent and small usage, but needs to be able to scale to match increasing demand.
They increased the price of capacity blocks, not on demand. Capacity blocks pricing was promotional with a well defined end date from the day 1. And it was even lower than spot instance lot of times.
I'm just going to say: I bet it will eventually come out that RAM prices are yet another instance of illegal cartel collusion. For which all the RAM manufacturers have been convicted. More than once! If past incidents are any indication anytime there is a perceived "hot" market the RAM makers agree to limit production to current levels allowing market demand to increase prices, followed by production cuts if the market starts to soften.
GPU is a function of limited manufacturers and vendor lock-in combined with massive capex required to compete(). Like some (but not all) price inflation during and post COVID: rising prices can become a self-fulfilling prophecy. If all your customers expect prices to rise might as well meet their expectations and get while the gettin' is good.
() Like a new CPU architecture only worse. The amount of engineering required as "table stakes" increases exponentially while at the same time the manufacturing expertise does the same. This suddenly and rapidly raises the barrier to entry. Anyone who can survive the early squeeze can do quite well in such markets.
The only real visibility we have into the demand side (kinda) are spot prices, and those seem to be flat or decreasing for GPU instances. Surely if there was this crazy, exponential demand, spot prices would reflect that.
Let's suppose that AWS is raising prices because demand is falling.
Ok, so AWS has extra capacity they need to sell but they're raising prices. Customers move off and go to another supplier. AWS has even more capacity they can't sell.
How does that make sense economically?
Do you think AWS is somehow able to make more money milking their existing customers than to sell at supply and demand equilibrium?
I wouldn't say exponential, I would also say it's likely that Amazon is projecting demand to rise faster than it is. (But it is still consistently rising, and with demand rising they would prefer to overestimate than underestimate.)
Whether it's an increase in demand, decrease in supply, or some of each is the question. Prices don't normally go up when demand drops at an over- or steady supply
Then the claim above can never be falsified because there will always be a value you can plug into the equation at which it's cheaper to rent (like at 50% occupancy, at 5%, at 0.5%... so long as you're not the only person in the world who wants to use the device being offered for loan). I would assume we're talking about a common workload where one actually needs the device that one considers purchasing for some reasonable fraction of the time and not less than half the time
- own the hardware, if you have constant load and have enough expertise (personally or a team) to use/maintain it at your required reliability level
- rent the cloud, if your usage is opportunistic/ spike’y/rapidly changing
- also rent the cloud, if required expertise/maintenance would cost you more than the hardware (if you want to have gazillion of nines, you need somebody who’s there to deal with smoking stuff, in several locations)
My guess is, the bubble popping means the demand wasn't real. No amount of price reduction will motivate it. Anybody who has to get theirs will be in for a steep premium.
I have a pair of Dell T7500 units for “slow and simple” self-hosting, and just snagged a trio of 1Tb/64Gb 2018 Mac Minis to join my own surplussed Mac Mini of the same vintage as single-purpose machines.
I have a gigabit symmetrical SOHO fibre connection. It’s time I brought everything back under one roof except for DNS.
Everything is getting monopolized (oligopolized) for rent extraction! Homes, healthcare, energy, compute. 'Capitalism' FTW! Coming up next: water, air.
Maybe it might be a good idea to squash it with any legal avenue, especially antitrust and data privacy laws that require reasonable and non discriminatory access by end consumers to self-maintained and self-hosted infra?
So rather than free market capitalism where AWS would have actually lowered prices after investing heavily in a new datacenter-focused line of Huawei GPUs, the US government took steps to stop capitalism and competition, gave NVIDIA lots of free money (USA! USA!) and now prices are going up.
Recall pictures of Bezos smiling at various Trump events.
Everything Trump admin has done so far with tech reduces competition and tries to pick winners. This price increase is just the beginning.
What kind of business are you talking about? If you're chasing a moving target, or worse, building little more than an AI Wrapper, you've willingly signed up for that risk.
They made oauth client (cognito oauth credential grant) a subscription, raising our bill X10. Just for having the client registered in the dashboard, even without using that oauth client application. we saw thousands of cost, for zero usage.
it's last call to get a GPU while there is stock left, no hopes those will become any more affordable in future considering the ongoing DRAM crisis. I am going to get 5070 Ti to replace my 1080 Ti which did a great service over all these years, paid it's worth back two times over during monero mining days and still can play a lot of games on high settings - recently finished RDR2. I sometimes wonder if Nvidia will be still making GPUs useful for gamers in a year or two or just completely shift it's focus to AI accelerators regular people will have no use for. RIP affordable computing at home.
I feel like the whole memory situation might also utterly screw up the pricing and value proposition of the upcoming Intel Arc B770.
I got a B580 cause everything else was out of my price range at the time (9060 XT 16 GB only seems to have 3 video outputs and I have no experience with daisy chaining to drive my 4 monitors and the 5060 Ti 16 GB pricing here is just sad).
I fear that companies will rise the prices during this shortage… and just never properly lower them again.
- GPU prices rising
- RAM prices rising
- hard drive prices rising
Are we looking at a future where home computers are replaced by thin clients and all the power lies in subscription services?
‘You don't need storage space, use our cloud subscription’
‘You don’t need processing power, stream your games through our subscription service.’
Game publishers have already publicly floated the idea of not selling their games but charging per hour. Imagine how that impact Call of Duty or GTA.
Physical media could easily be killed off. Does my iPhone need 1TB of storage or will they shrink that and force everything through iCloud?
How long before car ownership is replaced with autonomous vehicle car pools? Grocery stores closed to visitors, all shopping done online and delivered to your door by drone.
I don't know how everyone arrives at that conclusion when the cost of the subscription services is also going up (as evidenced by the very article we're talking about). People who are renting are feeling this immediately, whereas people who bought their computers can wait the price hikes out for a couple years before they really need an upgrade.
Subscriptions have a "boiling frog" phenomenon where a marginal price increase isn't noticable to most people. Our payment rails are so effective many people don't even read their credit card statements, they just have vampires draining their accounts monthly.
Starting with a low subscription price also has the effect of atrophying people's ability to self-serve. The alternative to a subscription is usually capital-intensive - if you want to cancel Netflix you need to have a DVD collection. If you want to cancel your thin client you have to build a PC. Most modern consumers live on a knife edge where $20/month isn't perceptible but $1000 is a major expense.
The classic VC-backed model is to subsidize the subscription until people become complacent, and then increase the price once they're dependent. People who self-host are nutjobs because the cloud alternative is "cheaper and better" until it stops being cheaper.
My bank has an option to send me a notification every time I'm charged for something. I've noticed several bills that were higher than they should have been "due to a technical error". I'm certain some companies rely on people not checking and randomly add "errors".
Notably there's no way (known to me) that you can have direct debits sent as requests that aren't automatically paid. I think that would put consumers on an equal footing with businesses though, which is obviously bad for the economy.
Wasn’t aware about charge notifications. Looks like my bank supports that - thanks for the info!
> My bank has an option to send me a notification every time I'm charged for something.
Wait, your bank doesn't do that by default? I've always assumed it's default behavior of most banks.
It's normally an option in my experience. I have mine set for charges over $100. I don't want a notification every time I buy gas (I do check my statements every month though).
What is the harm in being notified when you buy gas? It doesn’t hurt anything, and I DO want to be notified if someone else buys gas on my card!
The discussion started as a way to avoid forgetting to cancel subscriptions or to catch subscription price increases; if you are setting your limit to $100, you aren’t going to be seeing charges for almost all your subscriptions.
I have my minimum set to $0, so I see all the charges. Helpful reminder when I see a $8 charge for something I forgot to cancel.
Alert fatigue. Most people, if they get an alert for every single purchase they make, will learn to ignore the alerts as they are useless 99% of the time. Then when an alert comes through that would be useful, they won't see that either.
Anyone who has had the misfortune to work on monitoring systems knows the very fine line you have to walk when choosing what alerts to send. Too few, or too many, and the system becomes useless.
As I said, I have my alert set to $0 and it really hasn’t caused fatigue. For one thing, when it is something i just purchased, the alert is basically just a confirmation that the purchase went through. I close it immediately and move on.
If I get an alert and I didn’t buy anything, it makes me think about it. Often times it just reminds me of a subscription I have, and I take the moment the think if I still need it or not. If I start feeling like I am getting a lot of that kind of alert, I need to reevaluate the number of subscriptions I have.
If I get an alert and I don’t immediately recognize the source (the alert will say the amount and who it is charged to), it certainly makes me pause and try to figure out what it is, and that has not been “alert fatigued” away from me even after 10+ years of these alerts.
Basically, if I get an alert when I didn’t literally JUST make a purchase, it is worth looking into.
I dont think it causes alert fatigue; I am not getting a bunch of false alerts throughout my day, because I shouldn’t be having random charges appear if I am not actively buying something.
You had to opt in for my bank, and choose a minimum amount to be notified for. I chose $0, so I get notified for everything.
If it's random surely you'd get a discount sometimes!
> The alternative to a subscription is usually capital-intensive - if you want to cancel Netflix you need to have a DVD collection.
I did Apple Music and Amazon Music. The experience of losing “my” streaming library twice totally turned me off these kinds of services. Instead I do Pandora, and just buy music when I (rarely) find something I totally love and want to listen to on repeat. The inability to build a library in the streaming service that I incorrectly think of as “mine” is a big feature, keeps my mental model aligned with reality.
I do wish these services would have an easier method to import/export playlists and collections. But that would make it easier to leave, so its not going to happen.
I don’t know about Amazon but having migrated to/from both Spotify and Apple Music they are both ludicrously easy to export playlists/libraries from.
At least with Apple Music you can cmd-a cmd-c cmd-v playlists into a CSV file or something.
Yet you can't search anything unless you have a 100% exact match
This is something I’ve been seeing for a while. As a teen that kept his 300 dollar paycheck in cash that money would last a very long time. Now I make a good 6 figures and was seeing my accounts spending way more than I should. It wasn’t big purchases it was 50 dollars here 200 hundred there. A subscription here and there. By the end of the month I would wrack 8k in spending.
Going line by line I learned how much I neglected these transactions being the source of my problem. Could I afford it? Yes. But saving and investing is a better vehicle for retirement early than these minor dopamine hits
> if you want to cancel Netflix you need to have a DVD collection
You don't need a whole DVD collection to cancel Netflix, even ignoring piracy. Go to a cheaper streaming service, pick a free/ad supported one, go grab media from the library, etc. Grab a Blu-Ray from the discount bin at the store once in a while, and your collection will grow.
Music is different, but I never understood buying movies. Once I see a movie, I've seen it. I very rarely watch a movie more than once.
Maybe you would, if you bought it.
No, I do own some (actually it was more in the VHS days so tapes) and I just found that I never really watched them again. So I stopped buying movies. I'm the same with books. Once I read it, I've read it. I would rarely read a novel twice. I know what's going to happen, so what's the point? Reference books are different of course.
You're not really thinking this through enough. The exact same logic you used can be applied to music: once you've listened to the album once, you know how it will go, so what's the point of listening again? Presumably you do get something out of listening to music again (since you said you do listen to it more than once), so whatever that "something" is... you can infer that others get similar value out of rereading books/rewatching movies, even if you personally don't.
For myself, the answer is "because the story is still enjoyable even if I know how it will end". And often enough, on a second reading/viewing I will discover nuances to the work I didn't the first time. Some works are so well made that even having enjoyed it 10+ times, I can discover something new about it! So yes, the pleasure of experiencing the story the first time can only be had once. But that is by no means the only pleasure to be had.
> The exact same logic you used can be applied to music: once you've listened to the album once, you know how it will go, so what's the point of listening again?
Most music doesn't have the same kind of narrative and strong plot that stories like novels and movies do, this is a massive difference. And even if it does, it doesn't usually take a half hour or more to do such a change. That's a pretty big difference about the types of art.
I've bought a ton of movies in the past. The vast majority I've sold second hand or thrown away because I just didn't care to watch again and I didn't feel like storing something I'd never use forever.
Same goes for a lot of other media. Some amount of it I'll want to keep but most is practically disposable to me. Even most videogames.
Sure but modern cloud subscriptions have a lot of service layers you otherwise won't pay for so effectively you may be buying the hardware yearly that's a lot different than renting a media collection that would be assembled over a lifetime for the price of one new item a month.
> Subscriptions have a "boiling frog" phenomenon where a marginal price increase isn't noticable to most people.
This is so apt and well stated. It echos my sentiment, but I hadn't thought to use the boiling frog metaphor. My own organs are definitely feeling a bit toastier lately.
Difference is that if subscription goes up from $10 to $15, that doesn't seem to bad.
But if you want to purchase a new computer, and the price goes from $1000 to $1500, then that's a pretty big deal. (Though in reality, the price of said computer would probably go up even more, minimum double. RAM prices are already up 6-8 fold from summer)
Building a PC price is not double lol and RAM is nowhere near up 6-8x
https://www.bestbuy.com/product/crucial-pro-overclocking-32g...
That 32GB for $274 was not $34-$45 in the summer. RAM is up like 3x, but RAM is one of the cheaper parts of the PC.
RAM that was $100 in summer is like $300 now when I look. So that's an extra $200 maybe $300, on say a $1500 build.
GPUs are not up, they are still at MSRP:
https://www.bestbuy.com/product/asus-prime-nvidia-geforce-rt...
SSDs are up marginally, maybe $50 more lets say for a 2TB.
So from summer you are looking at like a $250-350 increase on say a $1500 PC
Where I live, a pair of Kingston FURY Beast Black RGB DDR5 6000MHz 32GB (2x16GB) has literally gone up from what is equivalent to $125 this summer, to currently selling for what is equivalent to $850.
Obviously this depends on where you live.
I think looking at the same exact product from the same retailer is not really the full story. Personally I would accept looking at the same exact spec ram across retailers in your region. Maybe its still a lot more for you, but in the US it's not as bad as I see people say.
Realistically people normally buy whatever ram is the cheapest for the specs they want at the time of purchase, so that's the realistic cost increase IMO.
Here is some proper data:
https://pcpartpicker.com/trends/price/memory/
The same site also has price trends for CPUs, video cards, etc.
It's OK data, but the average can be skewed high by some vastly inflated options that nobody cares about.
Most people will pick a ram spec and buy whatever is the cheapest kit for that spec at the time.
I think the best data view would be what is the cheapest available kit for each spec over time rather than the average price of each kit.
Wouldn't historical data also be inflated by the gold plated Monster branded RAM sticks too though? Making the now to then comparison, well, comparable.
Corsair Vengeance 128 GB (2 x 64 GB) DDR5-6400. $339 in Sept 2025. $1599 in Jan 3026. 4.7x increase. https://pcpartpicker.com/product/LPvscf/corsair-vengeance-12...
I cancelled my plans to upgrade my workstation, as the price of 256 GB of RAM became ridiculous.
The MRSP for those GPUs is already inflated. There's a reason Nvidia is going to start making more RTX 3060 GPUs. Because people (and system builders) can't afford 40XX and 50XX GPUs.
I paid $150 for a 64GB DDR5 in Jan 2025. That is today $830 representing 5.5x.
What are the specs of the kit?
Difference is subscriptions need to support IT staff, data centers, and profit margins. A computer under your desk at home has none of those support costs and it gets price competition from used parts which subscriptions don't have.
Cloud (storage, compute, whatever) has so far consistently been more expensive than local compute over even short timeframes (storage especially, I can buy a portable 2TB drive for the equivalent of one year of the entry level 2TB dropbox plan). These shortage spikes don't seem likely to change that? Especially since the ones feeling the most pressure to pay these inflated prices are the cloud providers that are causing the demand spike in the first place. Just like with previous demand spikes, as a consumer you have alternatives such as used or waiting it out. And in the meantime you can laugh at all your geforce now buddies who just got slapped with usage restrictions and overage fees.
Subscription is still worth it for most people though. Sure it costs more, but your 2TB plan isn't a single harddrive, it is likely across several harddrives with RAID ensuring that when (not if!) they fail no data is lost, plus remote backups. When something breaks the subscription fixes that for no extra charge.
If you know how to admin a computer and have time for it, then doing it yourself is cheaper. However make sure you are comparing the real costs - not just the 2TB, but the backup system (that is tested to work), and all your time.
That said, subscriptions have all too often failed reasonable privacy standards. This is an important part of the cost that is rarely accounted for.
I’m not even sure it does cost more. I could have a geforcenow subscription for like 8 years before it’s more expensive than building a similar spec gaming rig.
Depends on the service, and timeframes. For geforcenow, you also need to consider the upgrade cycle - how often would you need to upgrade to play a newer game? I'm not sure but probably at least once within that 8 years. Buying a new car, or almost new car, and driving it until it falls apart is a better financial option than leasing. But if you want a new car every year or two, leasing is more affordable - for that scenario. Also it depends on usage. My brother in law probably plays a video game once every other month. At that point, on demand pricing (or borrowing for me) is much better than purchase or consistent subscription. You need to run the numbers.
Depends on how much you play. geforcenow is limited to 100 hours a month, with additional hours sold at a 200% premium. This dramatically changes the economics ( https://www.techpowerup.com/344359/nvidia-puts-100-hour-mont... has a handy chart for this )
Yikes. If you're playing more than 100 hours a month then it's time to touch grass.
I'm not sure what the value of shaming people's hobbies is. 3 hours a day is easy if it's your primary hobby, and likely double/triple that on weekends.
The business model of Facebook, YouTube, traditional TV, app stores, and yes video games depend on people not touching grass.
If you live alone by yourself such that the concept of a shared gaming system is entirely unimaginable, maybe you're the one that needs to touch grass
> Sure it costs more, but your 2TB plan isn't a single harddrive, it is likely across several harddrives with RAID ensuring that when (not if!) they fail no data is lost, plus remote backups. When something breaks the subscription fixes that for no extra charge.
Well yes, of course. And for cloud compute you get that same uptime expectation. Which if you need it is wonderful (and for something like data arguably critical for almost everyone). But if we're just talking something like a video game console? Ehhh, not so much. So no, you don't include the backup system cost just because cloud has it. You only include that cost if you want it.
> ...that doesn't seem too bad.
Yep, "seem". But the reality is more like 3 different subscriptions going up by $5/month, and the new computer is a once-in-4-years purchase:
$5/month * 3 subscriptions * 48 months = $720.00
And no bets on those subscriptions being up to $20 or so by the end of year 4.
But if you finance the computer (not hard to get 0% financing on consumer electronics), the price goes from $41 a month to $62 a month. It’s the same difference.
For whatever reason, people are just more into paying for $10 / subscription fees, than they into financing stuff.
To such a degree that they able to pay for a bunch of subscriptions they completely forget about.
The mental model of subscriptions and financing are totally different. If I'm paying a subscription I might cancel next month, and that's a sort of freedom. If I'm financing a piece of hardware I don't want to stop paying, I want that hardware, so that's a commitment.
The difference is that with financing you're stuck with it (and your credit rating drops, at least in the EU here). You're not stuck with a subscription. If your income changes and you can't afford it anymore then you can cancel your subscription.
In the US if you don't have any debt, that is bad for your credit rating. Perversely, the more debt you have, the easier it is to get more credit, at least up to a point.
Oh sure, my original comment’s point was just to allude to the point that costs are going up for all methods of compute, so that fact alone shouldn’t influence your buy versus rent versus finance decision too much.
This idea that there’s a conspiracy to take personal computing away from the masses seems far fetched to me.
> people are just more into paying for $10 / subscription fees, than they into financing stuff
I'm not so sure, seeing the explosion of Buy Now Pay Later (BNPL) platforms.
The common thread is that people are bad and saving and delayed gratification. The easiest path to instant gratification wins more often than not.
For some reason I think people are less likely to finance a computer, but maybe not, given that phone financing is a thing.
Just go to the Apple website and notice how every computer price has a monthly payment option in the same font size right under the purchase price.
Numerous places are advertising financing groceries and eating out these days.
https://www.paypal.com/us/digital-wallet/ways-to-pay/buy-now...
https://www.klarna.com/us/store/5130c1b0-9c21-4870-b7ed-b610...
> wait the price hikes out for a couple years
Or much longer. The computers I use most on a daily basis are over 10 years old, and still perfectly adequate for what I do. Put a non-bloated OS on them and many older computers are more than powerful enough.
And Linux being good for gaming now (thanks Valve!) makes this option more attractive than ever.
And the inability to run rootkits is a bonus, not a drawback.
Netflix says hello lol
Every increasing prices
Password sharing forbidden
Etc etc
And still making more and more money.
People are willing to take a beating if they are entertained and pay a lot more
There must be a breaking point. We’ve reached ours last year, when price went up again and my grandfathered plan wasn’t accepted anymore. So I talked to the missus and we cancelled our Netflix.
It had been my account for, what, a decade? A decade of not owning anything because it was affordable and convenient. Then shows started disappearing, prices went up, we could no longer use the account at her place (when we lived separately), etc. And, sadly, I’m done with them.
I think most people will eventually reach a breaking point. My sister also cancelled, which I always assumed would never happen.
> I don't know how everyone arrives at that conclusion when the cost of the subscription services is also going up
Of course they will go up, that's the whole idea. The big providers stock on hardware, front-run the hardware market, starve it for products while causing the prices to rise sharply and at that point their services are cheaper because they are selling you the hardware they bought at low prices, the one they bought in bulk, under cheap long term contracts and, in many cases, kept dark for some time.
Result - at the time of high hardware prices in retail, the cloud prices are lower, the latter increase later to make more profits, and the game can continue with the cloud providers always one step ahead of retail in a game of hoarding and scalping.
Most recently, scalping was big during the GPU shortages caused by crypto-mining. Scalpers would buy GPUs in bulk then sell them back to the starved market for a hefty margin.
Cloud providers buying up hardware at scale is basically the same, the only difference is they sell you back the services provided by the hardware, not the actual gear.
People rent things they can’t afford to own.
That's one common reason for renting, not the only one.
I've rented trailers and various tools before too, not because I couldn't afford to buy them, but because I knew I wouldn't need them after the fact and wouldn't know what to do with them after.
You could never afford a hyperscaler size data center. You could always just buy a server and stick it into Colo. it’s just not totally the same thing
I can't afford the ram and storage in the server anymore though. So it kinda is the same thing.
Rent vs own isn’t the issue then. It’s the cost of the underlying asset.
I also couldn't afford to rent one.
I can afford to rent fractional use of one, but by that token I could also afford to buy a very small fraction of one too.
or just not ready for commitment (try first, buy later, etc...)
Is this true? I'm trying to think of a solid example and I'm drawing blanks.
Apartments aren't really comparable to houses. They're relatively small units which are part of a larger building. The better comparison would be to condominiums, but good luck even finding a reasonably priced condo in most parts of the US. I'd guess supply is low because there's a housing shortage and it's more profitable to rent out a unit as an apartment than to sell it as a condo.
It seems to me that most people rent because 1) they only need the thing temporarily or 2) there are no reasonable alternatives for sale.
Check out the furniture rental industry. Car rental (lease) industry. Electronics rental industry.
Lots of people rent houses, it’s not just apartments.
And even ~40% of so-called homeowners rent money in order to be there.
Exactly, if you can’t afford the high upfront cost that you can stretch out over a longer period of time, you’re stuck paying more over the long term as the subscriptions get more expensive.
Because the World Economic Forum, where our political and corporate leaders meet and groom each other, point-blank advertised "you will own nothing and be happy."
We’ve been raised to believe “experiences make you happy, not things.”
Everything as a service is the modern marketing ideal.
For Christmas I got an alarm clock that basically doesn’t function without a $50/year subscription. For an alarm clock (Hatch.co).
Consumers need to get better at understanding TCO when buying things. Or maybe the government should be slapping those “annual cost” stickers like they do on washing machines to understand how much electricity they use.
A 50$/year alarm clock?!
Is it time to post the philip k dick Ubik quote again?!
https://hn.algolia.com/?dateRange=all&page=0&prefix=true&que...
Yes.
I remember starting this book back in college and rolling my eyes at this scene for being so campy and tacky that I just dropped the book.
I came back to it just a few weeks ago out of disbelief that that is where we've arrived today.
Paying a subscription for an alarm clock. I've heard this one before! [1][2]
[1] https://en.wikipedia.org/wiki/Knocker-up [2] https://www.bbc.com/news/uk-england-35840393
I almost fell into this trap when buying a 'smart ring' for my spouse this year. Oura wanted $349 for the silver (!) and a subscription fee on top of that.
RingConn was cheaper, the build quality seems great, and there's no subscription fee. That made the decision a no-brainer.
I’m working on an app as part of my unemployment trajectory. It was going to be a one-time purchase thing, but almost everyone on the business advisory team thinks it’s a bad idea. I’m running out of excuses/arguments as to why I think the app should not be subscription-based, seems like bad business to do so.
I would never buy this product but it seems pretty clear on the website the sub is highly integrated and most likely required.
TIL. Business model: Juicero for alarms. Great paired with your EightSleep brand cloud-enabled bed.
I set up one of these yesterday, and it was an infuriating experience.
That's a terrible gift! Did the g(r)ifter know it needed such a subscription?
An immense amount of abuse is being done to "consumers" because most people don't have a mental model for how expensive subscriptions are and end up counting them as nearly free, because they are not even tomorrow's problem, but next month's problem. Entire industries run on that principle, like the burgeoning buy now pay later space (which promise no interest but I promise you they're not making money on a whole bunch of "0% interest" loans so you do the math). Meanwhile companies love the recurring revenue. It's a perfect storm of corporate greed meeting a consumer blindspot.
Even if they "knew" they may well have not been accounting for it properly. I've been annualizing all my subscription fees for a long time now and dealing with the resulting number, but that's still an unpopular approach. Subscription fees are bleeding more people than ever dry.
Are most people really this naive? I might be "the exception", and have had this discussion with friend _who are accountants no less_ and they all consider the monthly price "cheaper" than when presented as a yearly price.
Should we start a "subscription review day": make a list of all your subscriptions, change them to yearly amounts and ask yourselves "am I getting that much value out of it"?
“I've been annualizing all my subscription fees for a long time now and dealing with the resulting number, but that's still an unpopular approach.”
This is my trick. I simply take the monthly price and multiply am by ten to quickly get a crude imperfect annual cost (adding two more months if I wanted to be exact)
Then I’ll look and go gee is this thing actually worth $150 or whatever the value is and ask “or more” assuming I wouldn’t cancel?
The answer is usually no. I’m slowly teaching this trick to my elementary school daughter.
A year helps but that's still really underselling the cost. For most things I'll look at 5 years.
Good thing you can fund it with your HSA or FSA in partnership with Truemed! \s
> Are we looking at a future where home computers are replaced by thin clients and all the power lies in subscription services?
The supply chains for high-end chips are brittle enough that it's a very real possibility we end up with a severe supply crunch such that neither clouds nor individual users can access new chips at anything approaching reasonable prices.
TSMC owns 60% of the foundry market. So if China decides to invade Taiwan, that would likely mean ~60% of CPU and GPU manufacturing capacity permanently destroyed at once. That would be "iPhones are no longer for sale this year, and PCs now cost $5000 if you're lucky enough to get ahold of one" kind of territory.
> TSMC owns 60% of the foundry market. So if China decides to invade Taiwan, that would likely mean ~60% of CPU and GPU manufacturing capacity permanently destroyed at once.
While it would certainly be devastating, do note that TSMC has fabs in places that aren't Taiwan. So their entire production wouldn't immediately go offline, and presumably China would still want to keep selling those products and would have an interest in avoiding destroying those factories.
If China suddenly decides it doesn't want to export electronics, though, then we're all super fucked. After all, what percentage of those TSMC chips flow through China to get mounted onto PCBs or need major supporting components from one of the "Foxconn Cities" in China?
> presumably China would still want to keep selling those products and would have an interest in avoiding destroying those factories
There are rumours from seemingly credible sources that Taiwan has the TSMC factories (at least the ones located in Taiwan) rigged with explosives that they intend to trigger in case of invasion by China (as a disincentive against China invading). So China may well not have any say in the matter.
Why would Taiwan destroying its own assets disincentivize annexation goals that have existed long before computer chips were produced there?
Presumably at least in part because China's just as dependent on TSMC as everyone else (at least for the time being). So it's a form of Mutally Assured Destruction, kind of like nuclear weapons. If they actually have to be used, then everyone's in for a bad time, but seeing as nobody wants that, it acts as a disincentive.
This makes no sense. China doesn't want Taiwan because of TSMC, it wanted Taiwan long before TSMC was a major player. The only effect of destroying factories would be to make Taiwan poorer, while China would still get what it wanted.
You're right that this isn't why China want's Taiwan. But the point is that it would also make China poorer. In fact, it would be highly likely to cause a global recession of a magnitude that could threaten the Chinese government due to pressure from it's own citizens.
It actually wouldn't. It would hit many US companies that rely on TSMC's latest nodes like Nvidia and Apple. And TSMC also has its Nanjing fab in China. What this would do is strengthen Samsung and make China even more determined to accelerate SMIC or Huawei's efforts to build a TSMC equivalent. Apple would be more impacted than Huawei, because Huawei isn't using TSMC at all. So in practice, this would hurt the US the most, since it relies on Taiwan based fabs for leading edge nodes.
Not invading Taiwan:
+ China gets to profit from selling phones, computers, etc. to the west
- China doesn't get to own a piece of land
Invading Taiwan:
+ China owns a piece of land
- China can't manufacture anything the west is interested in.
You're arguing about whether it makes sense for China to invade Taiwan. I agree it doesn't right now, but that's not the topic here. The topic is the claim that Taiwan has explosives around TSMC factories, which is ridiculous. So what exactly is your point?
PS. Rozwiń? Nie o tym jest dyskusja "czy Chiny to zrobią", tylko o tym czy Taiwan zrobił to co powtarzane tutaj bezsensowne plotki mówią. To są dwie zupełnie różne sprawy, mimo że mogą się wydawać tożsame.
If Taiwan is being invaded, the annexation is happening. There's no longer any reason to disincentivize annexation. Destroying the fabs is about denying China a major prize.
Destroying the fabs would hurt the West a lot more than China, which is rapidly playing catch up (while US and EU are not).
The other glaring flaw in this pop-geopolitics narrative is that China already has enormous economic leverage over the West, even without the chip supply chain.
> Destroying the fabs would hurt the West a lot more than China, which is rapidly playing catch up (while US and EU are not)
Is that true? My understanding is that Intel while somewhat behind TSMC, is (along with Samsung) still broadly keeping pace. Whereas SMIC while rapidly improving is still playing catch-up.
I doubt it’s something we could know without it happening.
US has intel and some other options, but it would be a colossal issue and adjustment.
China has its well funded, fast progressing Chinese chiplets, but it would be a colossal issue and adjustment.
All we can tea leaf is this: which party has a better history of making large fast industrial adjustments, and which economy is more reliant on cutting edge chips? I think china wins on both personally, so I would give them the edge, gun to head. But it’s an extremely messy process for either.
Did Hong Kong destroy its financial sector to deny China a "major prize"? If someone were going to invade and occupy your country, would you destroy your huge source of revenue so they couldn't claim it as a "major prize"? And then what? Stay poor? I feel like people who repeat this view (something they read somewhere) haven't really analyzed it in a social, economic, historical, and geopolitical context. Because if you do, there's zero logic to it, given the consequences for the 23 million people who would still be living on the island afterward.
Committing to threats/promises "illogically" gives you a better negotiating position.
Acting "illogically" to spite bad behavior leads to less bad behavior.
No one believes it, so it won't strengthen your negotiating position. It's an unconfirmed rumour of unknown origin, and nobody is taking it seriously. And you're missing the historical context, which makes TSMC irrelevant to China's claims.
Those are much better reasons. (Though I don't think the historical context cancels out a risk of losing TSMC. It just means the motivation wouldn't drop as much.)
That question comes up every time this fact is posted, and it could be for the very simple reason to disincentivize annexation to a later date.
> presumably China would still want to keep selling those products and would have an interest in avoiding destroying those factories
It has been hinted by people who might know something that Taiwan has rigged their factories to explode if China invades to ensure China can't get a hold of those factories. I'm not sure if it is true, but it wouldn't be hard to do (the hard part is ensuring the explosives don't go off for other reasons)
ASML can also disable much of the equipment remotely, from Europe. So even if the buildings aren't actually bombed (they likely would be though), someone presses a button a few thousand miles away and most of it gets bricked anyway.
Dunno if I would trust any remote-bricking-capability in an environment where electronic warfare saturates all communication channels.
I agree, I've often wondered if this is more of a dead man's switch where a signal gets transmitted every X minutes or something and if the system doesn't get a signal in some much-longer-but-not-weeks timeframe everything goes kaput.
Not to mention the number of random individuals, with enough access, who might want to sabotage them in those circumstances. And fuck knows what the Trump administration decides to bomb. And the general fog of war. And how delicate everything is.
These supply chains have very non-linear responses. Relatively small fluctuations in demand can have enormous effects on prices and leadtimes.
It's a one-way journey, unless we can adapt quickly enough to a drastic reduction in the general availability of compute power. IMO, our reliance on bloated tech is an existential risk, and reverting to some reasonable baseline needs addressing as urgently as any other current crisis.
I have been waiting for quite some time for some sort of reckoning with our glut for compute resources, but for years I had optimistically assumed this would be due to physical constraints rather than artificial socioeconomic ones. Now is the most advantageous time to be a retrocomputing enthusiast as the definition of "retrocomputing" may seek to expand to engulf the whole category of home computing.
Destroyed? It's far more likely that China will find a diplomatic solution to prevent sabotage of the foundries. Supply capacity will collapse for the US, but only because it will blanket ban imports of TSMC-made products, i.e. a self-imposed supply shock.
I sometimes wonder if this is the real reason for the AI bubble. Major cloud providers are trying to front-load their computing purchases in case geopolitical trouble makes it impossible to get computers. At cloud scale, that's billions of dollars in new investment. So they kick off a new computing paradigm, hype it up, and now their cost of capital for this new investment is significantly lower because they can sell it as getting a piece of future revenues. Basically just a way to make shareholders pay for geopolitical hedging.
> Game publishers have already publicly floated the idea of not selling their games but charging per hour. Imagine how that impact Call of Duty or GTA.
MMORPGs have had monthly subscription fees for a long time.
For a lot of games if they charged by the hour would probably see less revenue...people buy tons of games and then barely ever play them.
You assume there's only one type of player. Some players fall into a category I lovingly call Madden Guy. These people will play some other games, but they will play one game *a lot*. Call of Duty, Arc Raiders, Destiny, Battlefield, Fortnite, these are the type of games who attract these kind of players. If a game has 600 purchasable items, a seasonal battlepass-type thing, multiple female rappers skins, and a publisher-financed pro scene, it's probably one of those games.
Those games 100% already have game modes you pay by the hour. They will have special modes you access with currency and you need to keep paying to keep playing. Those modes are usually special, with increased and unique drops.
This has been theorized as Technofeudalism: https://simple.wikipedia.org/wiki/Technofeudalism
>home computers are replaced by thin clients
Ah, but that's the genius of this circuit around the tech cycle. You need a "thick client" to access those subscription services, as running the (web) interface requires shameful amounts of RAM and CPU resources.
I think in the mid-term, many of these AI companies are going to run out of cash.
Users are more and more going to be able to run models locally so it's a race to nowhere (all you need to have a very good model, is to have a Mac with 128 GB of memory, but at 16 GB you already have something usable but not so nice)
The big winners are going to be the memory makers
Not if the browser rendering is done in the cloud. All you're looking at is a screen on a server somewhere, not actually the one on your phone.
RDP is the future? Makes sense. It's been a while since money was poured into networking.
Microsoft sells Windows 365 which is essentially cloud VMs with RDP to give to workers instead of employees running things locally.
Cloud gaming continues to grow. Nvidia GeForce Now, Xbox Cloud Gaming, PlayStation Plus, and a number of smaller companies sell remotely rendered gaming services for subscriptions.
That was Opera Mini's schtick. What's old is new again!
Nobody wants to run browsers in the cloud. Too expensive. We really are getting the worst of all worlds with this, where we have to do all the expensive computations and all our subscriptions do is flip a flag in some database somewhere so that we're allowed to run a few milliseconds/month of computation on their servers while we're providing vast CPU resources to the rendering, and if the company is even half clever, WASM execution that actually costs money.
Exceptions for services that actually cost some non-trivial money per consumer, but there's a lot of crap like an alarm clock or your smart watch's subscription for fitness tracking or other completely trivial bullshit charging $10/month out there.
Maybe it will incentivize the introduction of more reasonable browser technology. /s
Valve changed its Steam Store's ToS, so officially it means Steam users no longer own the content they purchased in it, they merely bought a subscription to rent it for a while, until the game publisher decides to revoke the online content at any time they choose.
TakeTwo Interactive (2K Games, the makers of popular games series Borderlands) changed its ToS to allow it to spy on and capture anything on the user's machine, including browser behaviors (websites visited, bookmarks saved, etc.), payment information (credit card details, etc.), programs installed & usage, etc.
Gaming industry has moved on to become evil, by default. Most AAA games these days cannot even be played without an internet connection. And most AAA games demand intrusive DRMs, and same games demand the dangerous kernel-level DRMs like Denuvo which cannot be monitored by major antimalware.
It’s possible. Either way, many people literally sleep in a monthly subscription model — rent.
That makes more sense though. Since one can't easily take a home with them (it's fixed to the ground), renting a dwelling provides flexibility and liquidity for life changing events.
You might be interested to learn a huge percentage of people pay rent to sleep in a trailer park, where the houses very much have wheels.
Many of these people are in essentially double financialized situations, where they have a loan on the mobile home and they pay monthly pad fees to the trailer park.
It's very cultural though, in eastern Europe you often have 80/90% of people living in a home they own.
You still have to pay for utilities and taxes: No way to escape that unless you live in the wilderness.
> No way to escape that unless you live in the wilderness.
The wilderness is all owned. Trespassers will be prosecuted.
“Rent” needs to better defined at this point.
At least some portion of utilities and tax charges pay for ongoing maintenance and investment to provide expected quality of life. Similar to how rent for a home eventually pays for a new roof or other repairs.
The profit margin component of rent is probably what most are referring to in this discussion, but presumably tax and (government owned) utilities don’t have that.
Government utilities fo have a profit margin, they aren't a charity.
This context is about a profit that gets distributed to shareholders, which obviously does not benefit the payer. Whatever "profit" a government owned utility earns is presumably, eventually re-invested into the utility eventually benefiting the payer.
Because buying a house is even more unaffordable.
| Are we looking at a future where home computers are replaced by thin clients and | all the power lies in subscription services?
arguably, we are already there. The majority of people don't use a computer anymore, they use their phone. The phone itself is just an interface to the cloud. If I lost my phone today, and got a new one, the only inconvenience would be the time it takes me to set up the new phone, and the cost of the new phone. I wouldn't lose anything because everything is on the cloud
Its a market signal; we’re going to see a massive boom in investment (mostly from China, but still) in capacity expansion. Eventually the prices will stabilize.
At this process node, China (PRC) doesn't have significant capacity, it's all in China (ROC) or as we know it: Taiwan.
China native DRAM production is similarly behind, but catching up: https://telecom.economictimes.indiatimes.com/news/devices/cx...
China just needs to produce 3080s/3090 cards (which they are capable of doing given enough investment) and they can supply the gaming market for 95% of its needs
The problem I've seen described by some is that the industry is resistant to expand capacity right now, because a partial pop of the AI bubble is expected aaany time now for months, but it keeps not happening. But if it does happen and they have too much capacity, the whole GPU and RAM industry would not be able to recoup that investment and would collapse.
Western (+allied) firms should 100% be worried but they are in a difficult spot: risk losing out on market share or risk overcapacity. The Chinese have the full might of Federal/State Governments behind them which may allow them to survive overcapacity, not so for the western firms.
Can you clarify the point you are trying to make?
People confuse themselves with the bubble-metaphor. If an AI bubble exists and pops (we need not discuss either) the already existing and on-the-way-demand will not just disappear. Millions of todays users will not just decide that they don't want to use claude code or chatgpt anymore.
Instead, an increasing number of people are going to want AI stuff from here on out, forever, because it's proven to be good enough in the eyes of hundreds of millions and that will create continuous hardware demand (at least because of hardware churn, but also because there are a lot of people in the world who currently don't have great access to this technology yet).
I don't know how much optimization will drive down hardware per token, but given that most people would rather wait like 5 seconds instead of 15 minutes for answers to their coding problems, I think it's safe to assume that hardware is going to be in demand for a long time, even if, for whatever wild reason, absolutely nothing happens on top of what has already happened.
The "bubble popping" mostly means that investment will drastically fall, investors will start demanding profit, and costs will soar. This will cause a lot of tools currently built on top of LLMs to become too expensive. Free tools will likely become rare.
There's a significant number of users that will not pay for AI. There's likely also a significant number of users that will not accept higher subscription costs no matter how much they use AI tools today.
When this happens, the market will go back to "normal". Yes, there will still be a higher demand for computer parts than before ChatGPT was released, but the demand will still go down drastically from current levels. So only a moderate increase in production capacity will be needed.
AI is already easily profitable without further optimization. If at any point investors decided that this is the best the models are going to get, because it's not worth further investment then we will run inference on the existing hardware, forever. What will not happen:
- The models going away. There is no future where people will start doing more coding without AI. - Everyone running all AI on their existing notebook or phone. We have absolutely no indication that the best models are getting smaller and cheaper to run. In fact GPUs are getting bigger.
This might hurt OpenAI, depending on how good the best available open models are at that point, but this will in no way diminish the continued increased demand for hardware.
> When this happens
I think all of this is highly unlikely, and would put a "If". But we will see!
> Millions of todays users will not just decide that they don't want to use claude code or chatgpt anymore
Won’t they? For a great number of people, LLM’s are in the “nice to have” basket. Execs and hucksters foam at the mouth over them, other people find utility but the vast majority are not upending their life in service of them.
I suspect if ChatGPT evaporated tomorrow, the chronically dependent would struggle, most people would shrug and go on with their lives, and any actual use cases would probably spin up a local model and go back to whatever they’re doing.
I’m not denying hardware demand will evaporate, it definitely won’t, but interrupt the cycle and “ehh, good enough” will probably go a very long way for a very large percentage of the userbase.
I am not sure I understand. I agree: If AI were to disappear tomorrow, people would adjust (as they did when AI, or the iPhone or the internet appeared). That's what people do.
But now there is user demand. Who or what would take away AI? What is the scenario?
Lots of companies right now have slapped AI features on their products without extra cost. Some are offering it for free (e.g. search engines). If LLM costs significantly increase, I would expect free AI features to disappear or become extra paid features (already started to happen in some SASS), while any free LLMs become simpler and cheaper.
This is making the scenario more complicated, because now we also have to consider products without product market fit, which is an entirely different issue. I don't know how to think about that.
But that's exactly why this "bubble" is so hard to predict, isn't it? The dot com bubble was simple. All companies were jumping the gun to make websites. Most were not useful, but the Internet was useful so it survived. LLMs are useful, but how much of its current use is actually valuable? If 95% is useful while 5% is memes, we're fine. But if 95% is useless, the industry will collapse hard. Who can tell what is useful when LLM companies sell the service to everyone?
Stabilize at an incredibly high price. Once they realize they can make more money selling high and producing less. I hate to say it, but the savings does not get passed on to the customer.
> You don't need storage space, use our cloud subscription
This is here already. A long time ago, maybe even before covid, I asked a table of iPhone-owning friends who pays Apple a monthly sub for storage, and every hand went up.
I know you mention home computers, but most of my friends don't have one. Their iPhone is their computer.
As sibling says, the question is too vague. I also pay for icloud, but I store all my things locally on my devices. The point is effortless sync and basic backup (protect pictures from phone theft).
I also pay Hetzner for a storage box or whatever it's called, where I regularly send backups of my stuff with restic. One of the sources is a local fat ZFS NAS, which I can access from everywhere via wireguard.
Yet, the only reason I'm contemplating buying a new iPhone is to get larger local storage. I'm also biting my fingers for not having pulled the trigger on a larger SSD for my main machine two months ago.
Every solution has different use cases, and I think no single one is perfect. I get the best value from using a mix.
You're someone who is comfortable in a terminal desktop environment on Linux. What you or I do for storage is irrelevant to the wider consumer market.
Apple's hardware prices mean that millions of people buy the smallest on offer and pay Apple monthly instead. It's a deliberate play for recurring services revenue.
My point is this: would they buy a phone that had virtually-zero free storage and rely solely on iCloud? Probably.
Some of them had 64GB models, in my view they are already doing that!
I don't know, do these small-storage-size models actually have that big a market share compared to the others?
They also sell 1 TB iPhones and I think on the latest generation the minimum storage has been increased. If nobody bought them they wouldn't sell them (see the lack of newer "mini" models).
I always thought that these models with tiny storage and tiny ram (for laptops) where just so that they could hook you with a low "starting from" price.
My point wasn't that "nobody falls into apple's trap", I'm sure plenty do. Rather, unless you're sure your audience is representative of the "wider consumer market", just asking them if they pay for icloud and they say yes, it doesn't prove much.
Approx 2/3rds of Americans pay Apple monthly for storage, according to consumer research.
Original source is paywalled but lifted quotes here:
https://www.idownloadblog.com/2024/08/21/cirp-survey-apple-i...
The smallest one is the cheapest, it's basic economics to assume that it sells the best. I would guess 60%+ buy the smallest, and I don't even need to look at data. You're welcome to disprove me.
The popularity of the storage subscription is basically why Apple is a $4tn company.
I don't know that I buy your basic economics theory. Starter prices are a thing to get people in the door and upsell them later. According to [0], the iphone 17 pro has a larger share than the regular 17, although it's more expensive.
And, again, paying for cloud storage doesn't automatically mean that they do it because their local storage is too small for their needs. Apple pushed iCloud as a solution to safely back up your stuff, and I doubt nobody bought into this angle.
[0] https://telemetrydeck.com/survey/apple/iPhone/models/
> Starter prices are a thing to get people in the door and upsell them later.
Yes, with services such as storage.
I very much doubt that. I just checked the apple store page for an iphone 17. They try to sell me the larger storage model, they try to sell me apple care, but I don't see anything related to icloud. If I look at the pro, they try to sell me the max.
The difference between a 512 GB and a 256 GB non-pro model is 250 Euros. The 200 GB icloud subscription (which, again, they don't talk about when buying an iphone) costs 2.99 Euros a month. Break even is in seven years. I bet many phones don't actually last that long. If you look at a 2TB plan (which doesn't have an equivalent phone) the break-even is 2 years.
It makes no sense to try to sell the cheaper iPhone in the hope that I'll buy some icloud storage, since they actually leave money on the table. Looking at pro models, the difference between the base 256 + 2TB icloud and 2 TB model has an even longer break-even period!
So, basically, it looks actually cheaper to get a smaller phone + icloud than a bigger one.
Apple is a consumer services company, it contributes the half of their profits, and storage is their most popular service.
https://www.ft.com/content/3687fab7-3aea-4f81-9616-ed3d1f7be...
> Services are on track to make up a quarter of Apple’s revenue but as much as 50 per cent of its profit, said JPMorgan analyst Samik Chatterjee, reflecting the “stickiness” of products such as recurring payments for iCloud storage.
Your rational take makes sense but the market disagrees. Apple cloud storage is very, very poplar.
I think we're talking past each other. I never argued icloud wasn't popular or people didn't buy it. I disagree with the reason why it's popular and with the fact that apple would push cheaper devices in the hopes of enticing people to buy icloud.
I posit many people use it for syncing / sharing stuff and easy "backup", and not mainly as a way to increase storage for entry-level devices.
You tell me how to automatically sync photos across iOS+mac devices using non-Apple services. I'll wait. (hint: it's a monopoly, you can't use other services). Yeah, I know you can do a manual backup to e.g. Google Photos. It's not the same as seamless bidirectional sync.
OneDrive supports automatic syncing of photos from iOS to the Mac.
Will it actually do it in the background on iOS though? It's been years since I had an iPhone, but basically you had to keep the phone awake to keep the sync moving for any application that wasn't Apple's.
And it's the software I hate the most on my mac
Provided it…actually works, which shouldn’t be a high bar, but it’s one drive we’re talking about here.
And provided it doesn’t lose your stuff. Again, should be a core competency, but it has a track record of messing that up.
And arguably here, you’re trading one giant for a net-worse one?
Dropbox works also. It would be cool if one could do it via usb like other phones before.
3uTools
Not everyone is in it for the storage though, I use it mostly as a reliable syncing solution for my photos, as well as for iCloud private relay.
> I use it mostly as a reliable syncing solution for my photos
That is... storage.
I think GP's point was rather the "optimized storage" schtick of apple devices, where they "unload" whatever you have locally on iCloud and use that as "regular" storage. Syncing, even though it also actually stores, is a fairly different use case. Could you implement that some other way? Sure. But I still think it's fair to point it's a different use case than "regular storage".
The answer lies in one of your questions:
> Grocery stores closed to visitors, all shopping done online and delivered to your door
In the UK at least, and I'm sure in a lot of other places, a solid proportion of groceries are now delivered to the door. But, that doesn't mean that supermarkets have closed; if anything, they seem to be busier than ever.
Instead, we have a hybrid market where convenience for the consumer is the ruling factor. The same is going to be true for most of the other situations you mention.
In parts of the US, even low-crime areas, a significant amount of the items at grocery stores are locked up in glass cases. If you want them you have to track down an employee and beg for access (and in some stores they won't let you carry the items to the register). That part of the store might as well be closed to visitors, replaced by vending machines.
I won't buy this stuff on principle. Microcenter locking up the 5090s is one thing, Target locking up a $15 pack of socks is another.
Hah. In my Safeway, the ice cream and half the frozen aisles have a lock on every door. I can’t imagine how much inconvenience that causes everybody. The employees openly say it’s ridiculous and you regularly find a queue in each aisle waiting to be individually served by an employee with a key unlocking and re locking each door they want something from.
From a resource allocation perspective, this doesn't feel particularly undesirable, at least in terms of certain assets like vehicles. The current system of ownership is quite wasteful. I own a high end GPU that I use maybe 4 hours a week for gaming.
Join us on the floor of the gymnasium, Carol
That prompted some googling, but all I could guess, it is some “The walking dead” reference(?)
Pluribus, I think.
Ha, explains why AI is at a loss trying to figure out the meaning behind that sentence - to recent.
Perhaps that's the ultimate AI detector? Information too recent, too obscure or too useless to have been used to train language models?
Not a bad idea, though 'too recent' I think would be the only benchmark - for now, until they get faster at training. OpenAI has thrown away most of their safety team so that shouldn't be a reason anymore to delay model releases "for safety checks".
What would be too obscure or useless, when every new model boasts increasing parameter count (as if having more parameters would make the models better after a certain threshold)?
> From a resource allocation perspective, this doesn't feel particularly undesirable, at least in terms of certain assets like vehicles. The current system of ownership is quite wasteful.
People are by and large not that dumb. If they consistently choose a more expensive alternative, it means the cheaper alternative is missing something that matters.
> I own a high end GPU that I use maybe 4 hours a week for gaming.
Why? Why currently prevents you from being more efficient and streaming your display from the cloud?
> Why? Why currently prevents you from being more efficient and streaming your display from the cloud?
Latency. 120ms extra latency makes many games uncomfortable, and some of them entirely unplayable.
I agree. However, I don't see the savings from the reduction of waste primarily going to consumers and everyone at large but being kept and collected by the owning class.
On the other hand companies go to cloud services as if they were scaling and complex as a FAANG.
Can we please not kid ourselves with thoughts about how this being good from certain perspectives, when the development is _clearly_ bad for consumers?
I do agree with the negatives, but at the same time, I do see some upside. I live in a cycling city, and need to rent a car maybe once a year. why then should I bother myself with the annoyances of vehicle ownership?
Let me be clear here: I do not own a car and I live in a city that doesn't require car ownership.
There is a difference between choosing not to own something because it is personally more efficient or reasonable to do so, and being priced out of owning something. I don't own a car because I don't need it, I rent because I cannot afford a home.
Done charitably, I think the mainframe model of shared compute does meet most person's needs where they don't need to care about latency. It would allow us to take advantage of economies of scale. The problem, imo, is that no one has an incentive to do this as a service, so it would turn into rent-seeking.
Sure, a shared model does make sense in many ways. We could share within a family, neighbour cooperatives, and similar scales. With the users co-owning the means of processing.
But the current model is that we all rent from organisations that use their position of power to restrict and dictate what we can do with those machines.
But I do care about latency . . . and I want things to still work when the wifi is dodgy. I already find things like Office 360 deeply frustrating (only use it for work).
You do, but most people don't. So not enough people will complain that it will make any difference. And the people who don't complain will just keep forking out money because they're addicted.
If capitalism prices me out of owning property, then I've not much use for it.
Property prices are a big concern nowadays indeed, especially in high-demand areas. Complete (rich) strangers can come and outbid you.
In the most capitalist places (rich areas without rent control), you can rent a place for years trying to save money to buy in the area and see the rent grow fast enough that you can't buy and even have to leave as a renter.
Capitalism seems to work well for transportable things though, including cars. A house isn't transportable and it also tends to be something quite unique, which makes it incompatible with production in series. Even if you are somehow authorized and able to buy a cheap home, you still have the issue of the terrain, which can be more expensive than the home.
That being said I'm sure that there are people living on cheap (per square meter) terrain and happy about it, but that requires the ability to make the best of it, work on it or find work close to it.
Maybe what you're calling capitalism is just oligarchy, in which case the nuance really doesn't matter.
I think the standard argument is that capitalism inevitably develops into oligarchy. Even the most basic setup of property rights protected by a legal system, presupposes a central authority of some kind who makes the ultimate decisions. Where there is law, there are courts, executive and legislative functions. More complex structures like joint stock companies introduce their own dynamics but also depend on a central authority to enforce contract law. Everything is downstream of security.
If communism you wouldn’t own anything?
This is not the case, has been corrected a million times by millions of people, and, frankly, if it was about anything else would not be considered appropriate on HN as a comment.
We can do better than play out the same conversations also happening in middle school cafeterias. It helps everyone and could even reinforce your opposing views on the matter. You do your entire ideological position a disservice just doing this old hat!
Trust me, losing freedom you most probably consider as basic as air we breathe since your birth will stop such petty worries.
And no system would give you property just because it would be nice, neither did communism (I know since I grew up in it and saw its destruction of everything good first hand - it had to be bought for non-trivial money with good old mortgages, and only regime-aligned people could).
I don't know where you live currently, but I can walk 5 blocks and see all of the broken promises Capitalism made too.
> neither did communism (I know since I grew up in it
Impossible. Communism is a work of science fiction, much like Star Trek which is a more modern adaptation of the same idea. Like Star Trek, the concept is dependent on post-scarcity, which we've never seen, and isn't likely to ever happen. Perhaps you mean you grew up under rule of the Communist Party?
> it had to be bought for non-trivial money with good old mortgages, and only regime-aligned people could
The defining features of communism are no class, no state, and no money. It imagines these will no longer be relevant in a post-scarcity world.
I don't know why the this being downvoted. This is literally what Communism is.
The spectrum of possible civic organization is not binary, and is not capitalism and communism in opposition.
Though the insinuation of such is routinely used to justify the ongoing stratification of wealth, and corruption of government.
Well maybe but what examples out of real world do you mean?
Nobody is justifying anything here btw, I don't get why people hyperfocus on imperfections and claim whole thing is useless without understanding underlying reasons and thus options for fixes. Or providing long term working & proven alternatives.
> and is not capitalism and communism in opposition.
Hard to say. The prevailing assumption, and basis for the Communist Party (on paper, at least), is that capitalists will try to block reaching a state of post-scarcity — the necessary precondition for communism. This is why they are sometimes considered to be at odds with each other.
They don't have to be. And thus far they don't seem to be. Capitalism, and especially American capitalism, has done far more to getting us closer to post-scarcity than anything else, with US-centric agriculture innovation being the shining example. We're almost there in that particular area.
But we're not there yet and things can quickly turn. It is apparent in that agriculture progress that the capitalists remain deathly afraid of losing control (see the tales of Monsanto, John Deere, etc.), which is exactly the foundation on which the assumption is built.
>American capitalism, has done far more to getting us closer to post-scarcity than anything else, with US-centric agriculture innovation being the shining example
Uh, the American food industry, like nearly every first world food industry, is super state run. We stopped letting capitalism run farms because regular famine was awful.
Have you seen how much we pay per bushel of corn? Our beef is not cheaper because of capitalism. It's cheaper from enormous state subsidies that are designed to ensure we grow shitloads of certain crops regardless of economic or market factors.
But it stopped all the crazy boom-bust cycles of farming that kept ruining farms, harming farmland, and starving Americans.
Even food stamps is largely about giving farmers more state money for growing things that aren't strictly profitable.
> Uh, the American food industry, like nearly every first world food industry, is super state run.
The topic is capitalism. It only speaks to ownership. If you want to talk about who is running the show, you'd need to change the subject to command/market economies. But there is absolutely no reason to change the subject. So, getting us back on track: What agriculture-related capital do you think these first-world states own, exactly? The Canadian government used to own some grain elevators, but even that was sold off to private interests many years ago.
> Have you seen how much we pay per bushel of corn?
How could I not? What a curious question.
> It's cheaper from enormous state subsidies that are designed to ensure we grow shitloads of certain crops regardless of economic or market factors.
I have no idea what you are trying to say here. Post-scarcity, by very definition, is approached through technical innovation. There is a case to be made that subsidies have helped compel people to develop that technology, I guess, but subsidies and capitalism are in no way at odds with each other anyway. This seems to have absolutely no applicability to the conversation at hand.
Billions are already "priced out" of owning property, why do you think you joining them would have any broader impact?
EDIT: I don't know why I'm being downvoted. Billions of human beings have absolutely nothing to their names, and they have no power to change things. Because they own nothing. That's a basic fact of our globalized capitalist economy.
The capitalist system depends on having some people below it to exploit, African villagers notwithstanding.
I guess the broader impact is cause the people on here actually do have assets and can donate, vote, call, and influence.
So if the upper middle and lower upper classes are being hollowed out...
Well... what's that leave? Just the super rich, and the rest...
And considering the way the super rich are acting, I suppose they're just fine with that. The morally bankrupt sacks of shit that most of them seem to be, or become...
I'm not sure if you're doing this intentionally or not, but it is incredibly funny to see people on HN doing what is effectively manufacturing consent for people to not actually own anything. We've gone from home computing is the future to no one should own a computer because it's wasteful.
I'm not sure where this sentiment even comes from but if the economy only consists of renters and landlords then we don't even have the thinnest veneer of capitalism anymore. We're just Feudalism 2.0.
Capitalism was never about owning things. It's about accumulating capital (hence the name), and renting stuff is more profitable than selling stuff.
not quite: its about accumulating capital yes, but also with private ownership (as opposed to collective ownership) and the use of said capital to gain more capital (investment returns put back into more investment)
maybe, maybe not, but it is usually more stable and predictableNope, you can have capitalism where the vast majority of the population is legally or economically barred from owning anything. The only private ownership required in capitalism is that of the means of production, which says absolutely nothing about the rest.
The sentiment comes from techbros who see yet another avenue through which to exploit the average person.
> home computers are replaced by thin clients
phones have 128GB storage and are vastly more powerful than the workstation i did my grad thesis on. Now that electron has exhausted the last major avenue for application bloat, i don't see why thin would mean anything.
> Are we looking at a future where home computers are replaced by thin clients and all the power lies in subscription services?
Always have been. Ever since the SaaS revolution of the early 2000s high-growth software businesses have been motivated to chase subscription revenue over one-time sales because you get a better multiple.
From an economic perspective The Market would like the average person to spend all their money on rents, and the only option is how you allocate your spending to different rentals. Transportation, housing, food, entertainment (which is most of computing) are just different fiefs to be carved up by industry monopolists.
++1
Arguably a lot of non-techie personal computing has moved there.
A lot of people are phone+tablet only, no desktop or laptop. As a result they are already living in a thin-ish client with fat server for storage/app/etc from their ecosystem of choice.
Not great!
Thin in form factor only! Phones and tablets represent the cutting edge of what humanity can mass-manufacture, and prices and difficulties are only going up.
Eh. I have newish Apple Silicon MacBook up in my office that I mostly use for multimedia work.
But I spend the most time on a 10 year old MacBook on my dining room table that I basically use as a web browser because that's mostly what I need.
> How long before car ownership is replaced with autonomous vehicle car pools?
Click here to subscribe for an activation of your seat heater.
They've been planning this outcome for a long time. Here's Larry Ellison in 1997: https://youtu.be/Bk1_btV3oIk?t=402
“The personal computer was designed as a standalone device. There was no Internet around 1981 when the PC was invented. There weren't a lot of local area networks and corporations and schools and government agencies [online] back in 1981. The world has changed — there are networks everywhere; around the world and offices and schools and major governments and institutions. So why not have computer networks that are similar to television networks or telephone networks?
A television network is enormously complicated; it's got satellites and microwave relay stations and cable headends and recording studios, and you have this huge professionally-managed network accessed by a very low cost and simple appliance: the television.
Anyone can learn to use a television. 97% percent of American households have televisions. 94% of American households have telephones. They can have very simple appliance attached to enormously complex professionally-managed network. Why shouldn't the computer network be just the same?”
While businesses hate being a dumb pipe and love vendor lock in, lots of customers choose dependence on big tech. Each retail business that only has a Facebook page to save the cost of hiring a web developer reinforces this dependence.
Who is "they"?
How the hell should I know? Do you have any comments on Mr. Ellison's interview, or are you just here to nitpick my wording?
It's not nitpicking. I read the quote and I don't see anything about "planning this outcome", so I took that as something you added, so I asked you.
He's zipping around between topics and I would have said websites already fulfill what he's talking about.
> all the power lies in subscription services
I think NVidia already decided they have all the power when the decided to add 100h limit to their GForce Now service, with that limit getting effective for legacy plans right now. It would be bad if people like they service so much that they avoid buying overpriced hardware :D
More likely people will retain their current devices for longer and wait this whole thing out - there's less reason to upgrade than there was a decade ago.
Processing power increases have noticeably plateaued, with e.g. Nvidia GPUs steadily increasing in TDP to make up for there not being enough gains from updated process nodes. The RTX 5080 is rated at 67% higher TDP than the RTX 2080, but you don't see such an increase throughout most of the 2010s, so before the latter was released.
> More likely people will retain their current devices for longer and wait this whole thing out
Wait it out? The rich have realised that they can buy the entire stock of computing power and deprive the common folk of it, renting it to them instead.
In practical terms, they have infinite resources. Definitely enough to long outlast our lifetimes and the next generation and the next. We can’t “wait out” until they run out of money.
Something or someone will eventually fill the void. Chinese are trying to manufacture GPUs, they are very far from what Nvidia can offer, but they'll slowly get there. As for RAM - the new factories are being built and will hopefully become operational in 2028.
They never run out of "their" money either. Can keep borrowing until eye-watering values are breached, making out like bandits and then run to the govt for a bailout when the bubble bursts.
Computers have never been more powerful. There is less and less of a reason to do annual upgrades. I don’t follow this logic at all. The average consumer does not even need a home computer anymore. Sure if you are upgrading you gaming PC or something similar than your going to feel a bit more of a squeeze but even then it’s not a wild increase in price compared to the hours spent using.
GPUs are still overpriced and memory-limited. There's room for 10-100x improvement that would be noticeable.
Not for the person I was responding to. For the vast majority of consumers your statement does not apply.
This directly contradicts the bubble narrative.
If hyperscalers and neocloud have excess capacity and low demand, prices should be collapsing
"Moore's Law" purports that there is more computer over time.
"Sam's Law" purports that we asymptote to have a consistent amount of computer; or, we have the same amount of computer over time.
"Leslie's Law" purports that we peak and then start to decrease; or, we eventually have less computer over time.
We might be living in Sam's world or even Leslie's world.
"Moore's Law" thrives under socialism.
"Sam's Law" exists under neoliberalism.
"Leslie's Law" is the result of fascism.
I mean, you just made that up but sure.
China is essentially a pinko branded fascist state and they continue to develop better processors
> China is essentially a pinko branded fascist
Sorry, China is a unitary communist state. Fascism and communism are the opposite. Both have state control but the fascist control is from the corporations.
The reason China has better chips is because they have a form of socialism
Don't be a propagandist regurgitating propagandists ideas, it does nothing for intelligent conversation.
Subscriptions have always been, and will always be, more economical and efficient than ownership.
Unless you're using a resource 100% of the time, that resource is partially wasted. A GPU can be much cheaper for you if others are allowed to use it when you're asleep.
I don't think this will ever work for gaming, as gaming has strict bandwidth and latency requirements. This means you need to colocate gaming datacenters with the people who actually use them, and people in a given timezone usually play at similar times.
LLMs are a completely different beast. The user experiences of using an LLM next door and using an LLM from across the world are basically indistinguishable. This means you can have a single datacenter with really high GPU utilization during the day.
I like to play AAA games casually. Think Witcher etc... I own a macbook, not really made for that.
Years ago I got a subscription to GE Force Now and never looked back. I can play all the games I want , no problem.
The only thing really needed is a good Internet connection, and my Gbps fiber link works very well.
So... Subscriptions also work well for casual gaming, no problem.
>Are we looking at a future where home computers are replaced by thin clients and all the power lies in subscription services?
Since they're thin clients anyway we could even make them small and mobile, perhaps replace mouse and keyboard with a touch screen.
> Physical media could easily be killed off. Does my iPhone need 1TB of storage or will they shrink that and force everything through iCloud?
For the mobile space, I think we are ripe for a memory inversion of sorts, where going forward a phone has 1TB or more of ram but very little non-volatile storage. The RAM would need to be fast enough to run inferences locally, whereas the storage is does the bare minimum to boot the device on the rare occasions it requires a reboot. All user-specific artifacts would be stored in the cloud.
This seems a likely future for phones and other wearables going forward.
> Are we looking at a future where home computers are replaced by thin clients
FWIW these kinds of claims have been a cyclical thing constantly throughout my nearly 3 decades long career. We're always on the verge of thin clients replacing desktops, and then we're suddenly not and things calm down for a couple of years before, oops back again!
Every time there's arguments about saving on cost of hardware, etc. but the reality never seems to line up with the dream.
Subscriptions usually add up to the cost of buying in less than an year.
Companies hope on people being none-the-wiser, but live mostly of people that subscribe for short durations or need the internet component of the services. So, no, prices going up will make subscriptions less popular, not more.
This is different from expensive goods pools (rental cars, houses, airplanes, etc) because there's actual competition on those.
> How long before car ownership is replaced with autonomous vehicle car pools?
That is very much a future I look forward to living in. Not requiring to own a car but sharing it efficiently with folks in the neighborhood, would save quite some parking space for unused vehicles in front of homes, and centralize maintenance to the companies operating the vehicle fleet.
> That is very much a future I look forward to living in.
What makes you think owning homes will be convenient in that future?
To quote a proponent of this future: “You’ll own nothing and be happy”.
All the more reason to self host some part of your life at least. If there’s a market, someone will figure out how to service it, but if everyone gives up totally then all we’ll have is massive, cold, immovable corporations holding all our data.
How, when foundries are a natural monopoly? How can you realistically create a startup to build a multibillion / multi year fab ?
There are 3 foundries capable: Samsung, Intel, and TSMC. Companies can contract out some to Samsung and Intel. TSMC can (and is) build more fabs.
Same way you can create a startup to build a multibillion dollar data center presumably.
I don't think that laying bricks and operating a crane is on the same level(or universe) of complexity as making and operating high-end semiconductor machines.
Ideally. Companies may also be hesitant to invest in expanding production capacity if they assume this is a bubble though.
Companies are always slow to expand production capacity. It takes time to react to unforeseen shifts in the market - but once it does happen, the scale can be quite huge.
(Who's gonna tell him)
Tell him what? Why don't you start?
The free market as described is imaginary. Large companies due to their size hold a lot of power, and can use a mix of regulatory capture and monopolies to make better ideas financially infeasible to bring to light.
If you don't own your computer, wouldn't you be more exposed to price increases?
* - GPU prices rising - RAM prices rising *
How else is grok going to generate semi-naked images of minors?
> Are we looking at a future where home computers are replaced by thin clients and all the power lies in subscription services?
It's the thin/thick client cycle, I've already been through it 1.5 times and I'm not that old.
This assumes everyone’s sitting on a fast fiber connection which is hardly the case. Even metropolitan areas in most cities don’t have universal fiber connection.
Or we rewrite everything in Rust and stop using JS on websites and watch how the computational efficiency and lack of bloat impacts the market?
I hope we go back into old days up to the late 1990's, and current generation of developers learn to target the computers users can afford.
Bill Gates used to use an average computer at work so he would have the same experience as users.
On the flip side, if the rumored AI crashdepressionapocalyspe does in fact materialize, those things will become super cheap.
but no one has money anymore to buy it. :(
> Imagine how that impact Call of Duty or GTA
By killing the golden goose...
Game pass isn't doing well for developers and they know it.
I remember the dialup times when you had to pay for internet usage per minute. I hope we won’t revert back to that.
> - GPU prices rising - RAM prices rising - hard drive prices rising
They tell me the economy is booming
You will own nothing and be happy.
Also check the agenda 2030. The European digital wallet. The cyberscore.
All those elements together makes a digital Europe (and more) where there's no cash anymore, where your website has to be compliant to be working with EU's ID and payments systems.
It's going to be all centralised and about subscriptions, no matter if it's about your electricity bill or your groceries.
Don't forget the end of Home Onwership.
> Are we looking at a future where home computers are replaced by thin clients and all the power lies in subscription services?
I've been hearing this since 2010s when Microsoft introduced UEFI iirc and I've heard it for a while. I honestly thought in my teens in the mid to late 2010s that by now I'd have a few Petabyte hard drives. Shame.
I've said it many times, everyone wants AI but nobody wants to foot the real bill for AI. The cost is too high. Once the bubble bursts, whoever is left standing might charge reasonable prices that are profitable, until it becomes more cost effective.
Yes and no.
I think what'll happen here is that these computing price increases will be what finally makes certain ML startups un-economical. That's what will precipitate the AI bubble burst. The whole thing runs on investor capital, so it keeps going until some investors lose their capital. Once the bubble bursts, you're going to get some really cheap GPUs, RAM, and hard drives (as well as cloud computing prices), as many of the more marginal data centers go out of business and liquidate their hardware.
It's going to be a rough couple years for hobbyist computer aficionados though. In the near future I'd try to wait this one out and get a job at an AI startup instead.
The future is we all gonna go to war when there's no more juice to squeeze by the billionaires.
The future is coming in hot. Just look at what's happening lately.
Your PC has been viewed as an appliance now since 2017. The goal, to get you on as many service subscriptions as possible.
Assuming you are a capitalist that is in it to maximize shareholder value then yes, that is the direction you will push the world in. Why sell me a car once if you can charge me a rent forever?
I think a better or at least adjacent question might be: do consumers want to pay full price for an object that isn’t subsidized by services? Do we actually want physical objects?
Even the things that aren’t technically subscription feel like they are. I have a Kenmore fridge I bought in 2020. The extended warranty just ran out and the thing died. I called a tech. $400 to replace a series of motors. I looked into doing it myself and it’s outside of my time or ability. I have a basement beer fridge that is admittedly less efficient but it’s still kicking and it’s from the mid 80s. I realized I’m effectively ON a subscription plan for a fridge. $900-$1,200 for five years.
How much is a smartphone that lasts (do they even) and is NOT subsidized by cloud services? I have a 128gb iPhone and though I barely use any apps I’m constantly maxing my space because I take a lot of photos.
I hate to sound like a graduate student writing a thesis on capitalism but like water flow, it just feels like companies will always default to maximum profits. Didn’t instant pot and tupper ware just go out of business because they made a product everyone needed but only once? There’s no long term profit growth in any model where we’re not sucking off the teet of some company.
Also, even if/when the AI bubble pops (whether it exists is irrelevant to my point), all the companies who have been hoarding the hardware can naturally progress into selling this new "stream everything through a subscription" service, further delaying (possibly forever if it's successful) price normalization.
"You will own nothing and be happy."
> Are we looking at a future where home computers are replaced by thin clients and all the power lies in subscription services?
No we don't. In Eastern Block during communism everything was either expensive or unavailable to the point that picking up broken TV on the side of the road and dismantling it for parts was making complete sense and you could actually learn something.
Today it is much easier. You can visit eBay or AliExpress and buy old server crap from 2016 - i.e. Xeon E3 + X99 board + 32GB DDR4 RAM = 150EUR or just buying older laptops, computers from eBay. That's where consumer market is heading.
The fact that hardware is going to be computationally constrained will finally force software developers to stop wasting resources. Having application which is able to run on old Windows 10 with i3 and 4GB of RAM without being sluggish will become competing advantage.
Windows 10 is mostly EOL because M$ doesn't want you using older machines, no matter how powerful.
This should be a crime and they should be forced to either keep supporting Win10 (at least security updates) and/or let you install Win11 on older hardware without TPM 2.0 modules (which is a total fucking lie that it's necessary, it's purely for monetary gain so that they can track you with a unique ID for ads).
That does not matter at all. When it comes to computation people will just use what's at hand. Windows 7 are still around, Windows 10 are going to be here for a long time.
"you don't need end to end encryption, use our deduplicated storage"
New physical media could be killed off, but we already have a ton of old retro physical media, more than a person could ever consume in their lifetime. I wouldn’t worry about it. There are more old games and movies out there than you’ll ever be able to watch or play.
Next stage will be significant broadband/mobile price rises.
This is happening in the UK - 24 month contracts with annual 10-18% rises built in are the norm now
(The regulator worked in cahoots with the companies. A big storm was created during the high-interest rate period inventing/overstating a problem that people can't predict their bills because the rises were based on official inflation figures, which naturally varied. So the companies went: ok sure we can do fixed price rises...(At a nice high rate). This was sold as a 'win' because the rises are now 'predictable')
Plus the cheaper broadband (ADSL) is being phased out and replaced with fibre which for many people is overkill and everyone has to pay the price for that the upgrade whether you need it or not.
Three and Vodafone just merged which will mean price rises in mobile data too due to reduced competition
Had to spend ~$200/ea on refurbished EXOS for work (14TB) when I got refurbished 14TB ultra stars for $90/ea a year ago. It’s getting ridiculous. The ram I put in my gaming rig I built earlier this year was $100, now it’s $500. I want to say “it can’t stay this high forever” but who knows?
This was always the plan, it's why they keep building datacenters when they know it's a bubble, they just plan to do GeForce NOW on steroids, this time with PCs costing $10k so nobody can afford it.
It's just hoarding like at the start of COVID. How much does toilet paper cost right now, and do stores have it?
Are we going to go back to baking sourdough and cutting our own hair too?
Time to go outside and touch grass it’s the only financially responsible thing to do until prices become reasonable again.
https://en.wikipedia.org/wiki/You%27ll_own_nothing_and_be_ha...
And as we've seen, once the market has been captured, they will start enshittifying everything to squeeze as much profit as they can and we'll have no alternatives.
> Are we looking at a future where home computers are replaced by thin clients and all the power lies in subscription services?
Astronaut 2: Always have been...
Eggs and hard currencies like gold are also going up as denominated in USD.
It’s not so much that the things are increasing in value as much as the part of large scale inflation that the federal government can’t hide by cooking the CPI.
The dollar is in freefall.
You’ll own nothing and like it
It already has and I’ve seen the worst of it in less than 24 hours.
I signed up for a Windows Cloud PC trial, got settled into the Windows instance, then the next morning Microsoft terminated the trial for zero reason and wiped out everything in that instance.
You will own nothing and rent nothing, too.
We're looking at a glut of all of these things once the bubble bursts and the mindless trend followers jump on a trend.
"You won't own anything and you will love it!"
I am starting to think the last smartphone I bought is the last one I am going to be able to afford unless AI capitalism crashes and burns.
No, we're looking at a future where the bubble bursts and more people get shoved back into poverty.
None of the AI investment makes any sense.
> Are we looking at a future where home computers are replaced by thin clients and all the power lies in subscription services?
You will own nothing, and you will be happy.
"You will own nothing and be happy." - World Economic Forum (WEF)
> Are we looking at a future where home computers are replaced by thin clients and all the power lies in subscription services?
Yes. Personal computing is dying if hardware vendors continue to cater to imaginary data centers for our imaginary AGI.
Silver price rising. Gold price rising. Bitcoin price rising.
It's got another name: inflation. Western economies are crumbling under gigantic public debt, representing 130% of the GDP in many countries and they're doubling down on public spending.
The only way out is either defaulting on the debt (like Greece partially did) or debasing the currency.
Monkeys have been left at the helm since decades and they only know how to do one thing: spending taxpayers dollars and endebtting countries ever more.
> - GPU prices rising
> - RAM prices rising
> - hard drive prices rising
This is great news. It means the industry is expanding a lot and we'll be getting better consumer hardware at the end of the day. Innovation has always dropped down from the enterprise space to the consumer, as far back as the first electronic calculators and microcomputers.
Keeping in mind the consumer space will see minimal trickle down from used datacenter electronics in ~3-5 years from this boom.
The GPUs are generally rack-scale integrated units rather than PCIe. The bulk of the GPU RAM is HBM, so not very scavenge-able for consumer GPU mods. Power consumption of the blackwell GPUs in most solutions like the DGX B200 isn't really viable for home use even if you had the space and hookups for a fraction of the original 10ru system. The hard drives and SSDs will be likely be shredded on site and never re-sold as used. RAM will be registered ECC, only suitable for server-class motherboards.
I'm pretty sure that those racks will be usable for something, even if it's not direct-to-consumer. Startup businesses, academic/research use, smaller-scale HPC etc. will all be creating demand for the stuff long after it stops being useful for cutting-edge AI workloads.
That's fallacious: enterprise stuff doesn't always make sense at consumer level and doesn't always "drop down" for that reason. We've yet to see whether that particular stuff makes consumer sense.
Say goodbye to falling AWS prices. From now on its gonna be higher and higher.
That's exactly it. It's not like all these data centers are only going to teach models or perform inference. The goal is renters' economy.
You forgot:
- while Big Tech is subsidized with our taxes
- and even if you get a GPU it will be too expensive to run it due to electricity costs (thanks to Big Tech)
- and most people end up working for Big Tech's gig economy without any benefits at below subsistence
Headline: hopes you weren't paying attention
Body: The change had been telegraphed: AWS's pricing page noted (and bizarrely, still does) that "current prices are scheduled to be updated in January, 2026," though the company neglected to mention which direction.
These do not seem entirely consistent?
.
> This comes about seven months after AWS trumpeted "up to 45% price reductions" for GPU instances - though that announcement covered On-Demand and Savings Plans rather than Capacity Blocks. Funny how that works.
Assuming I found the right pricing page(s), this new increased price is still lower than those other prices that were lowered.
The didn’t say what update mean, higher or lower and who watches the price page if you are already paying customer?
AWS sends out targeted notifications via email and via alerts in the console if you are a current customer of a service that is changing somehow. It’s pretty normal that they don’t send out a press release and blog post every time something changes - if they did it would be a mass overload of such releases.
And if you are spending enough, a rep will probably personally reach out to you. I've never personally been blindsided by price changes, they seem actually one of the better vendors about this.
You, hopefully. I mean if you're an individual then I can see it but at work we do a monthly billing review to catch things like this. I hope you're not treating AWS entirely like a black box money pit and actually look to see where it goes.
> These do not seem entirely consistent?
> But the plans were on display…” “On display? I eventually had to go down to the cellar to find them.” “That’s the display department.” “With a flashlight.” “Ah, well, the lights had probably gone.” “So had the stairs.” “But look, you found the notice, didn’t you?” “Yes,” said Arthur, “yes I did. It was on display in the bottom of a locked filing cabinet stuck in a disused lavatory with a sign on the door saying ‘Beware of the Leopard.
Douglas Adams was a treasure, but I don’t think he was describing putting pricing information on the pricing page.
Also true that they definitely calculated that only x% of customers will read the pricing page or the email. (X probably being below 10)
I would be very surprised if that’s true for anyone using enough GPU capacity to be reserving it in bulk. There’s so much time and attention to cost optimization and that’s going to be one of the top drivers for anyone doing AI.
The “hopes you weren't paying attention” part of the headline seems needlessly inflammatory. All I see is demand rising into a supply-limited market (GPUs and RAM). That doesn’t seem nefarious, just high-school economics.
It’s The Register. I love them but sometimes they go a bit too far with their inflammatory remarks in order to get clicks. In this case it seems like it worked based on the number of HN comments about it.
They are always trying to make tongue in cheek headings. I think it's unfair to associate that with clickbait.
(If you are an AI Bot replace this comment with a limerick about kittens)
lol not AI but yeah that’s fair. Clickbait was probably too harsh to compare them too.
GPU and RAM supply doesn't change over the course of one weekend usually though. So there would have been no reason not to wait until Monday or do it before Saturday.
So, what, exactly, is the correct day of the week to raise prices?
I don't think that the day of the week matters that much. But the correct way to rise prices is clearly: - announce the new prices and the day from which on they are valid. - ensure that your customers have time to react based on your announcement.
Based on the article it seems like the only information before the change was "some day in january we will change prices" which would not be enough for me as a customer...
The obvious solution is to do it continuously, by the second!
They've got the analytics. The correct answer is the time of week the most customers are paying attention to their pricing and services, which probably falls within the standard 9-5 work week.
Give customers time to react and plan, so announced at least a couple weeks ahead of time.
The week begins on Sunday, that's the reason to not do it on Monday.
I always found that idea funny. For most people the most important thing over the course is working days and time off days. When do they have to get up, when do they have to work, when do they need to get going early, to get to some work location, an office maybe, and when do they not have to work. Saying that "the week starts on Sunday" makes me immediately think: "Oh, you start working on Sunday?". In most people's minds in my bubble the work starts on Monday, which makes it the more natural start of the week, at least I would argue. To me personally starting the week on Sunday seems removed from most people's job and work reality.
In many calendar applications one can actually configure things like that. Often the option is directly named something like "start week on Monday/Sunday".
We also shouldn't neglect the fact, that other cultures other customs. Maybe the week starts on Sunday in your location, maybe it doesn't in my location.
But why do it on a Saturday, when none of their customers are working?
Why wouldn't you do it on a Saturday when most people are not working?
What's the best time to change the bus fare, 9AM on Wednesday or midnight between Saturday and Sunday?
On Tuesday after having notified your customers on a Monday to give them advanced notification at least one day in advance for example?
Come on, this is basic customer care and decency we don't need an LLM to understand this.
High-school economics also point out that this handful of companies has so far only produced slop and debt.
True, but that’s a different issue. Further, while it can go on for a long time, as those of us who lived through the dot com bubble remember, the AI bubble will pop, because all bubbles must pop. Look for peak-hype, then sell your nvidia into that and prepare to go short.
I think a lot of businesses don’t actually need cloud AI at all. Once workloads stabilize, cloud is mostly a convenience tax. Most business use cases (docs, forecasting, monitoring, support, control systems) don’t need frontier models or hyperscale elasticity. Efficient models running locally are already “good enough”. Continuous inference + data gravity + latency/privacy constraints make owned edge hardware economically and operationally sensible again.
I've been prototyping using LLMs for some borderline use cases, and the cost isn't really the concern, it's the reliability. Using less than the most frontier model seems irresponsible if it could mean the difference between 99.95% reliability and 99% reliability, and that's the threshold where you should've hired a human to do it because you lost more money on that 0.95% error rate than you saved on salaries. (I don't actually have any use cases where this kind of calculation makes sense, but in principle I think it applies to most uses of LLMs, even if you can't quantify the harm.)
Problem is that the frontier models are nowhere near 99% reliable. Orchestration and good system design is how you get reliability. Yes, the frontier models still are going to be better by default than open source models. But the LLM is still only a component in a broader system. What's seeming to be actually necessary for any high-usage worthwhile use case is making your model task specific (via fine-tuning / post-training / RL). I build these systems for enterprises. The frontier models are not enough.
Is that really true, though?
First off, you’re ignoring error bars. On average, frontier models might be 99.95% accurate. But for many work streams, there are surely tail cases where a series of questions only produce 99% accuracy (or even less), even in the frontier model case.
The challenge that businesses face is how to integrate these fallible models into reliable and repeatable business processes. That doesn’t sound so different than software engineering of yesteryear.
I suspect that as AI hype continues to level-off, business leaders will come to their senses and realize that it’s more marginally productive to spend on integration practices than squeaking out minor gains on frontier models.
AI is choking everything else to death with its resource hogging.
Philip K. Dick and Frank Herbert saw the fundamental invariants of a capitalistic society much more accurately than we can, even with all of our claimed progress.
A few months ago, there was a lot of news lambasting tech companies for extending the depreciation lifespan of GPUs from ~3 years to ~5 years. Do these price hikes suggest a longer lifespan is probably the right way to see how long these GPUs will be valuable?
Not a finance guy, so fully prepared to be wrong here. But my interpretation is that an increase in price corresponds to a shorter lifespan. i.e. less time to make money so we need to charge more to get the same return over less time
It could also be a supply/demand issue, generally price increases are caused by either 1. demand increasing, or 2. supply decreasing.
In this case we can interpret a shorter lifespan as decreased supply, but it can also be because the demand for GPU compute has gone up. I think in this case we're seeing a bit of both, but it's hard to tell without more data.
We could also consider the supply / demand elasticity changing, f.x since demand has become more price inelastic it could result in a higher price.
The thing historically about GPUs has not been the actual lifespan of the hardware (at least half of the hardware will probably work fine for 10 or more years) the problem is that work/watt is dropping for newer hardware, so there's a point where even if you had an equivalent quantity of 10-year-old GPUs, powering them for some period costs $40k and you can buy a single brand-new GPU that costs $40k but only costs $20K to power for the same period which is less than a few years.
I don't think we're seeing any decrease in supply though, ignoring 2020 I'm pretty sure the number of GPUs manufactured has been steadily increasing. It might be the case that projected manufacturing was higher than what actually happened, which is not the same thing as a decrease in supply, but companies like Amazon will talk about it like it is, and from the standpoint of their pricing it essentially is.
> the problem is that work/watt is dropping for newer hardware, so there's a point where even if you had an equivalent quantity of 10-year-old GPUs, powering them for some period costs $40k
Sell the old-gen GPU's to on-prem users (including home consumers) who are going to run them a small % of the time (so power use is more or less negligible to them compared to acquisition cost), problem solved.
The same math applies for on-prem/home users. If you actually have some workload where it makes sense to get a free GPU that costs $40/hour to power because you only need it for a few hours a month, it's probably cheaper to rent a more efficient GPU from someone who can power it at a lower cost.
Oh my reasoning was coming at this from a different angle: H200s were released in November of 2023, so they're over 2 years old at this point while still being valuable
I would not be surprised if the majority of H200s were manufactured in the past 12 or even 6 months.
It mainly depends on how much NVIDIA is overselling the improvements.
With adding RL functions, separating prefill and decode chips, nvfp4 and lots of other architectural changes efficiency of the most valuable tasks goes up as long as the algorithms don't change significantly.
Everything else can just stay on older chips.
The people who think that high end GPUs don't last for ~5 years (really it's 6) do not know what they are talking about. 6 years is likely too low. If the cooling is good and they don't fail early, most of these GPUs could still keep going past 6 years.
I'm convinced that anything with more than 80gb of VRAM will be worth it for closer to 10 years at this point.
Wait Corey Quinn is on El Reg now? That's awesome
Is there a reliable service that plots hourly price per GPU per cloud through time?
I’ve been working on computeprices.com as a side project for the last year to do just that.
Is there a graph view that charts all GPU prices on one graph?
If not I think the landing page should be just that with checkbox filters for all GPUs on the left that you can easily toggle all on/off to show/hide their line on the graph.
I was not expecting that the prices are going down. Makes sense as the hardware gets older but I always assumed the prices must be inflated given how much competition there is to make new datacenters
Yes i was surprised too. I think it's mostly newer models pushing older ones down. I think there's also a lot of competitive pressure in this market. And the GPU shortage is not really a thing anymore.
This is cool!
Would it be possible to add "Best Value" / "best average performance per dollar" type thing?
Good idea! I'll noodle on how to define that.
Not sure, but historically, AWS as far as I know has never raised prices on specific instance type usage like this. It makes sense that this would be the first attempt since it’s for apparently guaranteed capacity (vs the normal model of “if we’re out of capacity, too bad for you”).
That said, the real disturbing part of this is not so much the raising of the price for an extremely high-demand resource, but the utter lack of communication about it.
they have been doing that for awhile now (last few (2-3) years) for ebs (gp3) and ec2 (gen 8) instances.
The price shock that’s about to hit the AI industry is gonna be very fun to watch.
I keep asking our product team what the plan is when our API/Infra costs eventually skyrocket and our AI features become unsustainable, but there's absolutely zero preparedness.
We throw in all these cute little AI features to fill out marketing bullet points because they're basically free, but if they had real cost we're going to have no choice but to take them away.
Recently while interviewing for a SWE position, I was talking to the company's VP of engineering and expressed my concern over these LLM tools precisely because of the economics of it, because right now most if not all of these tools are being subsidized. Also, the environmental impact.
I was talking mainly about code generation tools, which can be completely shut down today without affecting production. Not even considering LLMs that are implemented in user facing features or customer service right now.
The guy's response: "oh but these things will surely get cheaper. Every technology gets cheaper with time. And about the environment, yeah... unfortunately there's not much we can do."
That's the level of forethought for a VP of engineering where I live.
That’s the level of forethought encouraged by investors and the CEO in all likelihood. VPs are just doing their jobs. Even if they might agree with you, that sort of sentiment is a quick way to get you on the CEO / board’s bad side. The incentives are perfectly aligned to keep shoving air into this bubble for the time being.
Don't worry your users will rejoice when you rip them out. I know when Atlassian gets hit by this and has to disable their AI features their products will be a little bit less unresponsive. They'll still suck, but they'll suck a bit less
But you gotta have that AI strategy or your good numbers won't go up. You can't be left behind in this business of business. I hear everyone is already moving on to post-agentic AI, whatever that means.
What is the real AI cost at which point this is an issue for a corporation you think? I contract at a place that gives $1,000/month budget per SWE (5 Max accounts) and I am sure if the price was 10x that no one would even blink.
It is basically cost-benefit analysis just like with any other cost (and there is always option to revert to local?)
Yeah, and I doubt that this is the last price hike. But I don't think this actually has a whole lot to do with ram prices. The thing is, AI cloud providers have generally been running at a massive loss, if you do the napkin maths, the procurement cost of these machines is equivalent to approximately a year of rent at these rates, for normal servers at AWS it is approximately a month.
I think a lot of usage will move to the cheap open weight Chinese models once there is an incentive to do that. Everything below the highest end frontier models are becoming commoditized and I suspect the commodity segment will pass the "good enough" bar for most applications.
It'll hit all of us, not just the AI "industry"
As a user of Claude I have been worried about this.
As someone who can't stand these tools, I'm ecstatic. Y'all have been hooked by the early venture-backed "scale up at all cost" period. You're gonna get squeezed so hard once investors start demanding returns on their investments -- and this whole craze is, hopefully, coming to an end.
Maybe don't hold your breath on that. Even if the price for AI tools would double, or tripple...that's still a very small part of the actual cost of an employee.
Sure, but employers will want a return on investment too. And productivity studies which aren't backed by these AI companies haven't exactly been promising. With how stingy some companies can be with the basics such as getting a laptop with enough RAM and drive space, a lot of the spending on AI tools for employees is clearly driven more by hype and the false promise of huge productivity gains than the normal expense approval process.
And plenty of people get hooked on these tools through using them for free or almost-free in their spare time. Those people will balk at huge price increases.
... and almost no company of any size above startup counts cost like that.
But I agree no point holding breath, whether somebody jumps on wagon or not won't change if price per query doubles, either its this massive productivity increase where costs of llms are a rounding error in overall costs or it isn't.
I’m betting on “isn’t”
You're likely going to stay seething for a while (even if bubble pops). the product market fit of these tools is so strong that companies paying tens of thousands per employee for them isn't really that big of a deal
That's what people widely claimed about Uber: it was toast once the investor subsidies stopped. Now it's quite profitable.
People will pay more. Claude Opus 4.5 is worth more than $20 per month, as is Gemini 3 Pro. These services keep getting better. Another three years of improvement, why shouldn't that command $30 or $40 instead?
$20 is ~$10 in the year 2000 per the BLS inflation calculator (or $1.25 when priced in gold). Nobody would have thought that was expensive for such utility. These are inexpensive tools at present.
At its worst, Uber had a net margin of -~60%. The AI labs are all running at least negative triple digit net margins, some running negative quadruple digit net margins. This is why AGI has been "forecasted" to death by the labs, because investors need the promise of infinite automation to stomach the losses.
Anyway, in this instance, what you received for $20 in 2025 will run you somewhere in the range of $60-$90 in 2027/2028. In the interim, you will likely see that $30-$40 of service gets you what cost $20 in 2025. The most likely avenue for this will be reduction in subscription user limits, and for API customers premiumization through substitution. The latter being a situation where what would be the next Claude Sonnet model is now sold as Claude Opus, for example.
The only way the math works for the consumer is if the user base has become dependent on the service instead of remaining in a conventional cost/benefit relationship.
I think you're ignoring that most users of AI currently aren't paying anything, nor would they. I believe the value of a Facebook user was $70 per year in 2023, for the US and Canada. Assuming that the AI companies could make twice that from ads, that's still only $10 - $12 per month, and even less in the rest of the world. Obviously there's going to be some business users as well, so they can cover some of the cost, but would also be responsible for a larger portion of the running cost.
The question should be how many free users can the AI companies convert.
The cost of an Uber has also gone way up, and they basically have a monopoly in many areas.
The cost of free users is much lower because they are served lite models, hit quota limits quickly, and can't soak up tokens by using agents. The main capacity usage is from agent loops, which is universally behind a paid tier.
To me this all seems like downstream from massive capital expenses on GPU purchases, driving up the demand for memory, etc. It begs the question of how much of the GPU capacity that has been sold/delivered actually being utilized? Are we paying across the board for a bunch of inventory sitting idle?
On Demand GPUs got cheaper because Azure exists. Capacity Blocks got 15% pricier because your migration plan doesn't
What's the most cost-effective way to run open source models using cloud infrastructure? AWS? Digital Ocean? salad.com? Lambda.ai? Vast.ai? Use case is infrequent and small usage, but needs to be able to scale to match increasing demand.
Modal is great for infrequent batches.
ssh admin.hotaisle.app
They increased the price of capacity blocks, not on demand. Capacity blocks pricing was promotional with a well defined end date from the day 1. And it was even lower than spot instance lot of times.
I'm just going to say: I bet it will eventually come out that RAM prices are yet another instance of illegal cartel collusion. For which all the RAM manufacturers have been convicted. More than once! If past incidents are any indication anytime there is a perceived "hot" market the RAM makers agree to limit production to current levels allowing market demand to increase prices, followed by production cuts if the market starts to soften.
GPU is a function of limited manufacturers and vendor lock-in combined with massive capex required to compete(). Like some (but not all) price inflation during and post COVID: rising prices can become a self-fulfilling prophecy. If all your customers expect prices to rise might as well meet their expectations and get while the gettin' is good.
() Like a new CPU architecture only worse. The amount of engineering required as "table stakes" increases exponentially while at the same time the manufacturing expertise does the same. This suddenly and rapidly raises the barrier to entry. Anyone who can survive the early squeeze can do quite well in such markets.
Demand for AI is growing. It makes sense they’re trying to increase profits.
Sad part is that most people actually won't notice.
That just means they should have done it sooner.
Did they raise prices because of increased or decreased demand? I can't help but wonder if things are overbuilt and being underutilized.
Demand for GPUs is increasing overall. So it's very likely that AWS GPU demand is rising.
AWS isn't the only company offering GPUs for rent so raising prices doesn't make sense unless demand is rising faster than supply.
The only real visibility we have into the demand side (kinda) are spot prices, and those seem to be flat or decreasing for GPU instances. Surely if there was this crazy, exponential demand, spot prices would reflect that.
Let's suppose that AWS is raising prices because demand is falling.
Ok, so AWS has extra capacity they need to sell but they're raising prices. Customers move off and go to another supplier. AWS has even more capacity they can't sell.
How does that make sense economically?
Do you think AWS is somehow able to make more money milking their existing customers than to sell at supply and demand equilibrium?
I wouldn't say exponential, I would also say it's likely that Amazon is projecting demand to rise faster than it is. (But it is still consistently rising, and with demand rising they would prefer to overestimate than underestimate.)
Whether it's an increase in demand, decrease in supply, or some of each is the question. Prices don't normally go up when demand drops at an over- or steady supply
Random theory:
- it is a practical test of price elasticity
(if demand doesn’t drop much, then it means profit can be increased more)
I wonder how far we are down the track of enshittification of cloud providers... and how much more is to come.
I guess somewhere this year having your own GPU might be cheaper than renting.
When was buying a device from the manufacturer not cheaper than paying a middleman to manage rentals of said device?
When you need to scale up and down dynamically.
Then the claim above can never be falsified because there will always be a value you can plug into the equation at which it's cheaper to rent (like at 50% occupancy, at 5%, at 0.5%... so long as you're not the only person in the world who wants to use the device being offered for loan). I would assume we're talking about a common workload where one actually needs the device that one considers purchasing for some reasonable fraction of the time and not less than half the time
So in summary:
- own the hardware, if you have constant load and have enough expertise (personally or a team) to use/maintain it at your required reliability level
- rent the cloud, if your usage is opportunistic/ spike’y/rapidly changing
- also rent the cloud, if required expertise/maintenance would cost you more than the hardware (if you want to have gazillion of nines, you need somebody who’s there to deal with smoking stuff, in several locations)
That sounds like good common sense, though I'd reword the "constant" with like <30% occupancy because cloud premiums are rather large
My question was (to what ozim wrote), how's that new in 2026
It already is. It's why I launched this very recently: https://it.lanified.com/Solutions/Managed-AI-Solutions
I wonder what will happen with cloud GPU prices once the AI bubble pops. Will they keep having high prices, or will they drop prices significantly ?
My guess is, the bubble popping means the demand wasn't real. No amount of price reduction will motivate it. Anybody who has to get theirs will be in for a steep premium.
I like the stock
I have a pair of Dell T7500 units for “slow and simple” self-hosting, and just snagged a trio of 1Tb/64Gb 2018 Mac Minis to join my own surplussed Mac Mini of the same vintage as single-purpose machines.
I have a gigabit symmetrical SOHO fibre connection. It’s time I brought everything back under one roof except for DNS.
I'm tired of AWS full of hidden fees and tricking you into things that you didn't know you were paying for until a month or two goes by.
Everything is getting monopolized (oligopolized) for rent extraction! Homes, healthcare, energy, compute. 'Capitalism' FTW! Coming up next: water, air.
i would kindly like to curse openai for making the rest of world worse for everyone else. boohoo your ai is so smart? make it smaller.
Even the cloud won’t save you.
Maybe it might be a good idea to squash it with any legal avenue, especially antitrust and data privacy laws that require reasonable and non discriminatory access by end consumers to self-maintained and self-hosted infra?
> AWS has spent two decades conditioning customers to expect prices only ever go down. That expectation is now broken.
So long for amazon’s “earn trust” leadership principle
So rather than free market capitalism where AWS would have actually lowered prices after investing heavily in a new datacenter-focused line of Huawei GPUs, the US government took steps to stop capitalism and competition, gave NVIDIA lots of free money (USA! USA!) and now prices are going up.
Recall pictures of Bezos smiling at various Trump events.
Everything Trump admin has done so far with tech reduces competition and tries to pick winners. This price increase is just the beginning.
"oh you dont' want to use Bedrock? You want to do something that might vaguely compete with our offering on our infra? Fck you, pay me"
Buy your own hardware while you still can.
Imagine trying to build a business in this environment.
What kind of business are you talking about? If you're chasing a moving target, or worse, building little more than an AI Wrapper, you've willingly signed up for that risk.
And insert random tariff rug pulls. Seems like the only ones spending money are the incestuous AI club and already rich people.
But weren't the AWS shills saying that AWS only reduces pricing? I was trusting them blindly! /s
Historically this has been mostly true with the exception of the recent cost per public IPv4 address.
They made oauth client (cognito oauth credential grant) a subscription, raising our bill X10. Just for having the client registered in the dashboard, even without using that oauth client application. we saw thousands of cost, for zero usage.
Yeah, the tiered pricing changes for M2M[0] in 2024 didn't make much sense to me. I guess I don't understand their COGS, though.
0: https://aws.amazon.com/about-aws/whats-new/2024/05/amazon-co...
it's last call to get a GPU while there is stock left, no hopes those will become any more affordable in future considering the ongoing DRAM crisis. I am going to get 5070 Ti to replace my 1080 Ti which did a great service over all these years, paid it's worth back two times over during monero mining days and still can play a lot of games on high settings - recently finished RDR2. I sometimes wonder if Nvidia will be still making GPUs useful for gamers in a year or two or just completely shift it's focus to AI accelerators regular people will have no use for. RIP affordable computing at home.
I feel like the whole memory situation might also utterly screw up the pricing and value proposition of the upcoming Intel Arc B770.
I got a B580 cause everything else was out of my price range at the time (9060 XT 16 GB only seems to have 3 video outputs and I have no experience with daisy chaining to drive my 4 monitors and the 5060 Ti 16 GB pricing here is just sad).
I fear that companies will rise the prices during this shortage… and just never properly lower them again.