The economics of AI are terrible. The exact inverse of what VCs normally say they strive for.
It’s capital intensive, doesn’t easily scale without linear costs, is basically a commodity, seemingly has negative cashflow forever, and customers aren’t willing to pay what it actually costs. It’s basically the worst tech business model and yet somehow VCs caught a spell and threw every available penny into the fire.
Watching this all unravel is going to be interesting.
> It’s capital intensive, doesn’t easily scale without linear costs, is basically a commodity, and customers aren’t willing to pay what it actually costs
What are some VC funded companies that come to mind that do not fit this profile?
Yeah, which IMO is actually the problem. They got so addicted to the sugar hit of SaaS they lost their way. The game change, the risk profile dropped drastically, but they kept their own economics and upside the same and made an absolute fortune.
The type of investment called out by the start of this thread sounds exactly like the type of speculative "we don't even know if this is technically possible, much less if there's even a market for it" that Silicon Valley was founded on.
> speculative "we don't even know if this is technically possible, much less if there's even a market for it" that Silicon Valley was founded on
yes and that's the entire point of Silicon Valley and why you can't have another Silicon Valley in Europe or Asia where almost no one is willing to suspend belief and reality like this.
VCs become rich not because they are smarter and better than Warren Buffet but because they are willing to place bets that Buffet would never take on.
If Uber operated like most AI startups its business model would be to buy new cars every 6 months, sign agreements with every automaker to buy every car they make for the next 5 years despite having neither the cash nor cash flow to pay for that, and sell rides for a fraction of what all this costs them. All while providing a service only moderately better than the boring taxi service that costs the same price and that taxi service just released an app to hail rides. It would then claim it’s worth 100x the valuation of the taxi service.
> If Uber operated like most AI startups its business model would be to buy new cars every 6 months, sign agreements with every automaker to buy every car they make for the next 5 years, and sell rides for a fraction of what all this costs them. All while providing a service only moderately better than the boring taxi service that costs the same price and that taxi service just released an app to hail rides. It would then claim it’s worth 100x the valuation of the taxi service.
Uber and Lyft literally "sell rides for a fraction of what all this costs them" upto early 2025. I don't know how old you are but I started to use them back in 2012. I was one of their early users when you could sit next to the driver. From then upto 2025, Uber and Lyft used to literally cost me less than what just gas would cost me to drive the same distance.
This is true too: "It would then claim it’s worth 100x the valuation of the taxi service" - there was literally a website (might be up even now) that literally made this argument and why Uber was doomed to fail anytime now.
Your first two points are also true for Uber Lease - a product that was rolled out for drivers - they could choose to lease specific models and Uber in turn signed agreements with specific automakers. An exact parallel for these points, esp. the circular financing is hard to make because they are not even the same product but I imagine one would agree, on the whole, we are in the conception stage of LLMs and this industry is brand new and figuring things out.
This same pattern had happened when PCs were becoming mainstream. That was 40 years ago so memory of it might be faint or lost. Also VCs weren't really a huge deal back then either but the fundamentals were the same.
We are all overreacting over a nascent industry, upset that it's increasing prices for everyone and pushing humans out of jobs. This is understandable but the conclusion is wrong. We are letting emotions drive decisions.
You make some valid points, but the bigger fact remains that Uber survived because its model was not capital intensive. Yes it burned cash for a while but on one-off costs like setting up new markets. Had it signed contracts to be buying new cars very 6 months in the way AI startups have committed
to future expenses it would have gone bankrupt.
> that Uber survived because its model was not capital intensive. Yes it burned cash for a while
I would say you are either very young or unfamiliar with Uber as a business. Uber was and continues to be extremely capital intensive and you need to qualify your "while". Uber burned cash for 14 years.
The one argument I can agree with is Uber didn't burn as much cash as Anthropic or OpenAI does but they are very different businesses with very different impact on humanity.
OK and which of the companies that I listed above do not fit the profile below and specifically which portions:
> It’s capital intensive, doesn’t easily scale without linear costs, is basically a commodity, and customers aren’t willing to pay what it actually costs
I think the closest analogy we have of the AI boom is the mid 90s browser wars.
First all the money went into the browsers like Netscape. Then the browser got commoditized, but that didn't mean the value went away. A better business model came along, and lo and behold, the Chrome browser went open source. The browser became free, and the scale of adoption meant a lot of value could still be extracted.
A free browser product didn't mean the value of Chrome went away. Rather the opposite. It's a gateway for buying stuff through Google search.
Marc Andressen was the "Mosaic browser" guy before he was the investor guy, so the folks in the early days did not lose all their money. Quite the opposite. Many are now billionaire oligarchs.
LLMs will likely go the way of the browser... commoditized, then free, then appropriated to shove ads into your face. I predict a lot of safety and interpretability research will be leveraged to one day push ads, just like "Golden Gate Claude" could be manipulated to think of itself as the Golden Gate Bridge.
> I think the closest analogy we have of the AI boom is the mid 90s browser wars.
Depends on how old you are. This analogy makes sense if you were a teenager in the mid 90s. I am older than that so to me, this same pattern had happened when PCs were becoming mainstream. That was 40 years ago so memory of it might be faint or lost. Also VCs weren't really a huge deal back then either but the fundamentals were the same.
We are all overreacting over a nascent industry, upset that it's increasing prices for everyone and pushing humans out of jobs. This is understandable but the conclusion is wrong. We are letting emotions drive decisions.
On the models yes. All indications are models will basically be a free commodity soon.
On the datacenter side the closest analogy is the fiber buildout of the .com bust. While that fiber ended up being useful decades later it’s unclear what value a bunch of half-completed datacenters will bring if their owners go bust.
Am I understanding this correctly as an idiomatic reference to an enchantment or curse? If so, I'd be interested to know where the expression "catching a spell" comes from.
(I'm really interested in the phrasing and I hope you don't take this the wrong way. For me personally, the metaphorical idea that someone with inexplicable behavior has been bewitched is a normal thing to express, but I'd be more likely to say they "fell under" a spell than that they "caught" it. I believe in Mandarin the usual idiom uses 中 "be struck [by a projectile]".)
I think it's more nuanced than that. Yes, it's true that the economics are horrible. But it's also true that consumer startups are hard. And then when you tack on legal fees and product changes due to policy concerns, it becomes impossible. Maybe the lesson is that the world isn't ready for a product that targeted minors with fake friends based on existing IP. Maybe it never will be.
> It’s not clear whether AI giants like OpenAI and Anthropic, or the AI industry in general have gotten too big or too important to fail, but Character’s bankruptcy shows that the runway for second-tier players is over.
But the current title definitely baits more people into the article
Does it though? It arguably illustrates how the economics of building custom specialized small sub-frontier LLMs and building an service around them that can easily be replicated by open tools wrapping either frontier LLMs for equal or better quality or personally-controlled open LLMs for privacy doesn't work, especially in a market where privacy is a selling point.
But I think that it is dubious whether that is a good proxy for “the economics of AI” broadly.
Which I am not saying are good—I just don't think Character.ai has much probative value on the broad question on way or the other.
It's a bit silly to write an article like this without mentioning the company was hollowed out by Google in 2024.
It is difficult to make a company in a fast-growing sector work when all of your technical staff and leadership are working for Google, and your remaining staff can see the writing on the wall and are checking out.
For what it's worth Michael Burry is pretty much short on the entire thing. At some point(I think mid 2027) those data centers bankrolled by loans will have to start paying rent to investors who bought those loans.
If companies that are basing X% of their future earnings on new AI based products don't pay up a bunch of those data centers are not gonna be re-leased. It's gonna be a problem. An seasoned analyst can take educated guesses as to what that X is: investments made in pure chips by Nvidia/AmD vs how much capital investment— not VC — Faang(i.e ones that are just using the chips for building software) are putting into them. The difference between investment devoted to pure DC build out minus what they are spending is a good proxy to understand the overinvestment.
The trade surplus of Taiwan, China, Korea, Japan etc including chips has been sky rocketing. This is like Bill of materials(BoM) and it's at least 3-4 trillion in chips for AI alone, summing it up over the past 3 years.
Not worth much. Hasn’t Burry called a number of bubbles since the great financial crisis, none of which have come to pass? Seems like a broken clock to me (he’ll still be right twice a day.)
The headline doesn’t match the story. It is TBD whether the economics of AI actually work or not. A small company (that wasn’t vertically-integrated) going bankrupt doesn’t demonstrate a whole lot, other than that some paths aren’t viable for an AI-based business.
Character.ai is hardly representative of the AI market. It is at best a tiny sliver. There's coding tools, business process applications and more. Now the AI market likely has some challenges around financing build outs given recent inflation driven interest rates etc. But that's very different from what afflicted character.ai
There are a tonne of AI companion apps out there now, lot of them focused on nsfw, it's going to be interesting to see if character.ai is an isolated case, or if it's a domino effect.
I doubt the NSFW-focused ones have taken on venture funding. They're probably leaner shops with a dramatically smaller userbase that is more likely to actually pay.
Before Grok locked down its video generation tools the amount of high-quality and stunningly realistic looking porn you could generate was insane. You’re going to need something like that in order to make billions of dollars in this space. I don’t think anyone is going to take the time/money and deal with all the legal implications that comes from training a video generation model like though.
Elsewhere on this page I argued how this reminds me of the PC wars of the mid 90s.
We are all overreacting over a nascent industry, upset that it's increasing prices for everyone and pushing humans out of jobs. This is understandable but the conclusion is wrong. We are letting emotions drive decisions.
Well said. The anti-AI sentiment is strong in a few camps on HN. Praying that AI will go away so as a software engineer, only familiar with the SDLC, you become relevant again is not a winning strategy. AI is like a soup with many bubbles, some will pop while new ones begin.
> Character.ai’s bankruptcy is the one of the first dominos to fall in the long-foreseen AI bubble.
Kind of speculative, and a very bold speculation at that. I don’t have anything else to say, but one company that was obviously mismanaged doesn’t really get my attention.
Groups desperate to prove an increasingly absurd suggestion are latching onto this because actual evidence of overinvestment or of broader correlated issues is thin on the ground.
News is less about inspection and more about speculation these days. Ironically, journalists are largely paid the same way Ai is, per word. The incentives might be better if both were "salaried"
That's overly generalized. Perhaps they became a buyer after google stripped top staff, but before that they were certainly a producer, and had many advances in cutting edge inference at the time.
The conclusion of the article is wrong - that the Economics of AI Don't Work.
A better report would have been how the "Economics of AI Wrappers Don't Work".
Character.ai was likely paying frontier LLM API rates (which are known to be ridiculously expensive at 90% margins) and then had their own R&D plus lawsuit overhead. Most people can use a $20 Claude or ChatGPT sub to build a completely custom Character.ai. Once Character.ai employed MBA Strategy #3 (aka enshittification), churn killed the company (as it should)
---
What doesn't work is attempting to build an AI Wrapper. Those companies had a good run 2022-2025, but with how good both open weights and cheap frontier models have become, AI Wrappers can no longer demand fat margins.
I'm sure people will now argue what exactly AI Wrappers are in the comments below, but for now I'm more interested in understanding what the HN hivemind's take is on the topic as a whole.
Known to someone who doesn't trade on a heavily AI indexed market (in 2026 all companies, even Tyson or Chevron is an AI play but not as heavily AI indexed as those say 2 degrees away from Anthropic or OpenAI) - likely not.
So if the point you're making is your typical barber doesn't know about this, I am guilty.
If you want to do your own research into this, there are a lot of private reports and publications on the topic.
If you don't want to spend any money, there is even a recent public analysis using a 92% API gross margin.
Anthropic's October 5, 2026 analysis also points to a 92% API gross margin that SemiAnalysis based their report on.
OpenAI reported paid-user compute margin reached 70% in October 2025 - a year ago.
DeepSeek has similar reports. The frontier labs are comfortably in the fat inference margin territory.
Sort of. Big fan of 404media usually, but they need to "look through the bubble" to understand this can and likely will be replicated on device when local models and inference catch up, which they will eventually (Apple Intelligence, Hermes, OpenClaw, etc). Slap iCloud and Google Storage sync on the app, and you've got your own character bots in your device ecosystem.
"eventually" is doing a lot of work here. Like sure, yes, eventually. But that could be later this year or it could be 50 years from now. We don't know the answer to that. Assuming it's going to happen in a timeline short enough to matter for this conversation isn't worthwhile. It's like saying eventually we'll have cold fusion, so why is everyone skeptical?
It may be, I didn't say otherwise. I said we don't know for a fact that this moment will come in the near future, so just handwaving that it's obvious is folly.
There are already llms on phones now, they are being embedded / distributed by iOS and Android. You can find HN stories and comments about the struggle to disable or the joy of having them. Tons of Show HN about getting models to run on all sorts of devices, including phones
There's a difference between having "an llm" on a small device and having an llm of the class being discussed in this comment thread.
As I said, it may happen within a year. But maybe life takes a different direction. The person to whom I responded said "eventually", and yes, "eventually" it will happen. but that's not a useful time horizon to plan around because it could be a lot of things.
If you build your app today to use an LLM API with credits so the economics works (or support existing LLM providers who can provide an API key for use within the app, to not expose the app to negative usage economics), but it has the capability to switch to a local API using existing known primitives, I think that is very useful in this context, no?
The hard part is refining the UX, imho, while supporting pluggable LLM support on the backend. If the future never arrives of cheap on device inference, still works if you pay as you go and for those users willing to pay for their usage directly through their cloud inference provider of choice or through the app (unless the future is "cloud inference is unsustainable at consumer price points and on prem inference never arrives"). Just my two cents.
This is fundamentally what Apple did with new Siri, powered by Gemini today. They will eventually flip the switch to an in house solution. Agree the wildcard that still remains is "can this be done on devices eventually?"
> Apple weighed using its own AI models for the LLM version of Siri , and also tested options from OpenAI and Anthropic, but it decided to go with Gemini after deciding Anthropic's fees were too high. Apple already has a partnership with Google for search results, with Google paying Apple around $20 billion per year to be the default search engine option on Apple devices.
> Though Apple is planning to rely on Google AI for now, it plans to continue working on its own models and will transition to an in-house solution when its LLMs are capable enough. Apple is already working on a 1 trillion parameter cloud-based model that could be ready as soon as 2026. Apple is unlikely to publicize its arrangement with Google while it develops in-house models.
So, your contention is that 404media isn't taking into consideration a hypothetical where they wouldn't burn millions of dollars each month running something.
Yes, which I say as a subscriber and donor to 404media (they do very good journalism work imho) and someone who leans more towards AI skeptic while being a heavy user in specific domains. "It's not a business, it's a feature."
Durable storage + sufficiently performant local LLM hardware + appropriate models [last two are wildcards and hard to predict the forward curve imho] = this experience locally. Remember when mobile video hardware acceleration wasn't good? And now, it is very good.
None of that has any relevance to what is happening right now. Character.ai is in chapter 11 bankruptcy, and "oh they should have done it differently, and 404media isn't taking into consideration a completely unproven hypothetical technology that wouldn't have bankrupted them" is not a useful discussion, nor a good critique of their coverage.
> Character.ai's Bankruptcy Documents Show the Economics of AI Don't Work
My comments:
"I don't agree, here is why."
This specific model doesn't appear to be economically sustainable and/or commercially viable. That is a reasonable statement and assertion, based on the evidence presented. Does not mean the economics of AI in general do not work. That's a very broad statement! Nuance, words mean things.
Therefore, we should ask ourselves and discuss, "Under what circumstances could|would the economics work? Are there any? Or are there none?" We are discussing the problem space! Be more curious.
Some (most?) journalist skimp on the nuance. Character.ai is not representative of the entire Ai ecosystem and market. There are a lot of other factors that played a role here than the Ai economics themselves.
It seems harder to sell on device ai as a subscription, I would guess? although the much lower inference costs mean maybe you could just sell it as a one time purchase
By on device you mean running locally on a mobile device like an iPhone? If so we're a really long way from a chatbot that's engaging to talk to. There's just not enough processing power to do it
No-one's going to want to talk to a 7B model for more than 3 minutes
I think we're a breakthrough away. I don't think processing power is an issue. A 7B model has a lot of cruft (like knowledge of code syntax) that a character sim doesn't need. It should be possible to extract just the relevant parts of a larger and more powerful model and run it with the same resources you would run a general 7B model. We just don't have the ability to isolate and extract weights like that yet.
Yes, current gen hardware was not built to run such models, and are obviously inefficient and bad at it.
Significant investment is being spent on producing hardware that is more suitable for running mid-sized (100B-ish) models. Starting with flash chips with wide enough interfaces to provide the weights (NAND flash can be ~5x more energy-efficient for reads than DRAM), and going into more exotic solutions like compute-in-memory.
Yes, iphone, ipad, mac. Maybe your phone makes the inference requests to either Apple servers (requiring iCloud+ and credit based?) or your Mac (Tailscale equivalent).
My assumptions are that inference costs continue to decline (and/or the difficulty level of routing requests declines) and hardware acceleration continues into the future. These assumptions might be wrong, but that is the idea the thesis is based around.
(have built workflows and harnesses that currently target LLM provider APIs under the assumption I'll be able to point "locally-ish" eventually, my career had a strong focus on distributed computing about 10-15 years ago, think managing data and work clusters for LHC detector data processing, and I grew up on hobbyist distributed.net and BOINC projects, all the lenses I see this problem through; queues, jobs, sync vs async performance optimization, etc; recovered sysadmin|infra/devops engineer a lifetime ago)
Isn't the majority of AI compute training and not inference? Saying "It doesn't matter if the AI companies fold because I can run models on my computer" is like saying "I don't care if the farm closes because I get my food from the grocery store"
You are correct, but I think it's much more like consumer culture doesn't and never will care about any of these details.
Specifically with regard to AI, the consumers aren't even aware of the "grocery store" either. They're being served from a silver platter. What matters most is the quality of results, not the methods. You can disagree all you want for whatever (good and compelling) reasons, but it doesn't change the fact that many people do not even buy groceries now and doordash everything.
If it turns out that you can put a mobile kitchen in their backyard that cooks up everything without them realizing, it makes no difference to the consumer.
Realistically, yes there's a finite amount of training data, hence the plateau in capability we've seen for at least a year. I think we're likely to see a consolidation and compression of all of this into local models. The economics don't support the expected growth of the current arrangement either. There simply isn't that much distinct data out there as initially assumed.
I have been seeing this more and more with “top” tier publications but “Character.ai’s bankruptcy is the one of the first dominos to fall in the long-foreseen AI bubble.”
A simple grammatical mistakes like this should have been caught on review/edit phase.
Either that or writers are relying too much on AI.
AI doesn’t make mistakes like that generally, I’ve almost never seen the grammar butchery from them I’ve seen from even major publications in recent years. I wish they’d let AI copy edit.
The economics of AI are terrible. The exact inverse of what VCs normally say they strive for.
It’s capital intensive, doesn’t easily scale without linear costs, is basically a commodity, seemingly has negative cashflow forever, and customers aren’t willing to pay what it actually costs. It’s basically the worst tech business model and yet somehow VCs caught a spell and threw every available penny into the fire.
Watching this all unravel is going to be interesting.
Keep dreaming. AI is absolutely booming!
Once the models are good enough, the companies can stop selling access to models and start directly competing with the rest of the economy.
> It’s capital intensive, doesn’t easily scale without linear costs, is basically a commodity, and customers aren’t willing to pay what it actually costs
What are some VC funded companies that come to mind that do not fit this profile?
Literally any software company, which is what VCs loved for 20 years.
Yeah, which IMO is actually the problem. They got so addicted to the sugar hit of SaaS they lost their way. The game change, the risk profile dropped drastically, but they kept their own economics and upside the same and made an absolute fortune.
The type of investment called out by the start of this thread sounds exactly like the type of speculative "we don't even know if this is technically possible, much less if there's even a market for it" that Silicon Valley was founded on.
> speculative "we don't even know if this is technically possible, much less if there's even a market for it" that Silicon Valley was founded on
yes and that's the entire point of Silicon Valley and why you can't have another Silicon Valley in Europe or Asia where almost no one is willing to suspend belief and reality like this.
VCs become rich not because they are smarter and better than Warren Buffet but because they are willing to place bets that Buffet would never take on.
any SAS company 4 years ago
Is Uber a SAAS? Palantir? AirBnB? Netflix? Amazon/AWS?
If Uber operated like most AI startups its business model would be to buy new cars every 6 months, sign agreements with every automaker to buy every car they make for the next 5 years despite having neither the cash nor cash flow to pay for that, and sell rides for a fraction of what all this costs them. All while providing a service only moderately better than the boring taxi service that costs the same price and that taxi service just released an app to hail rides. It would then claim it’s worth 100x the valuation of the taxi service.
> If Uber operated like most AI startups its business model would be to buy new cars every 6 months, sign agreements with every automaker to buy every car they make for the next 5 years, and sell rides for a fraction of what all this costs them. All while providing a service only moderately better than the boring taxi service that costs the same price and that taxi service just released an app to hail rides. It would then claim it’s worth 100x the valuation of the taxi service.
Uber and Lyft literally "sell rides for a fraction of what all this costs them" upto early 2025. I don't know how old you are but I started to use them back in 2012. I was one of their early users when you could sit next to the driver. From then upto 2025, Uber and Lyft used to literally cost me less than what just gas would cost me to drive the same distance.
This is true too: "It would then claim it’s worth 100x the valuation of the taxi service" - there was literally a website (might be up even now) that literally made this argument and why Uber was doomed to fail anytime now.
Your first two points are also true for Uber Lease - a product that was rolled out for drivers - they could choose to lease specific models and Uber in turn signed agreements with specific automakers. An exact parallel for these points, esp. the circular financing is hard to make because they are not even the same product but I imagine one would agree, on the whole, we are in the conception stage of LLMs and this industry is brand new and figuring things out.
This same pattern had happened when PCs were becoming mainstream. That was 40 years ago so memory of it might be faint or lost. Also VCs weren't really a huge deal back then either but the fundamentals were the same.
We are all overreacting over a nascent industry, upset that it's increasing prices for everyone and pushing humans out of jobs. This is understandable but the conclusion is wrong. We are letting emotions drive decisions.
You make some valid points, but the bigger fact remains that Uber survived because its model was not capital intensive. Yes it burned cash for a while but on one-off costs like setting up new markets. Had it signed contracts to be buying new cars very 6 months in the way AI startups have committed to future expenses it would have gone bankrupt.
> that Uber survived because its model was not capital intensive. Yes it burned cash for a while
I would say you are either very young or unfamiliar with Uber as a business. Uber was and continues to be extremely capital intensive and you need to qualify your "while". Uber burned cash for 14 years.
The one argument I can agree with is Uber didn't burn as much cash as Anthropic or OpenAI does but they are very different businesses with very different impact on humanity.
Yes?
OK and which of the companies that I listed above do not fit the profile below and specifically which portions:
> It’s capital intensive, doesn’t easily scale without linear costs, is basically a commodity, and customers aren’t willing to pay what it actually costs
I think the closest analogy we have of the AI boom is the mid 90s browser wars.
First all the money went into the browsers like Netscape. Then the browser got commoditized, but that didn't mean the value went away. A better business model came along, and lo and behold, the Chrome browser went open source. The browser became free, and the scale of adoption meant a lot of value could still be extracted.
A free browser product didn't mean the value of Chrome went away. Rather the opposite. It's a gateway for buying stuff through Google search.
Marc Andressen was the "Mosaic browser" guy before he was the investor guy, so the folks in the early days did not lose all their money. Quite the opposite. Many are now billionaire oligarchs.
LLMs will likely go the way of the browser... commoditized, then free, then appropriated to shove ads into your face. I predict a lot of safety and interpretability research will be leveraged to one day push ads, just like "Golden Gate Claude" could be manipulated to think of itself as the Golden Gate Bridge.
> I think the closest analogy we have of the AI boom is the mid 90s browser wars.
Depends on how old you are. This analogy makes sense if you were a teenager in the mid 90s. I am older than that so to me, this same pattern had happened when PCs were becoming mainstream. That was 40 years ago so memory of it might be faint or lost. Also VCs weren't really a huge deal back then either but the fundamentals were the same.
We are all overreacting over a nascent industry, upset that it's increasing prices for everyone and pushing humans out of jobs. This is understandable but the conclusion is wrong. We are letting emotions drive decisions.
On the models yes. All indications are models will basically be a free commodity soon.
On the datacenter side the closest analogy is the fiber buildout of the .com bust. While that fiber ended up being useful decades later it’s unclear what value a bunch of half-completed datacenters will bring if their owners go bust.
> and yet somehow VCs caught a spell
Am I understanding this correctly as an idiomatic reference to an enchantment or curse? If so, I'd be interested to know where the expression "catching a spell" comes from.
(I'm really interested in the phrasing and I hope you don't take this the wrong way. For me personally, the metaphorical idea that someone with inexplicable behavior has been bewitched is a normal thing to express, but I'd be more likely to say they "fell under" a spell than that they "caught" it. I believe in Mandarin the usual idiom uses 中 "be struck [by a projectile]".)
FOMO
I think it's more nuanced than that. Yes, it's true that the economics are horrible. But it's also true that consumer startups are hard. And then when you tack on legal fees and product changes due to policy concerns, it becomes impossible. Maybe the lesson is that the world isn't ready for a product that targeted minors with fake friends based on existing IP. Maybe it never will be.
That seems even less nuanced: bad economics AND they had a potentially illegal product.
Perhaps it's more that "consumer startups that want to become a unicorn" are hard.
So a perfect military expense.
> I think it's more nuanced than that.
So did they, they just left it for the conclusion
> It’s not clear whether AI giants like OpenAI and Anthropic, or the AI industry in general have gotten too big or too important to fail, but Character’s bankruptcy shows that the runway for second-tier players is over.
But the current title definitely baits more people into the article
People made the exact same argument for years where Facebook stock stayed around 20 bucks post ipo. I did too and sold like an idiot!
Does it though? It arguably illustrates how the economics of building custom specialized small sub-frontier LLMs and building an service around them that can easily be replicated by open tools wrapping either frontier LLMs for equal or better quality or personally-controlled open LLMs for privacy doesn't work, especially in a market where privacy is a selling point.
But I think that it is dubious whether that is a good proxy for “the economics of AI” broadly.
Which I am not saying are good—I just don't think Character.ai has much probative value on the broad question on way or the other.
It's a bit silly to write an article like this without mentioning the company was hollowed out by Google in 2024.
It is difficult to make a company in a fast-growing sector work when all of your technical staff and leadership are working for Google, and your remaining staff can see the writing on the wall and are checking out.
The company was hallowed out broadly because it was a dead startup walking. It just took this long for an official bankruptcy to happen.
For what it's worth Michael Burry is pretty much short on the entire thing. At some point(I think mid 2027) those data centers bankrolled by loans will have to start paying rent to investors who bought those loans.
If companies that are basing X% of their future earnings on new AI based products don't pay up a bunch of those data centers are not gonna be re-leased. It's gonna be a problem. An seasoned analyst can take educated guesses as to what that X is: investments made in pure chips by Nvidia/AmD vs how much capital investment— not VC — Faang(i.e ones that are just using the chips for building software) are putting into them. The difference between investment devoted to pure DC build out minus what they are spending is a good proxy to understand the overinvestment.
Here is some data for example from a macro standpoint https://x.com/Brad_Setser/status/2110415047268933826?s=20
The trade surplus of Taiwan, China, Korea, Japan etc including chips has been sky rocketing. This is like Bill of materials(BoM) and it's at least 3-4 trillion in chips for AI alone, summing it up over the past 3 years.
> for what it’s worth
Not worth much. Hasn’t Burry called a number of bubbles since the great financial crisis, none of which have come to pass? Seems like a broken clock to me (he’ll still be right twice a day.)
Yes, separate from the “do AI startups have a viable business?” issue is the datacenter debt bomb ticking on the horizon.
The headline doesn’t match the story. It is TBD whether the economics of AI actually work or not. A small company (that wasn’t vertically-integrated) going bankrupt doesn’t demonstrate a whole lot, other than that some paths aren’t viable for an AI-based business.
Character.ai is hardly representative of the AI market. It is at best a tiny sliver. There's coding tools, business process applications and more. Now the AI market likely has some challenges around financing build outs given recent inflation driven interest rates etc. But that's very different from what afflicted character.ai
There are a tonne of AI companion apps out there now, lot of them focused on nsfw, it's going to be interesting to see if character.ai is an isolated case, or if it's a domino effect.
I doubt the NSFW-focused ones have taken on venture funding. They're probably leaner shops with a dramatically smaller userbase that is more likely to actually pay.
Before Grok locked down its video generation tools the amount of high-quality and stunningly realistic looking porn you could generate was insane. You’re going to need something like that in order to make billions of dollars in this space. I don’t think anyone is going to take the time/money and deal with all the legal implications that comes from training a video generation model like though.
The economics of AI? That’s like saying “the economics of the internet”.
Pets.com failed, therefore the internet is doomed. No one will ever make a successful business out of it.
Bingo!
Elsewhere on this page I argued how this reminds me of the PC wars of the mid 90s.
We are all overreacting over a nascent industry, upset that it's increasing prices for everyone and pushing humans out of jobs. This is understandable but the conclusion is wrong. We are letting emotions drive decisions.
Well said. The anti-AI sentiment is strong in a few camps on HN. Praying that AI will go away so as a software engineer, only familiar with the SDLC, you become relevant again is not a winning strategy. AI is like a soup with many bubbles, some will pop while new ones begin.
> Character.ai’s bankruptcy is the one of the first dominos to fall in the long-foreseen AI bubble.
Kind of speculative, and a very bold speculation at that. I don’t have anything else to say, but one company that was obviously mismanaged doesn’t really get my attention.
Groups desperate to prove an increasingly absurd suggestion are latching onto this because actual evidence of overinvestment or of broader correlated issues is thin on the ground.
News is less about inspection and more about speculation these days. Ironically, journalists are largely paid the same way Ai is, per word. The incentives might be better if both were "salaried"
Being paid per word doesn't cover all of the incentives. One needs to remain employed, which is the same incentive that salaried employees have.
character.ai is a buyer of tokens, not a producer of tokens. very different businesses
That's overly generalized. Perhaps they became a buyer after google stripped top staff, but before that they were certainly a producer, and had many advances in cutting edge inference at the time.
The conclusion of the article is wrong - that the Economics of AI Don't Work.
A better report would have been how the "Economics of AI Wrappers Don't Work".
Character.ai was likely paying frontier LLM API rates (which are known to be ridiculously expensive at 90% margins) and then had their own R&D plus lawsuit overhead. Most people can use a $20 Claude or ChatGPT sub to build a completely custom Character.ai. Once Character.ai employed MBA Strategy #3 (aka enshittification), churn killed the company (as it should)
---
What doesn't work is attempting to build an AI Wrapper. Those companies had a good run 2022-2025, but with how good both open weights and cheap frontier models have become, AI Wrappers can no longer demand fat margins.
I'm sure people will now argue what exactly AI Wrappers are in the comments below, but for now I'm more interested in understanding what the HN hivemind's take is on the topic as a whole.
They weren't paying frontier rates. They had frontier-level staff to train their own bespoke, hyper-optimised[0] LLMs until they didn't.
[0]: https://blog.character.ai/optimizing-ai-inference-at-charact...
> which are known to be ridiculously expensive at 90% margins
Is that known?
depends on the person.
Known to someone who doesn't trade on a heavily AI indexed market (in 2026 all companies, even Tyson or Chevron is an AI play but not as heavily AI indexed as those say 2 degrees away from Anthropic or OpenAI) - likely not.
So if the point you're making is your typical barber doesn't know about this, I am guilty.
If you want to do your own research into this, there are a lot of private reports and publications on the topic.
If you don't want to spend any money, there is even a recent public analysis using a 92% API gross margin.
Anthropic's October 5, 2026 analysis also points to a 92% API gross margin that SemiAnalysis based their report on.
OpenAI reported paid-user compute margin reached 70% in October 2025 - a year ago.
DeepSeek has similar reports. The frontier labs are comfortably in the fat inference margin territory.
Do you have information to the contrary?
Related:
Character.ai has initiated voluntary Chapter 11 bankruptcy
https://news.ycombinator.com/item?id=50068036
Sort of. Big fan of 404media usually, but they need to "look through the bubble" to understand this can and likely will be replicated on device when local models and inference catch up, which they will eventually (Apple Intelligence, Hermes, OpenClaw, etc). Slap iCloud and Google Storage sync on the app, and you've got your own character bots in your device ecosystem.
> which they will eventually
"eventually" is doing a lot of work here. Like sure, yes, eventually. But that could be later this year or it could be 50 years from now. We don't know the answer to that. Assuming it's going to happen in a timeline short enough to matter for this conversation isn't worthwhile. It's like saying eventually we'll have cold fusion, so why is everyone skeptical?
I believe it is much closer than you realize
https://huggingface.co/Qwen/Qwen-Image-2.1-Turbo (7B)
It may be, I didn't say otherwise. I said we don't know for a fact that this moment will come in the near future, so just handwaving that it's obvious is folly.
There are already llms on phones now, they are being embedded / distributed by iOS and Android. You can find HN stories and comments about the struggle to disable or the joy of having them. Tons of Show HN about getting models to run on all sorts of devices, including phones
There's a difference between having "an llm" on a small device and having an llm of the class being discussed in this comment thread.
As I said, it may happen within a year. But maybe life takes a different direction. The person to whom I responded said "eventually", and yes, "eventually" it will happen. but that's not a useful time horizon to plan around because it could be a lot of things.
If you build your app today to use an LLM API with credits so the economics works (or support existing LLM providers who can provide an API key for use within the app, to not expose the app to negative usage economics), but it has the capability to switch to a local API using existing known primitives, I think that is very useful in this context, no?
The hard part is refining the UX, imho, while supporting pluggable LLM support on the backend. If the future never arrives of cheap on device inference, still works if you pay as you go and for those users willing to pay for their usage directly through their cloud inference provider of choice or through the app (unless the future is "cloud inference is unsustainable at consumer price points and on prem inference never arrives"). Just my two cents.
This is fundamentally what Apple did with new Siri, powered by Gemini today. They will eventually flip the switch to an in house solution. Agree the wildcard that still remains is "can this be done on devices eventually?"
https://www.apple.com/newsroom/2026/09/siri-ai-a-profoundly-...
https://www.macrumors.com/2025/11/05/apple-siri-google-gemin...
> Apple weighed using its own AI models for the LLM version of Siri , and also tested options from OpenAI and Anthropic, but it decided to go with Gemini after deciding Anthropic's fees were too high. Apple already has a partnership with Google for search results, with Google paying Apple around $20 billion per year to be the default search engine option on Apple devices.
> Though Apple is planning to rely on Google AI for now, it plans to continue working on its own models and will transition to an in-house solution when its LLMs are capable enough. Apple is already working on a 1 trillion parameter cloud-based model that could be ready as soon as 2026. Apple is unlikely to publicize its arrangement with Google while it develops in-house models.
So, your contention is that 404media isn't taking into consideration a hypothetical where they wouldn't burn millions of dollars each month running something.
Yes, which I say as a subscriber and donor to 404media (they do very good journalism work imho) and someone who leans more towards AI skeptic while being a heavy user in specific domains. "It's not a business, it's a feature."
Durable storage + sufficiently performant local LLM hardware + appropriate models [last two are wildcards and hard to predict the forward curve imho] = this experience locally. Remember when mobile video hardware acceleration wasn't good? And now, it is very good.
None of that has any relevance to what is happening right now. Character.ai is in chapter 11 bankruptcy, and "oh they should have done it differently, and 404media isn't taking into consideration a completely unproven hypothetical technology that wouldn't have bankrupted them" is not a useful discussion, nor a good critique of their coverage.
Title assertion:
> Character.ai's Bankruptcy Documents Show the Economics of AI Don't Work
My comments:
"I don't agree, here is why."
This specific model doesn't appear to be economically sustainable and/or commercially viable. That is a reasonable statement and assertion, based on the evidence presented. Does not mean the economics of AI in general do not work. That's a very broad statement! Nuance, words mean things.
Therefore, we should ask ourselves and discuss, "Under what circumstances could|would the economics work? Are there any? Or are there none?" We are discussing the problem space! Be more curious.
Some (most?) journalist skimp on the nuance. Character.ai is not representative of the entire Ai ecosystem and market. There are a lot of other factors that played a role here than the Ai economics themselves.
It seems harder to sell on device ai as a subscription, I would guess? although the much lower inference costs mean maybe you could just sell it as a one time purchase
By on device you mean running locally on a mobile device like an iPhone? If so we're a really long way from a chatbot that's engaging to talk to. There's just not enough processing power to do it
No-one's going to want to talk to a 7B model for more than 3 minutes
I think we're a breakthrough away. I don't think processing power is an issue. A 7B model has a lot of cruft (like knowledge of code syntax) that a character sim doesn't need. It should be possible to extract just the relevant parts of a larger and more powerful model and run it with the same resources you would run a general 7B model. We just don't have the ability to isolate and extract weights like that yet.
Yes, current gen hardware was not built to run such models, and are obviously inefficient and bad at it.
Significant investment is being spent on producing hardware that is more suitable for running mid-sized (100B-ish) models. Starting with flash chips with wide enough interfaces to provide the weights (NAND flash can be ~5x more energy-efficient for reads than DRAM), and going into more exotic solutions like compute-in-memory.
Yes, iphone, ipad, mac. Maybe your phone makes the inference requests to either Apple servers (requiring iCloud+ and credit based?) or your Mac (Tailscale equivalent).
My assumptions are that inference costs continue to decline (and/or the difficulty level of routing requests declines) and hardware acceleration continues into the future. These assumptions might be wrong, but that is the idea the thesis is based around.
(have built workflows and harnesses that currently target LLM provider APIs under the assumption I'll be able to point "locally-ish" eventually, my career had a strong focus on distributed computing about 10-15 years ago, think managing data and work clusters for LHC detector data processing, and I grew up on hobbyist distributed.net and BOINC projects, all the lenses I see this problem through; queues, jobs, sync vs async performance optimization, etc; recovered sysadmin|infra/devops engineer a lifetime ago)
I can have full conversations with Gemma4 on my phone, no server requests, with voice.
Excellent, thank you for the data point substantiating my thesis.
Isn't the majority of AI compute training and not inference? Saying "It doesn't matter if the AI companies fold because I can run models on my computer" is like saying "I don't care if the farm closes because I get my food from the grocery store"
It's more like saying 'I don't care if monsanto goes under because I will save my own seeds to plant next year's crops'.
You are correct, but I think it's much more like consumer culture doesn't and never will care about any of these details.
Specifically with regard to AI, the consumers aren't even aware of the "grocery store" either. They're being served from a silver platter. What matters most is the quality of results, not the methods. You can disagree all you want for whatever (good and compelling) reasons, but it doesn't change the fact that many people do not even buy groceries now and doordash everything.
If it turns out that you can put a mobile kitchen in their backyard that cooks up everything without them realizing, it makes no difference to the consumer.
Realistically, yes there's a finite amount of training data, hence the plateau in capability we've seen for at least a year. I think we're likely to see a consolidation and compression of all of this into local models. The economics don't support the expected growth of the current arrangement either. There simply isn't that much distinct data out there as initially assumed.
[flagged]
I have been seeing this more and more with “top” tier publications but “Character.ai’s bankruptcy is the one of the first dominos to fall in the long-foreseen AI bubble.”
A simple grammatical mistakes like this should have been caught on review/edit phase.
Either that or writers are relying too much on AI.
(Edit: fixed a grammatical mistake. See?)
AI doesn’t make mistakes like that generally, I’ve almost never seen the grammar butchery from them I’ve seen from even major publications in recent years. I wish they’d let AI copy edit.