> About a year ago, Groq announced a $1.5 billion infrastructure investment deal with Saudi Arabia. They also secured a $750 million Series D funding round.... Then in maybe one of the best rug pulls of all time, in July they quietly changed their revenue projections to $500 million. A 75% cut in four months. I’ve never seen anything like that since the 2008 financial crisis.
Not following the core argument here. Author seems to be comparing valuation in funding rounds to revenue projections. Revenue projection was revised downward, valuation was not.
Good point about not running the proprietary models, but that doesn't preclude strategic fit with Nvidia.
Groq is not a publicly traded company and has no legal reporting requirements. Sure, their projected revenue numbers they gave to investors dropped from $2b in February to $500m in July, but a later funding round in September showed it wasn't significant to how insiders saw the company. Contrary to what this article would imply, their valuation more than doubled from $2.8b last year to $6.9b this year after Groq's latest round of investment in September (after their revenue adjustment). Considering they increased revenue from $90m to $500m and got a $1.5b commitment from Saudi Arabia, I really don't see this being 'hype'.
> Then in maybe one of the best rug pulls of all time, in July they quietly
changed their valuation to $500 million. A 75% cut in four months. I’ve
never seen anything like that since the 2008 financial crisis.
Not sure where the author is getting their information from but there is seemingly little correlation between the investment rounds quoted in this post and other online sources. No mention for example of the Series E that valued Groq at $6.9bn.
It’s good that you have done that but does that section make sense? It says there was a 75% revenue cut, but the prior number is valuation, not revenue. Or was valuation and revenue the same at that time (seems unlikely)?
Edit: some searching about suggests that Groq initially projected $2 billion in revenue for 2025, later cutting that forecast to $500 million. That appears to have be what this article is trying to say.
This is not new in any way. Famously, Cisco has done this for decades, having been on a nonstop mad acquisition spree since the nineties, and more than once even acquiring companies that started as Cisco spin-out.
Also many of Google’s flagship products come from acquisitions. Eg Android, Docs, YouTube, their entire ad network, Firebase, DeepMind, lots more.
This isn’t easy! Equally famously, Microsoft routinely botches acquisitions, eg Skype, Nokia etc. Seems to me the only MS acquisitions that don’t fail are the ones they mostly leave alone (eg LinkedIn, GitHub).
Almost the entire biotech industry has been this way for decades once the small molecule patent cliff hit pharma and the R&D costs for therapies skyrocketed. If you look at biotech IPOs, the majority of the startups IPO pre-revenue, long before they’re even legally allowed to sell anything.
Which is totally fine: anyone who is a biotech investor knows this and everyone makes tons of money in this arrangement. Investors (both public and private) take on the science risk and some of the regulatory risk, and the pharmaceutical companies provide a guaranteed (big $$$) exit and take over scaling manufacturing to bring a drug to market. Most people with retirement accounts and pensions and index funds rarely touch this stuff except as a diversification strategy that pools the risky stuff to get the upside on the whole industry.
It's the same in medical devices. Most startups take it from idea through R&D then go public or are acquired right as they go through FDA approval or submit for it.
It's low risk from the acquirer's point of view. Somebody else paid for that research, you just get to buy it once it's proven itself sufficiently to your liking.
The tv series Silicon Valley has a good episode where they discuss the importance of a start-up not having any revenue. Being pre revenue apparently means unlimited potential, with any level of revenue being bad, as you always have to grow it.
I don't exactly disagree. But the word "obvious" doesn't work very well during a bubble. Sure, yes, the current revenue doesn't justify the purchase price. But that doesn't mean that anything justifies the purchase price.
We can't work backward rationally from "this deal makes sense" and get to "here's why". Corporate acquisitions often don't work that way, even when there's no bubble. The price is often just not justified at all. By anything.
In many cases they're just capitalizing testosterone.
It could mean different things I guess, but here’s my take:
If you do very risky R&D in a big corpo then the risk creeps into other things: other projects might look at the R&D and say, “we will just use that when it’s done”. It’s a lazy kind of move for tech leaders to make, because it makes you look like a team player, and if the R&D goes belly up then you have someone else to blame. This ultimately leads to risky R&D in a big corpo not being compartmentalized as much as it should be. If it fails, it fails super hard with lots of fallout.
But risky R&D at a startup is compartmentalized. Nobody will say they use the output of the startup until there are signs of life, and even then healthy caution is applied.
The FTC requested significant increases for technology and economic analysis for FY2025 ($535M), but was given a static budget with plans to cut by 11%. FTC chair Ferguson reduced staff from 1,315 to 1,221 and aims to reach around 1,100 through attrition to align with lower budgets.
We could just keep using "First world" vs "Third world" as above. There are also multiple NGOs that measure these things and break down the "ratings" into component parts.
On the other hand, an advocate for greatly increased corruption might claim that corruption can't be measured at all. Or, hypothetically, they might strictly use un-anchored non-metrics like "the other guy does it too", "any is too much", "omg look over there!", etc.
how exactly is different…? I give you Trump, you give me Biden. I give you Scott, you give me Pelosi. I give you Bush, you give me Clinton… it is not different but regardless of whether it is or isn’t no one is defending it. it is just a disservice to everyone to think somehow magically things are all this different now than before, same crap different toilet paper
Ok, give me biden. Where are his lists of corruption scandals[1]? His public statements about taking bribes?
I'm not quite sure how to explain this very obvious point: biden and his government was not corrupt in any meaningful sense and trump and his government is extremely corrupt and pretending that they're the same is both factually wrong and has the effect of protecting trump and his corruption.
The point isn't that anyone is above reproach, the point is that all you're doing is normalizing the increased awfulness of the republican corruption. And normalizing it means that it is more likely to continue happening and less likely to be punished.
If you're supposedly unhappy about clinton "corruption" why aren't you really mad about trump?
This whole "oh everything is the same nothing can improve" attitude is literally a favored tactic of the most corrupt governments. They want you to think that way because it means they'll never be held accountable. Any time people start talking about improving things they're met with an endless deluge of "oh it's all the same nothing can change" which is, of course, a self-fulfilling prophecy.
[1] The best the fairly obvious house republican "investigation" into joe biden could manage was some vague statements about his son getting paid for having the last name biden, which may or may not be illegal, but certainly seems unethical, but more importantly, ISN'T THE SITTING PRESIDENT OF THE UNITED STATES. Like, it is so incredibly obvious that words fail me that the president being corrupt matters A LOT MORE than his son being corrupt. Like, a lot a lot a lot more.
It's different because it's all about that now. The Clintons had their scandals, the "pay to play" lists etc. We all know they are in bed with the moneymen. But it didn't define their administration, and they were pretty hush-hush about it.
Trump on the other hand is completely open about this. He even brags about making money from deals, something that was previously considered a huge conflict of interest. He appoints people based on loyalty alone, not knowledge or experience. He bullies countries into compliance with mafia tactics ("appease me or else..." tariffs or even war like venezuela and greenland). It's a huge moral shift where that is no longer unthinkable. The US used to have values. It was a country that was at least trying to be the good guy.
Also, the constitution used to be holy. Now Trump is flaunting the 1st amendment on a daily basis (limiting LGBTIQ+ speech, establishing America as a "christian country" which is explicitly forbidden). I think all these developments are very concerning. I don't live in America but considering it is still a big world power it does worry me.
I love the spirit of your comments but IMO it is misguided
The US used to have values. It was a country that was at least trying to be the good guy.
This really is all wrong. One might think this based on pitches from different times but all Empires are evil by their definition and America has always been that, always
> This really is all wrong. One might think this based on pitches from different times but all Empires are evil by their definition and America has always been that, always
Again, the problem with this train of logic is you inevitable condemn everyone and everything as evil, at which point the word completely loses its meaning. Evil is only useful as a term if there are actually things that are not evil.
America has certainly done immoral, unethical and frankly evil things. It's also done moral, beautiful and even heroic things. It's a big complicated entity made up of literally millions of people and trying to summarize it as "good or evil" is pointless.
The reason this nuance matters is that we want, need to encourage doing good and the first step to doing that is to actually be able to distinguish between good and evil.
just do a simple thing - ballpark how many lives of innocent people has America taken, lets just say since WWII. then lets see after you ballpark this whether you still think we are (or ever were) “good guys”
The major difference is the disappearance of shame.
However, the greatest enablement was the overblown cynicism large swaths of the american elites had towards the national proclaimed values. When you think everything is cynical even when it is not then the next step is to have governments that are completely cynical.
> I don't understand how market regulators allow this.
The US government is literally for sale. Businesses know that this window is limited and are executing antitrust manuvers left and right while they can.
the window is un-limited so there is not rush. the government has been for sale for a long time and will continue to be so regardless of who is “running the country”
> the government has been for sale for a long time
The government has been under significant influence of corporations for a long time: this is true. But now bribes are being accepted unabashedly. Presumably, hopefully, this won't last beyond the current administration. To equate the two is dishonest.
Can you tell me the last US president to accept literal bars of gold and jumbo jets from foreign monarchs? Or the last one who ran a crypto coin, pardoned a crypto billionaire who he claimed not to know?
Under current DoJ antitrust guidelines, there's nothing to stop a future administration from reviewing any anti-competitive actions ignored by the current one as part of an anti-competitive series of actions: https://www.justice.gov/atr/merger-guidelines/applying-merge...
So those businesses either know, or expect, that either:
a) these guidelines will be changed in a way that makes them hard or impossible to revert (i.e. through legislation or a Supreme Court judgement); or
b) there is little risk of a future change of administration.
Or (c) that any future administration is going to have a lot of more pressing concerns that will drown out seriously relitigating past mergers and acquisitions, and any concerns they do have will most likely be mollified with agreed remedies that sacrifice far less than the value of doing the merger.
Very few administrations do everything they theoretically could under the law and their own guidelines (even the ones that also do lots that violates both.)
Well there's also a c) - Whatever they get away with now they will have in pocket, and whatever penance they will have to do with a future administration will take years and years of legal back and forth to actually pan out, by which time it will be watered down so any fine will dwarf the profits made during this period.
Also, if they manage to reach "too big to fail" status by that point, whatever punishment will be nothing more than a slap on the wrist.
Funny how everyone shits on Nvidia's monopoly when we've got Google walking around after winning a monumental antitrust case regarding their Android/Chrome/Google information monopoly.
> Funny how everyone shits on Nvidia's monopoly when we've got Google walking around after winning a monumental antitrust case regarding their Android/Chrome/Google information monopoly.
... are you implying people around here don't give google flak for monopolistic business practices? That doesn't square with my experience, here.
I know no one wants to hear this, but this “acquisition “ is nothing of the kind. It’s just Nvidia hiring the four or five guys they need without having to take on the rest of groq. Which, as it turns out, is worthless without those four or five guys.
This is what happens when companies figure out they don’t have to buy out other companies. They just need to pay off shareholders for the right to hire key employees. Which is convenient, since the key four or five guys are usually pretty big shareholders.
It’s no longer necessary to monopolize a market. You can monopolize intellectual capital by just paying ungodly sums of money. The rest will take care of itself.
> It's a shame. Groq was really great. Nvidia is stifling innovation here.
I don't share your view. Groq continues to exist. Nvidia did not take any or their hardware, so the same Groq you access on OpenRouter will exist tomorrow or one year from now. If anything, they'll significantly increase their presence, since they just got $20 billion in cash.
As for Nvidia stifling innovation: one can argue that they do the opposite. They hired key personnel from Groq (including their founder and CEO, Jonnathan Ross). These people agreed to the move, presumably for the money, but most likely also because they think they can deliver even more if they have access to Nvidia's resources. So, in terms of overall innovation, it will most likely go up.
But you can say that they stifle independent innovation. Maybe, but the case for that is not that open and shut as it might seem. They entered a non-exclusive licensing agreement with Groq. Which means Groq can provide their "secret sauce" to other interested entities, maybe Apple, maybe Intel or AMD, maybe OpenAI, maybe Oracle. The number of companies who could be interested in their tech is quite high.
Or simply, Groq, with the many billions in unencumbered cash they just received will decide to go for version 2.0 of their tech, or they can significantly expand the GroqCloud. Their valuation just went from $6.5B to significantly higher than $20B. They can pursue an IPO, or they can issue debt. There are countless possibilities for Groq now.
> I don't share your view. Groq continues to exist. Nvidia did not take any or their hardware, so the same Groq you access on OpenRouter will exist tomorrow or one year from now. If anything, they'll significantly increase their presence, since they just got $20 billion in cash.
The linked article expects differently:
> Nvidia’s buying them with their insanely inflated war chest. They don’t want a chunk taken out of their market share. They can’t afford to take that chance. So it’s like they’re just saying: “Shut up, take the $20 billion, walk away from this project.”
How much this is true I can't really verify myself but it certainly sounds concerning.
> But you can say that they stifle independent innovation.
But this is exactly what a market watchdog is supposed to prevent. A market with one player (or two) is no market. And Groq was going in a decidedly different direction than Nvidia.
The linked article echoes my worries in other ways as well e.g. worker displacement, explosion of energy usage. I often equate it with the dotcom era, I worked on this thinking we made the world better. But the endgame, with the Google, Meta, pervasive tracking etc is much more dystopian. Especially considering the societal effects. Enshittification, corporate rule, polarisation due to social medias promoting "engagement" and thus conflicting content that get people riled up.
I don't want the same to happen with AI here and it feels like they are already aligning the stars to make exactly that happen.
> To visualize $1.5 billion: if you cashed that check out in $100 bills and stacked them one on top of another, it would reach a five story building. For ordinary plebeians like us, at the average US salary of around $75K, you’d need to work 20,000 years to earn that.
The average US salary isn't $75k btw. That figure is usually quoted from the reported median household income in 2022[1]. The median personal income, which is the figure that should* be quoted, was around $45k for 2024[2].
Why do you think? Not because of any output of the company, of course.
But because buying it helps perpetuate the hype and money cycle of the 'AI' trend for awhile longer. It may not look like it directly, but a purchase like this keeps Nvidia's stock up in the future, which is all investors care about.
If this is true, is it just the HN community that understands this? Otherwise, wouldn’t it make sense that the market understands this already and doesn’t fall for the hype? It doesn’t pass the smell test for me that it’s that transparent of a play for hype. What am I missing?
AI is real and it's also hyped. There's circular financing and real money involved. Groq has good tech and smart people and Nvidia is also taking a competitor off the board. People who only see one side have a lack of imagination.
B) All info the OP(= author) knows is known to the professionals dealing with the due diligence. They decided to do so while looking at data which is not available the public. So assuming they know some things why we don’t know is not a far fetched idea.
It's not - I prefer to record a YouTube video first (I speak better than I write) https://www.youtube.com/watch?v=2po-s2yOCcg Then I run that through a transcription app (so AI I suppose but still my words, shout out to Scriber Pro which I found via HN: https://news.ycombinator.com/item?id=45591222) then I edit down and try to change my verbs (visual to written) where necessary.
What’s more concerning is the growing trend of big tech companies “acquiring” a startup’s leadership team and IP and screwing over all the employees holding equity.
My understanding is Groq failed to deploy their second-gen chips on time, which caused their stock to deflate.
Groq's primary advantage over Cerebras and SambaNova, as I see it, is they don't fabricate on TSMC. That's attractive to Nvidia, who doesn't want to give up any of their datacenter GPU allocation.
You are missing the point that it is a strategic acquisition to kickstart a new vertical that they have struggled with: serving inference. They have tried to organically grow this and do weird things like inference within their other customers’ clouds.
It certainly isn’t a “panic” as nvidia is so flush with cash. This is a minuscule amount of money for them.
chips designed to run ResNET? I guess the haskell compiler they built is impressive (it made it so 8 racks of chips designed to run ResNET can run llama 70b with extremely low latency).
Edit: my information might be old, I don't know if they successfully taped out their second gen chip or not. Can anyone corroborate?
> At that time, the company was valued at $2 billion. Hello bubble. Then in maybe one of the best rug pulls of all time, in July they quietly changed their revenue projections to $500 million. A 75% cut in four months.
No, the author is just stupidly spreading misinformation. Looking through their other posts, it looks like he has an agenda to prove that we're in an AI bubble.
The bubble take is tired. This was regulatory arbitrage: IP licensing instead of acquisition to dodge CFIUS/antitrust. The $13B premium to avoid years of hold up while enriching Chamath and giving Trump's AI Czar a Christmas present. So many other things at play here than just "AI bubble so big it will boom".
The writing style here is so belittling, and frankly stupid.
E.g. "billion is so big!", uh, I've heard of a billion before, and then comparing the value of a company to a single person's salary, as if that was very relevant.
Go talk to someone outside of tech this week, preferably someone working in the trades or something else that's less dependent on a computer, and ask them about their AI use. You'd be surprised how new a lot of the tech concepts in this article are to people that might have only heard of ChatGPT.
Was just walking past a construction site and heard some of their banter. Didn’t realise the common man could debate the benefits of an LPU over GPGPU so eloquently. One of them even compared SRAM vs DRAM as being like a cheetah vs an injured antelope ;)
> About a year ago, Groq announced a $1.5 billion infrastructure investment deal with Saudi Arabia. They also secured a $750 million Series D funding round.... Then in maybe one of the best rug pulls of all time, in July they quietly changed their revenue projections to $500 million. A 75% cut in four months. I’ve never seen anything like that since the 2008 financial crisis.
Not following the core argument here. Author seems to be comparing valuation in funding rounds to revenue projections. Revenue projection was revised downward, valuation was not.
Good point about not running the proprietary models, but that doesn't preclude strategic fit with Nvidia.
Yeah the point is you aren't supposed to mislead investors with overly rosy revenue projections. That's fraud (but probably hard to prove).
If they told the investors privately then they're probably fine, but I doubt they did.
In private companies, it's much more caveat emptor, than in the public markets.
Groq is not a publicly traded company and has no legal reporting requirements. Sure, their projected revenue numbers they gave to investors dropped from $2b in February to $500m in July, but a later funding round in September showed it wasn't significant to how insiders saw the company. Contrary to what this article would imply, their valuation more than doubled from $2.8b last year to $6.9b this year after Groq's latest round of investment in September (after their revenue adjustment). Considering they increased revenue from $90m to $500m and got a $1.5b commitment from Saudi Arabia, I really don't see this being 'hype'.
src: https://www.reuters.com/business/groq-more-than-doubles-valu...
> Then in maybe one of the best rug pulls of all time, in July they quietly changed their valuation to $500 million. A 75% cut in four months. I’ve never seen anything like that since the 2008 financial crisis.
Not sure where the author is getting their information from but there is seemingly little correlation between the investment rounds quoted in this post and other online sources. No mention for example of the Series E that valued Groq at $6.9bn.
Thank you for calling this out. I have edited and added a footnote there. Their valuation wasn't 500mil it was their revenue projection.
The 6.9bn valuation is accurate though https://www.prnewswire.com/news-releases/groq-raises-750-mil... https://www.reuters.com/business/groq-more-than-doubles-valu...
It’s good that you have done that but does that section make sense? It says there was a 75% revenue cut, but the prior number is valuation, not revenue. Or was valuation and revenue the same at that time (seems unlikely)?
Edit: some searching about suggests that Groq initially projected $2 billion in revenue for 2025, later cutting that forecast to $500 million. That appears to have be what this article is trying to say.
https://www.investing.com/news/company-news/groq-slashes-202...
Ya well, startups are just low risk R&D facilities in service of big tech now https://centreforaileadership.org/resources/opinion_startups...
This is not new in any way. Famously, Cisco has done this for decades, having been on a nonstop mad acquisition spree since the nineties, and more than once even acquiring companies that started as Cisco spin-out.
Also many of Google’s flagship products come from acquisitions. Eg Android, Docs, YouTube, their entire ad network, Firebase, DeepMind, lots more.
This isn’t easy! Equally famously, Microsoft routinely botches acquisitions, eg Skype, Nokia etc. Seems to me the only MS acquisitions that don’t fail are the ones they mostly leave alone (eg LinkedIn, GitHub).
Almost the entire biotech industry has been this way for decades once the small molecule patent cliff hit pharma and the R&D costs for therapies skyrocketed. If you look at biotech IPOs, the majority of the startups IPO pre-revenue, long before they’re even legally allowed to sell anything.
Which is totally fine: anyone who is a biotech investor knows this and everyone makes tons of money in this arrangement. Investors (both public and private) take on the science risk and some of the regulatory risk, and the pharmaceutical companies provide a guaranteed (big $$$) exit and take over scaling manufacturing to bring a drug to market. Most people with retirement accounts and pensions and index funds rarely touch this stuff except as a diversification strategy that pools the risky stuff to get the upside on the whole industry.
It's the same in medical devices. Most startups take it from idea through R&D then go public or are acquired right as they go through FDA approval or submit for it.
I don't understand what "low risk" means here. For a start-up, 99% risk of failure is low. What are we comparing that to?
It's low risk from the acquirer's point of view. Somebody else paid for that research, you just get to buy it once it's proven itself sufficiently to your liking.
Well usually they only get acquired when they have something the purchaser wants, revenue was obviously not the get here
The tv series Silicon Valley has a good episode where they discuss the importance of a start-up not having any revenue. Being pre revenue apparently means unlimited potential, with any level of revenue being bad, as you always have to grow it.
I have been in actual conversations where the topic was whether to avoid revenue to prevent being measured on it...
That show was very on the nose about a great many things.
Yeah. That's totally real. (But I get that it was also funny, and I loved that show.)
I don't exactly disagree. But the word "obvious" doesn't work very well during a bubble. Sure, yes, the current revenue doesn't justify the purchase price. But that doesn't mean that anything justifies the purchase price.
We can't work backward rationally from "this deal makes sense" and get to "here's why". Corporate acquisitions often don't work that way, even when there's no bubble. The price is often just not justified at all. By anything.
In many cases they're just capitalizing testosterone.
Yeah I'm with you, didn't mean to imply there was any sort of underlying wisdom or truth behind the choice. People just love to rationalize
It could mean different things I guess, but here’s my take:
If you do very risky R&D in a big corpo then the risk creeps into other things: other projects might look at the R&D and say, “we will just use that when it’s done”. It’s a lazy kind of move for tech leaders to make, because it makes you look like a team player, and if the R&D goes belly up then you have someone else to blame. This ultimately leads to risky R&D in a big corpo not being compartmentalized as much as it should be. If it fails, it fails super hard with lots of fallout.
But risky R&D at a startup is compartmentalized. Nobody will say they use the output of the startup until there are signs of life, and even then healthy caution is applied.
Low risk for the large companies
Totally. I understand what the claim is. I don't understand why anyone would believe it.
Maybe the loses are limited to the amount of the investment in the startup? No risk of consuming more resources than intended before dying.
If those things were integrated into the giant there would be political risk of it eating all of the money of the giant.
Big pharma has worked like this for a while now
Extremely well put - this is my assumption as well having worked at many of them!
It's a shame. Groq was really great. Nvidia is stifling innovation here. I don't understand how market regulators allow this.
The FTC requested significant increases for technology and economic analysis for FY2025 ($535M), but was given a static budget with plans to cut by 11%. FTC chair Ferguson reduced staff from 1,315 to 1,221 and aims to reach around 1,100 through attrition to align with lower budgets.
Oversight hearing is worth a listen to get a better idea on how the current administation is harming regulators: https://www.youtube.com/watch?v=0NZxkvYaVuk
This is what happen when your government is run primarily thru corruption.
sorry, not corruption! retainer fees and timely stock purchases. different thing!
so same as always, eh? or different this time around?
Yeah, it's always corrupt. No, it was not always this corrupt. We're aiming for third world. That's going to be a new experience.
how do you measure levels of corruptness?
We could just keep using "First world" vs "Third world" as above. There are also multiple NGOs that measure these things and break down the "ratings" into component parts.
On the other hand, an advocate for greatly increased corruption might claim that corruption can't be measured at all. Or, hypothetically, they might strictly use un-anchored non-metrics like "the other guy does it too", "any is too much", "omg look over there!", etc.
Yes, it is different. Pretending it is the same is just another way to defend the corrupt ones.
how exactly is different…? I give you Trump, you give me Biden. I give you Scott, you give me Pelosi. I give you Bush, you give me Clinton… it is not different but regardless of whether it is or isn’t no one is defending it. it is just a disservice to everyone to think somehow magically things are all this different now than before, same crap different toilet paper
Ok, give me biden. Where are his lists of corruption scandals[1]? His public statements about taking bribes?
I'm not quite sure how to explain this very obvious point: biden and his government was not corrupt in any meaningful sense and trump and his government is extremely corrupt and pretending that they're the same is both factually wrong and has the effect of protecting trump and his corruption.
The point isn't that anyone is above reproach, the point is that all you're doing is normalizing the increased awfulness of the republican corruption. And normalizing it means that it is more likely to continue happening and less likely to be punished.
If you're supposedly unhappy about clinton "corruption" why aren't you really mad about trump?
This whole "oh everything is the same nothing can improve" attitude is literally a favored tactic of the most corrupt governments. They want you to think that way because it means they'll never be held accountable. Any time people start talking about improving things they're met with an endless deluge of "oh it's all the same nothing can change" which is, of course, a self-fulfilling prophecy.
[1] The best the fairly obvious house republican "investigation" into joe biden could manage was some vague statements about his son getting paid for having the last name biden, which may or may not be illegal, but certainly seems unethical, but more importantly, ISN'T THE SITTING PRESIDENT OF THE UNITED STATES. Like, it is so incredibly obvious that words fail me that the president being corrupt matters A LOT MORE than his son being corrupt. Like, a lot a lot a lot more.
https://oversight.house.gov/the-bidens-influence-peddling-ti...
It's different because it's all about that now. The Clintons had their scandals, the "pay to play" lists etc. We all know they are in bed with the moneymen. But it didn't define their administration, and they were pretty hush-hush about it.
Trump on the other hand is completely open about this. He even brags about making money from deals, something that was previously considered a huge conflict of interest. He appoints people based on loyalty alone, not knowledge or experience. He bullies countries into compliance with mafia tactics ("appease me or else..." tariffs or even war like venezuela and greenland). It's a huge moral shift where that is no longer unthinkable. The US used to have values. It was a country that was at least trying to be the good guy.
Also, the constitution used to be holy. Now Trump is flaunting the 1st amendment on a daily basis (limiting LGBTIQ+ speech, establishing America as a "christian country" which is explicitly forbidden). I think all these developments are very concerning. I don't live in America but considering it is still a big world power it does worry me.
I love the spirit of your comments but IMO it is misguided
The US used to have values. It was a country that was at least trying to be the good guy.
This really is all wrong. One might think this based on pitches from different times but all Empires are evil by their definition and America has always been that, always
> This really is all wrong. One might think this based on pitches from different times but all Empires are evil by their definition and America has always been that, always
Again, the problem with this train of logic is you inevitable condemn everyone and everything as evil, at which point the word completely loses its meaning. Evil is only useful as a term if there are actually things that are not evil.
America has certainly done immoral, unethical and frankly evil things. It's also done moral, beautiful and even heroic things. It's a big complicated entity made up of literally millions of people and trying to summarize it as "good or evil" is pointless.
The reason this nuance matters is that we want, need to encourage doing good and the first step to doing that is to actually be able to distinguish between good and evil.
just do a simple thing - ballpark how many lives of innocent people has America taken, lets just say since WWII. then lets see after you ballpark this whether you still think we are (or ever were) “good guys”
The major difference is the disappearance of shame.
However, the greatest enablement was the overblown cynicism large swaths of the american elites had towards the national proclaimed values. When you think everything is cynical even when it is not then the next step is to have governments that are completely cynical.
> I don't understand how market regulators allow this.
The US government is literally for sale. Businesses know that this window is limited and are executing antitrust manuvers left and right while they can.
the window is un-limited so there is not rush. the government has been for sale for a long time and will continue to be so regardless of who is “running the country”
> the government has been for sale for a long time
The government has been under significant influence of corporations for a long time: this is true. But now bribes are being accepted unabashedly. Presumably, hopefully, this won't last beyond the current administration. To equate the two is dishonest.
Can you tell me the last US president to accept literal bars of gold and jumbo jets from foreign monarchs? Or the last one who ran a crypto coin, pardoned a crypto billionaire who he claimed not to know?
No that’s completely ridiculous.
Under current DoJ antitrust guidelines, there's nothing to stop a future administration from reviewing any anti-competitive actions ignored by the current one as part of an anti-competitive series of actions: https://www.justice.gov/atr/merger-guidelines/applying-merge...
So those businesses either know, or expect, that either:
a) these guidelines will be changed in a way that makes them hard or impossible to revert (i.e. through legislation or a Supreme Court judgement); or
b) there is little risk of a future change of administration.
Or (c) that any future administration is going to have a lot of more pressing concerns that will drown out seriously relitigating past mergers and acquisitions, and any concerns they do have will most likely be mollified with agreed remedies that sacrifice far less than the value of doing the merger.
Very few administrations do everything they theoretically could under the law and their own guidelines (even the ones that also do lots that violates both.)
A quick google search indicates the average tenure of a CEO is ~7 years.
I wonder if there should be a c) There is a lack of meaningful planning beyond the current status quo.
Well there's also a c) - Whatever they get away with now they will have in pocket, and whatever penance they will have to do with a future administration will take years and years of legal back and forth to actually pan out, by which time it will be watered down so any fine will dwarf the profits made during this period.
Also, if they manage to reach "too big to fail" status by that point, whatever punishment will be nothing more than a slap on the wrist.
what was great about Groq? The token speed? Last I checked they aggressively throttled how many requests you could make.
Funny how everyone shits on Nvidia's monopoly when we've got Google walking around after winning a monumental antitrust case regarding their Android/Chrome/Google information monopoly.
How do the market regulators allow that?
My first grade teacher used to claim that two wrongs didn't make a right.
> How do the market regulators allow that?
Same way I reckon. Both are bad.
> Funny how everyone shits on Nvidia's monopoly when we've got Google walking around after winning a monumental antitrust case regarding their Android/Chrome/Google information monopoly.
... are you implying people around here don't give google flak for monopolistic business practices? That doesn't square with my experience, here.
One wrong doesn't make another wrong right.
Market regulators are working hard to ensure regulatory capture for the big players.
There’s regulators??
What market regulators?
I know no one wants to hear this, but this “acquisition “ is nothing of the kind. It’s just Nvidia hiring the four or five guys they need without having to take on the rest of groq. Which, as it turns out, is worthless without those four or five guys.
This is what happens when companies figure out they don’t have to buy out other companies. They just need to pay off shareholders for the right to hire key employees. Which is convenient, since the key four or five guys are usually pretty big shareholders.
It’s no longer necessary to monopolize a market. You can monopolize intellectual capital by just paying ungodly sums of money. The rest will take care of itself.
Maybe it was not the right term, "acquisition". But really the end-result is the same.
> It's a shame. Groq was really great. Nvidia is stifling innovation here.
I don't share your view. Groq continues to exist. Nvidia did not take any or their hardware, so the same Groq you access on OpenRouter will exist tomorrow or one year from now. If anything, they'll significantly increase their presence, since they just got $20 billion in cash.
As for Nvidia stifling innovation: one can argue that they do the opposite. They hired key personnel from Groq (including their founder and CEO, Jonnathan Ross). These people agreed to the move, presumably for the money, but most likely also because they think they can deliver even more if they have access to Nvidia's resources. So, in terms of overall innovation, it will most likely go up.
But you can say that they stifle independent innovation. Maybe, but the case for that is not that open and shut as it might seem. They entered a non-exclusive licensing agreement with Groq. Which means Groq can provide their "secret sauce" to other interested entities, maybe Apple, maybe Intel or AMD, maybe OpenAI, maybe Oracle. The number of companies who could be interested in their tech is quite high.
Or simply, Groq, with the many billions in unencumbered cash they just received will decide to go for version 2.0 of their tech, or they can significantly expand the GroqCloud. Their valuation just went from $6.5B to significantly higher than $20B. They can pursue an IPO, or they can issue debt. There are countless possibilities for Groq now.
The people now working for Nvidia will keep innovating but now with monopolistic pricing.
The $20B will be paid out to investors. Maybe GroqCloud will keep $1B to keep the lights on for a few years.
> I don't share your view. Groq continues to exist. Nvidia did not take any or their hardware, so the same Groq you access on OpenRouter will exist tomorrow or one year from now. If anything, they'll significantly increase their presence, since they just got $20 billion in cash.
The linked article expects differently:
> Nvidia’s buying them with their insanely inflated war chest. They don’t want a chunk taken out of their market share. They can’t afford to take that chance. So it’s like they’re just saying: “Shut up, take the $20 billion, walk away from this project.”
How much this is true I can't really verify myself but it certainly sounds concerning.
> But you can say that they stifle independent innovation.
But this is exactly what a market watchdog is supposed to prevent. A market with one player (or two) is no market. And Groq was going in a decidedly different direction than Nvidia.
The linked article echoes my worries in other ways as well e.g. worker displacement, explosion of energy usage. I often equate it with the dotcom era, I worked on this thinking we made the world better. But the endgame, with the Google, Meta, pervasive tracking etc is much more dystopian. Especially considering the societal effects. Enshittification, corporate rule, polarisation due to social medias promoting "engagement" and thus conflicting content that get people riled up.
I don't want the same to happen with AI here and it feels like they are already aligning the stars to make exactly that happen.
> To visualize $1.5 billion: if you cashed that check out in $100 bills and stacked them one on top of another, it would reach a five story building. For ordinary plebeians like us, at the average US salary of around $75K, you’d need to work 20,000 years to earn that.
No, we don't need to visualize that.
Hahaha why are people so mad at this visualization? I thought it was pretty cool!
The average US salary isn't $75k btw. That figure is usually quoted from the reported median household income in 2022[1]. The median personal income, which is the figure that should* be quoted, was around $45k for 2024[2].
[1] https://fred.stlouisfed.org/series/MEHOINUSA646N [2] https://fred.stlouisfed.org/series/MEPAINUSA646N
Most people are not doers but losers. When someone does anything, they like to throw s*t at it.
Doers are perpetually disliked by losers simply because they can do stuff while them can't.
It's natural that the losers' actions are aligned towards making doers disappear, but it's usually a very low level and pathetic threat to doers.
I enjoyed reading your article and hope you have more stuff coming :). Keep it going!
Why do you think? Not because of any output of the company, of course.
But because buying it helps perpetuate the hype and money cycle of the 'AI' trend for awhile longer. It may not look like it directly, but a purchase like this keeps Nvidia's stock up in the future, which is all investors care about.
If this is true, is it just the HN community that understands this? Otherwise, wouldn’t it make sense that the market understands this already and doesn’t fall for the hype? It doesn’t pass the smell test for me that it’s that transparent of a play for hype. What am I missing?
AI is real and it's also hyped. There's circular financing and real money involved. Groq has good tech and smart people and Nvidia is also taking a competitor off the board. People who only see one side have a lack of imagination.
A) What’s 20b comparing to the extravagant current valuation of Nvidia at 4.64t?
https://claude.ai/public/artifacts/8c395eb5-8d22-431f-b6ba-0...
B) All info the OP(= author) knows is known to the professionals dealing with the due diligence. They decided to do so while looking at data which is not available the public. So assuming they know some things why we don’t know is not a far fetched idea.
Since this article mentions SRAM, which not everyone knows a lot about, this could be helpful for anyone interested: https://people.freebsd.org/~lstewart/articles/cpumemory.pdf
It's a throwaway company used to tie up and defer account receivables for stock reporting, no?
If your goal isn't four times greater than what you're likely to achieve, than you might as well not have goals.
— A motivational speaker, probably
Revenue targets are meaningless, especially in hyped fields.
This article was good, but the blaming the us-east-1 outage on layoffs doesn't seem accurate.
Thank you for reading. Outages could be vibe regressions, could be something else. I'm definitely making an opinionated leap there.
Opinionated and uneducated. A favorite pastime of the internet it's just a shame that we give so much attention to such blogposts.
The article was terrible, it read like someone desperately justifying there already made conclusions with anything possible.
Thank you for reading it.
Gell-Mann Amnesia Effect.
Nothing against the author, the content is good. I just wish this wasn't fluffed out with AI.
It's not - I prefer to record a YouTube video first (I speak better than I write) https://www.youtube.com/watch?v=2po-s2yOCcg Then I run that through a transcription app (so AI I suppose but still my words, shout out to Scriber Pro which I found via HN: https://news.ycombinator.com/item?id=45591222) then I edit down and try to change my verbs (visual to written) where necessary.
What made you think there was AI in it?
Killing your competition is priceless though
Kind of weird to base the bubble on how tall a stack of $100 bills is...
I just thought it was cool
That probably explains why the Groq board accepted the deal.
It’s amusing to see repeated “it’s a bubble” takes.
One of them will surely be right eventually!
The man who said the market's about to crash this year for the past 40 years finally proven right!
What’s more concerning is the growing trend of big tech companies “acquiring” a startup’s leadership team and IP and screwing over all the employees holding equity.
Yes, this was a defensive move from Nvidia.
My understanding is Groq failed to deploy their second-gen chips on time, which caused their stock to deflate.
Groq's primary advantage over Cerebras and SambaNova, as I see it, is they don't fabricate on TSMC. That's attractive to Nvidia, who doesn't want to give up any of their datacenter GPU allocation.
Author seems to be conflating valuation and projected revenue
You are missing the point that it is a strategic acquisition to kickstart a new vertical that they have struggled with: serving inference. They have tried to organically grow this and do weird things like inference within their other customers’ clouds.
It certainly isn’t a “panic” as nvidia is so flush with cash. This is a minuscule amount of money for them.
They understand the value of the tech
chips designed to run ResNET? I guess the haskell compiler they built is impressive (it made it so 8 racks of chips designed to run ResNET can run llama 70b with extremely low latency).
Edit: my information might be old, I don't know if they successfully taped out their second gen chip or not. Can anyone corroborate?
We've all been waiting for info about the second gen chip. I assume Nvidia saw something privately (good or bad) that triggered this acquisition.
And it was a no brainer
In three years China will be making far cheaper clones of Nvidia chips
Current administration just handed it to them for a bribe
https://www.bbc.com/news/articles/ckg9q635q6po
They'll tear them down and x-ray/electron-microscope it and they've gotten exceeding great at cloning chips
Underrated comment. We are damaging our economy and giving advantages to our real competitors
> At that time, the company was valued at $2 billion. Hello bubble. Then in maybe one of the best rug pulls of all time, in July they quietly changed their revenue projections to $500 million. A 75% cut in four months.
I feel like I'm missing something here…
No, the author is just stupidly spreading misinformation. Looking through their other posts, it looks like he has an agenda to prove that we're in an AI bubble.
The bubble take is tired. This was regulatory arbitrage: IP licensing instead of acquisition to dodge CFIUS/antitrust. The $13B premium to avoid years of hold up while enriching Chamath and giving Trump's AI Czar a Christmas present. So many other things at play here than just "AI bubble so big it will boom".
Here's my take on what actually happened: https://ossa-ma.github.io/blog/groq
The writing style here is so belittling, and frankly stupid.
E.g. "billion is so big!", uh, I've heard of a billion before, and then comparing the value of a company to a single person's salary, as if that was very relevant.
Go talk to someone outside of tech this week, preferably someone working in the trades or something else that's less dependent on a computer, and ask them about their AI use. You'd be surprised how new a lot of the tech concepts in this article are to people that might have only heard of ChatGPT.
That is who I'm writing for.
Was just walking past a construction site and heard some of their banter. Didn’t realise the common man could debate the benefits of an LPU over GPGPU so eloquently. One of them even compared SRAM vs DRAM as being like a cheetah vs an injured antelope ;)
And they aren’t reading you, so lots of wasted words.
Brother nobody is reading me I only have like 150 substack followers.
Exactly. Lots of condescension.
It was probably less about revenue targets and more about pre-emptively removing potential future competition while it is still relatively cheap.
This is missing the point. Even at a clipped revenue projection - $500m at $2b is a 4x multiple on revenue.
A 4x for an AI cloud+infra play that targets speed and cost? Where do I send the check?
If NVIDIA believes it can take this and scale it, $20b is a no brainer.
Agreed - but why not pay 14B (double their highest valuation)?
The American Dream /s
Emphasis on Dream!