RU version is available. Content is displayed in original English for accuracy.
Advertisement
Advertisement
⚡ Community Insights
Discussion Sentiment
46% Positive
Analyzed from 4284 words in the discussion.
Trending Topics
#nvidia#more#money#coreweave#don#big#company#gpus#problem#circular
Discussion (154 Comments)Read Original on HackerNews
Nvidia invested $2b into CoreWeave for 9% equity stake. CoreWeave is spending $35b in CapEx in 2026. Therefore, Nvidia's investment is only 5.7% of CoreWeave's single year CapEx. The other $32b is coming from other sources that isn't Nvidia. This is hardly circular.
Nvidia invests in Neoclouds because it's a hedge against hyperscalers having too much power, ie designing and prioritizing their own chips, and not fully using Nvidia's rack design. Neoclouds give hyperscalers competition. Neoclouds accept Nvidia investments because it allows them to secure Nvidia chips first, which is a competitive advantage since new Nvidia chips have been as much as ~5-20x more efficient than old Nvidia chips.
Nvidia was planning to directly compete against hyperscalers through DGX Cloud. They cancelled public DGX Cloud access when they found that investing in Neoclouds would accomplish the same goals without having to compete against their biggest customers.
If you're Nvidia, it's smart because Neoclouds that you have a large stake in will deploy your full stack from GPUs to networking to storage racks. They will share valuable usage data back to you so you can design a better next generation. Hyperscalers are likely a lot less cooperative, prefer to use their own designs if possible, and will guard their usage data.
- you fund a new company and sign long terms contracts with it - this new company uses the money you gave it and a lot of debt (backed by long term contracts) to build datacenters and buy a lot of GPU - your figures look great
What happens when they run out of debt or funds? If they reach some kind of profitability it's not a big deal, but if not ...
EDIT
Forget to mention the buyback of unused capacity problem: what happens to your figures when you have to buy back tons of unused GPUs?
It being that size, lasting for that long, and the total lack of viable products created by it are the problem. Financing only adds leverage, that makes every loss or profit larger.
If their business model thinks they can make a profit doing it this way, why stop them?
The core problem here seems to be that people think your supplier having an equity stake in your company is wrong or risky.
It's irrelevant.
> If their business model thinks they can make a profit doing it this way, why stop them?
I don't think someone needs to stop them, but there are some legit questions that need an answer:
- what happens to all these companies when growth decelerate or stop?
- what happens to nvidia stock when it has to buy back unused gpus?
- what are the risk that a sectorial financial crisis turn into a major economic crisis?
If these were all private entities, I think it'd be okay.
But they're public entities and they're using the pittance of investment as a force multiplier on their stock price, which they're then regularly using to raise capital.
A lot of dumb money in retail investors (as well as corporate) are a big reason this valuations bubble is occuring - which is really the elephant in the room. It's not that the tech isn't real. It's that the valuations behind it have already priced in maybe a decade of profit that hasn't come close to materializing for the LLM vendors; although, the shovel sellers and makers are doing phenomenal - and they have a vested interest to keep the party going with many sweetheart financing/equity deals.
But "invest in companies that may grow your own TAM" is an ancient strategy. Sometimes it works, sometimes it doesn't (like any strategy).
I'm not disagreeing with you, just saying it's business as usual.
Who gives a ** if you've seen it before, it's now a large scale issue. Stop trying to downplay it like it's a book you've read the second time.
What is the end of this sentence?
And if it is, it's not a problem!
And if it's a problem, it doesn't affect me!
And that was rife with scams, chicanery, and nonexistent investments. As well as needing lots of GPU-filled power hungry data centres.
So I think a lot of people are viewing the AI boom through the same lens.
There are a few outliers like Meta's basket of currency crypto attempt and Sam Altman's World Coin.
Meanwhile, the entire tech industry has embraced LLMs one way or another.
And tech venture capitalists, Altman and Musk were big boosters of cryptocurrencies.
To an outsider, cryptocurrency cones from the tech industry, despite the fact Apple and Google didn’t bet big on it.
It can be useful and also be a complete and utter fucking scam the way it's "produced" and sold.
With just these 2 comments, now I'm really gonna read that article.
https://isaiprofitable.com/
The only profitable company is the one running the scam.
"Furthermore, in the case of CoreWeave, Nvidia has also provided a significant financial backstop against unsold GPU capacity. Under the agreement with an initial value of $6.3 billion, “in instances where [CoreWeave’s] datacenter capacity is not fully utilized by its own customers, NVIDIA is obligated to purchase the residual unsold capacity through April 13, 2032.” In other words, Nvidia is committed to purchasing unsold GPU capacity if CoreWeave is unable to find another buyer. With an initial value of $6.3 billion, there is the potential that the arrangement could become larger over time."
I don't know how Nvidia is handling Coreweave GPU sales revenue in their accounting, but it sounds to me like it should have a pretty big asterisk attached to it. It's more like a consignment arrangement than an actual sale. And it obviously creates a huge incentive for Coreweave to over-order GPUs, since there's no risk (I doubt they're paying cash up front).
The sale of the GPUs by Nvidia to CoreWeave is real. CoreWeave pays Nvidia cash and becomes the owner of the asset, so it's properly booked as a sale. If it can't sell capacity, the GPUs are not returned to Nvidia.
CoreWeave is using debt to make the purchases but the backstop provided by Nvidia ostensibly helps it get better loan terms. That doesn't change the accounting.
If Nvidia has to purchase unused capacity, it simply becomes an operating expense for Nvidia.
Nvidia's exposure is the $6.3 billion backstop obligation and the equity it holds in CoreWeave.
According to the article, the $6.3B is a floor, not a ceiling. And it's not clear whether CoreWeave is actually paying cash or getting the GPUs on credit. If the full amount is getting booked, it's an accounting loophole that's being exploited. If GM sells Hertz a million cars, but says "Hey, we'll buy these back if you can't rent them," can GM book all those cars as actual revenue? What if Hertz only has to pay 10% up front and the rest in 5 years?
How are we still saying there is no outside money flowing in? Demand is so great that no one has any extra capacity.
And clearly, the more compute we have, the better the results. AI intelligence has not hit a ceiling yet. More compute means more training, more inference, more thinking, more verification, more multi-agent work.
Depends if they actually got the $2b in real money. There's a difference.
It's a big deal if no money was involved. Nothing even entered the company directly. Some deals have structured with Special Purpose Vehicles where money goes to the SPV. The SPV buys GPUs with it (from Nvidia). GPUs is loaned back to the company involved. So this company is stuck with this GPU rental, which may or may not be what they want and not $2b.
This sounds like a bad deal? So Nvidia had to sweeten the deal and promise min utilization on those GPUs by renting it themselves even if they don't need it.
So what's income and what's expense here?
That's the problem. It's inflated and messed up.
https://investors.coreweave.com/news/news-details/2026/NVIDI...
One aspect of the profitability might be the utilization and the pricing a few years down the line for slightly older hardware. Already now it seems like the increased processing you get from newer devices versus the cost difference makes something like an H100 or even A100 significantly less desirable than newer more powerful ones. As an individual, I am happy to be able to get an H200 on demand, but the B200 or B300 can do so much more work with optimized software and models for only modestly more cost that if those become available then from a business perspective you really have to prefer that if you can keep it occupied.
Then with Vera Rubin being like 3 times more effective or whatever, that adds a new layer of gradual obsolescence. So the question is can they keep the pricing up on the older ones a few years down the line enough to fill out the end of those expected payback periods.
The real boogeyman for a neocloud that has heavily invested in expensive Nvidia hardware might be a variation of that beyond Nvidia with startups that have even more dramatic efficiency increases pushing the leading edge even further. For example, if companies like Mythic AI and d-Matrix could somehow rapidly rapidly scale, that would push prices down for all of Nvidia hardware that is significantly less efficient.
I guess so far it doesn't look like any startups with really big efficiency breakthroughs are even close to being able to scale like Nvidia though, especially with the manufacturing and power crunch. But I suspect some of that is because of favoritism and strong arming protecting investments rather than a free and fair ecosystem.
They don't expect to keep the prices flat over time, and everyone involved will have planned for this. Prices are highest when they're the newest and greatest (part of why it's valuable for neoclouds to be first in line for new models), and drop year by year as newer GPU models can do equivalent work at lower cost.
You can see a pretty cool dataset of this at [1]; H100 prices where $3/hr in 2023, and dropped linear-ish to $1.75/hr by 2025. And also the notable exception that prices are up this year due to shortage.
[1] https://semianalysis.com/gpu-pricing-index/
Micheal berry doesn’t know shit about GPU pricing or depreciation schedules. A100 demand is very high and easily 2 dollars an hour for reserved right now.
B200 and Vera reubin don’t help much if you don’t benefit from quantization, and that’s exactly my situation and many other AI research orgs situation.
A100s are going to continue making money per hour until 2030. Mark my words.
think about stuff like pork barrel funding for aerospace, which props up jobs, which generates funding for political campaigns that perpetuate pork barrel funding.
But really it's all more like the railroad panic of the 19th century. Investing too much, too soon, in something that does have promise, but not if you can't pay for it.
For almost everything else, the answer is no. No one else would pay the real costs to run them.
It'll require the whole industry to shrink down massively compared to what we're seeing now - down to a profitable (and much smaller) core.
If there is any data to support this, please share.
/s because Poe's law is increasingly inescapable
CoreWeave feels very YC-ish. I thought I had an in as a referral for a position there and got interviewed by someone who knew a lot of my peers where I worked. Dude seemed to ask very textbook style questions that you would only learn if you went to a school system for this particular position/subject. I guess I didn't answer to their satisfaction despite knowing more than them on almost everything else. I suppose I'm still bitter seeing as I interviewed with them three times for two different roles. Absolutely wild.
FWIW, I’ve referred someone 3 times to the same position because I’m very sure he would be a good fit, and I’ve seen his work.
But for trivial reasons (“He doesn’t seem enthusiastic enough” and “the other candidates are better at promoting/selling themselves”) , a couple of managers that are above me in seniority (and directly in the hiring loop) just refuse to pass said person.
In the end he’s stopped applying, and I feel shitty for referring him.
That puts a cap on surplus (potentially unused?) datacenter capacity that's around by the time the AI bubble pops.
Any surplus after a pop will be sold for market value and lead to more new cloud provider startups and co-location options.
[0]: https://en.wikipedia.org/wiki/California_gold_rush
LLMs are actually useful, people are willing to pay for access to them, and they do genuinely enable things that were unrealistic or impossible before. The advances in image, video, and sound models since 2020 are also striking but likely won't be as transformative as LLMs.
That being said, I don't think it's unlikely that we'll see a plateauing of progress followed by a strong crash/correction in the market a-la dot com. The Allbirds situation absolutely has echoes of pets.com.
I also feel that commodification is coming for models, training/inference hardware, and software (e.g. CUDA), as it has for nearly everything else useful in tech. So I expect valuations driven by unique advantages here to be eroded over time (Think Sun and SCO after Linux on cheap x86 servers became the norm).
It is. The GPUs go on to be used to get loans to then get more GPUs.
Now I've got the feeling they don't have huge amounts of GPUs sitting in their DCs, but rented for Opex. In case the bubble pops they might get it at discount as CapEx (like Amazon did with dark fiber after the dotcom bubble).
Nobody lives in GPUs and what was the ratio of equity/debt for the toxic assets in 2007?
Yup. Add to that the decade worth of ZIRP following the 2007ff crash and Covid... all that money has to exit the system again eventually.
Financing is circular because creating a liability for one party (debt) creates an asset for another (the bank) off of which more debt can be secured
A bank / financier sells trust and reassurance. They otherwise invent most money from thin air.
It may be fine, or not. It it has been a frequent type of manipulation to obfuscate the real accounting situation.
Manyfacturers aren't artificially restricting supply, they're running fabs full-tilt. You could want them to build more fabs to meet demand. Which they are, but at a more modest rate than what you would want, because those manufacturers have been burned in previous boom-bust cycles. Never mind that fab-construction lead times are measured in years.
And what's stopping you from fabricating & selling RAM? I've read it's very profitable! Oh yeah, it takes many $B to pull a SOTA fab out of the ground.
Vendors price-gouging? Probably. Wouldn't you?
TLDR; it's not a monopoly issue. This is a high-tech specialized market where a ridiculous spike in demand is near-impossible to cater for. You want some new RAM-heavy gadgets? Shell out $, adjust your RAM 'wants', or be patient.
Ram was a commodity and now it's not. Markets are not supposed to decommodidy things because a single company got so big that it could buy half the worlds supply. That's not a normal situation or a healthy market, and people can feel however they want about it. They had a previous good taken from them.
It’s also by many accounts a bit of a weird company to work for, but they can afford to pay above-market for many roles.
Certainly looks like they were trying to get out, and were rich enough to actually pull it off (that can't have been cheap). Also they deserve some serious kudos for actually trying to protect them.