Forgot your password?
typodupeerror
The Almighty Buck Businesses

Nvidia In Talks With OpenAI To Guarantee $250 Billion Financing For Data Center (reuters.com) 41

An anonymous reader quotes a report from Reuters: Nvidia is in talks to provide roughly $250 billion in financing guarantees for OpenAI as part of a massive data center project, the Wall Street Journal reported on Sunday. The backstop from Nvidia would help the ChatGPT maker lease a 10-gigawatt project that SoftBank's energy subsidiary is developing in southern Ohio, the newspaper said, citing people familiar with the matter. For OpenAI, a deal would be the first step toward controlling its own infrastructure instead of renting it from Microsoft, Amazon, and Oracle, while for Nvidia, it would guarantee demand for its chips for years to come.

The project is expected to cost more than $500 billion in total, including the chips inside the data center, according to the WSJ. The $250 billion guarantee covers the data center lease and debt financing, but would not cover the Nvidia chips inside the center, the WSJ said, adding that the chipmaker was also discussing financing OpenAI's chip purchases worth up to $350 billion. Nvidia's backing would support financing vehicles aimed at reassuring lenders about the project's funding, the report added. The first phase of the project is expected to be finished in 2028, with around 800 megawatts of power, the Journal said.
The report notes that the U.S. government will control access to the power, while Japan will fund it separately under a trade agreement tied to Tokyo's $33 billion investment in a natural gas plant. Commerce Secretary Howard Lutnick will reportedly help determine who receives access.

Nvidia In Talks With OpenAI To Guarantee $250 Billion Financing For Data Center

Comments Filter:
  • with a quarter trillion than putting a shitload of people out of work.
    • Re: (Score:3, Insightful)

      by gweihir ( 88907 )

      I think this may be an entirely temporary effect in most (not all) cases. The cult-like claims of the ones praying on the great LLM altar become more obvious nonsense every day.

    • by allo ( 1728082 )

      It looks like companies in capitalism don't care about "us as a species" but about the market.

  • When this all comes crashing down. As it must. And little of value will be lost.

    • Even once the bubble pops, that's only the financial side of it... AI won't be going anywhere.

      • by gweihir ( 88907 )

        You think somehow magically the mountains of money needed to keep it running and updated will appear? Talk about a childish fantasy ...

        Yes, sure, some will remain, like maybe what you get for free with search results at this time. And there will be specialist models. But generally available general LLMs of current sizes and power? No. Not a snowball's chance in Hell.

    • Probably. Google, Amazon and Microsoft all have their own chips for AI and probably won't need Nvidia in the long run. Anthropic will probably have its own chips, too. So when OpenAI tanks the round tripping is going to be terrible for Nvidia. But if TSMC hoards cash they'll be able to buy Nvidia for peanuts and make a bigger profit off all those chips being sold to gamers.

    • by allo ( 1728082 )

      Because nobody will have a use for compute anymore?

      • by gweihir ( 88907 )

        The term "compute" serves to obscure that this is not general-purpose CPU power. Some people fall for that deception. This stuff is about as valuable as crapcoin miner hardware after that crapcoin crashed.

        • by allo ( 1728082 )

          It's a huge amount of GPUs. That's what you need for most "big data" problems, from AI to huge simulations and to the weather forecast. If you want to even for crypto mining, but this scales (by definition) to the point it becomes unlucrative, so rather do something useful with that compute.

          • by gweihir ( 88907 )

            Nope. It is NOT GPUs. They are, at best, GPU-like. And no, GPUs not generally useful as compute hardware. They are useful for some very special, very limited things.

            • by allo ( 1728082 )

              You're wrong. Things like let's say a B200 is exactly the same GPU as you would use for the numerical weather simulation.
              Google has used TPUs, but I think even they seem to have moved most workloads to GPUs again, probably to streamline things.

  • by MachineShedFred ( 621896 ) on Monday July 27, 2026 @11:43AM (#66259046) Journal

    1. OpenAI wants to build a ridiculous amount of data centers, using a ridiculous amount of GPUs.
    2. Nvidia wants to sell a ridiculous amount of GPUs.
    3. OpenAI runs into a funding wall
    4. Nvidia gives them loan guarantees to build data centers, where they will be getting back a significant amount of that loan for their own products, while charging interest on the whole lot

    Sounds like OpenAI is looking to get purchased by Nvidia at severe discount when they can't pay the loan.

    • by Whateverthisis ( 7004192 ) on Monday July 27, 2026 @12:48PM (#66259138)
      You're missing a big one:

      5) Nvidia doesn't have the money to guarantee these loans. Nvidia has in current assets around 125B; about 50% to guarantee these loans.

      We have a major issue here. These guys had all this circular finance with equity, where Nvidia would invest in OpenAI who then turn around and bought chips from them, adding to NVidia's revenue. Very little cash was exchanged, the equity was exchanged, but then the equity went up (in a private valuation) because OpenAI was spending more on Capex. Now people are calling BS on this circular finance non-sense, so they're turning to debt.

      But the debt is a problem. Bloomberg came out (non-paywall source from Futurism [futurism.com]) with a report saying they're using Enron-style accounting, essentially shell companies to hold the debt while they own 20% of the shell, so the debt stays off their books. Debt financing like this however is extremely reckless; it is the source of the Enron crash, the Lucent technologies crash (Lucent also got into trouble with risky customer financing and crashed big time), the 2008 financial crisis, etc.

      These guys are headed for a fall and will bring the economy into recession with it, right when the US government is too over-leveraged too.

      • Where, in the middle of all this exchanged money and new created value, is all the tax revenue?

        right when the US government is too over-leveraged too.

        Ah, right.

    • I think there are a few improtant subtleties to it. Namely that Nvidia is providing a [loan] guarantee and not an actual loan. It is also unclear if this guarantee is on the data center, or for the power purchase agreement. So, when OpenAI can't pay for whatever it is, Nvidia is responsible for payments. Depending on the terms of the deal, the risk could actually be quite low for Nvidia. Something similar goes for financing chips with a >300% margin-- payments after the a year or so are pure profit.

  • Oh golly more circular deal making to prop up the hype that are in no way like Enron. ;-)

  • The best team at OpenAI left and started Anthropic. OpenAI is on the downtrend, even Microsoft is getting out of their deals with them. I'm surprised Nvidia would consider vendor financing of that size. ChatGPT may still be the 'name brand' but I don't meet or see anyone besides high school students writing papers that still use ChatGPT over Claude, Grok, or Gemini.
    • And absolutely nobody is using Llama.
    • Someone needs to fail. This bubble needs to pop yesterday! The OpenAI guys seem nice and all, but we need a big public failure in AI. I want to be able to buy devices without a massive AI tax. The bullshit hype is insufferable, but the chip cruch is killing us all...promising stuff few of us want to get these overleveraged HALF TRILLION datacenters built...while taking things we enjoy: our devices, gaming-consoles, etc and making them expensive and in short supply. There's NO SIGN they will succeed at
  • No it isn't, they are leasing that 10Gwhat? data center (oh sorry, 'project', apparently calling them a data center is bad publicity).
  • by CEC-P ( 10248912 ) on Monday July 27, 2026 @12:52PM (#66259144)
    How about they all fuck off, give us our RAM and GPUs back, and they can't build a datacenter until they can build their own chips and RAM modules from scratch at a separate FAB and supply their own power privately. How about that? Anyone in on that idea?
  • It sounds like the guarantee is on a PPA and not data center build based on the summary. That is a pretty important distinction as a PPA is a very different kind of risk than a compute facility (especially given grid shortfalls for projected growth).

  • I remember the buildups to the dotcom crash and the 2008 crash. There were lots of signs and doom sayers and there were also lots of people pulling their money out of the NASDAQ and out of banks exposed to real estate bullshit, respectively. This time it seems like the whole world has gone insane and is drunk on meme stock cognac. Anything that involves AI and/or Elon Musk is making paper millionaires. It's like someone went back in time and removed every American equities disaster from the history curricul

  • All of these motherfuckers need to be rounded up and hauled off to jail.

  • "Who's more foolish? The Fool? Or the fool who follows?" -- or maybe Vizzini from The Princess Bride, "...Plato, Socrates, Sam Altman? Morons, all of them." Well, something like that. How about, "Burn, baby, burn, Token Inferno..." No? Well, maybe you were just born too late.

"How many teamsters does it take to screw in a light bulb?" "FIFTEEN!! YOU GOT A PROBLEM WITH THAT?"

Working...