Nvidia (NVDA) and Microsoft (MSFT) don't just sell OpenAI chips and cloud time. They also fund the bill.

Nvidia (NVDA) and Microsoft (MSFT) don't just sell OpenAI chips and cloud time. They also fund the bill.

Key points

  • OpenAI's priced chip and cloud deals add up to about $710 billion, and three of its suppliers, Nvidia, Microsoft, and Amazon, also own its stock.
  • Nvidia's $100 billion OpenAI plan was never signed. Nvidia put in $30 billion instead, in a round that valued OpenAI at $852 billion.
  • OpenAI now plans about $600 billion of compute spending through 2030, though Sam Altman was still citing $1.4 trillion of commitments in August.
  • The first $18 billion phase of OpenAI's Broadcom chip still has no announced Microsoft purchase commitment, while Nvidia is guaranteeing part of an Ohio data center lease for OpenAI.

Nvidia (NVDA) sells GPUs to CoreWeave (CRWV). Nvidia also owns about 9% of CoreWeave. And in September 2025, Nvidia agreed to buy any CoreWeave cloud capacity left unsold, an order worth $6.3 billion to start. That makes Nvidia CoreWeave's supplier, shareholder, and customer at the same time.

That's a circular AI deal. A company invests in its own customer, sells that customer product, and in some cases buys that customer's output back. By mid-2026 it was the thing Wall Street couldn't stop arguing about, because versions of this loop show up almost everywhere OpenAI touches, and the dollar figures run into the hundreds of billions.

Nvidia and Microsoft own pieces of what they sell to

OpenAI is where the biggest version of this plays out. In September 2025, Nvidia and OpenAI announced a plan for up to $100 billion, paid out as OpenAI deployed 10 gigawatts of Nvidia hardware, in exchange for non-controlling shares. Two months later, Nvidia's own quarterly filing said there was "no assurance" of a definitive agreement. By early February 2026, the talks had stalled, and the original $100 billion plan was never signed.

What actually happened is smaller and built differently. OpenAI announced a $110 billion funding round on February 27, 2026, and closed it at $122 billion on March 31: Amazon put in up to $50 billion, with $35 billion of that tied to unspecified milestones and to OpenAI going public, SoftBank $30 billion, and Nvidia $30 billion, valuing the company at $852 billion after the money. Nvidia's $30 billion is a straight equity stake, not tied to gigawatts the way the original plan was, and the $100 billion figure people still quote is more than three times too big.

Then Nvidia found a new way to sit on both sides. On August 17, 2026, Nvidia said it will guarantee part of the lease and power payments on a data center in Pike County, Ohio, that OpenAI will lease for 20 years and fill with Nvidia chips, built and owned by SoftBank's SB Energy, in which Nvidia is also investing $1.5 billion. The initial buildout is 4.25 gigawatts with an option to reach 8. The guarantee phases in from 2028 to 2030, and Nvidia rejects the label "circular financing." Nvidia's 10-Q for the quarter ended July 26 puts $90.7 billion of other companies' stock on the balance sheet, a further $25 billion of equity commitments unfunded, and guarantees capped at $105 billion for the OpenAI site off the balance sheet entirely. The same filing shows one direct customer at 16% of second-quarter revenue.

Microsoft (MSFT) got there a different way. Microsoft had committed $13 billion to OpenAI since 2019 before the two sides tore up the old arrangement and started over on October 28, 2025. What came out the other side was a new corporate shell for OpenAI's for-profit business, and Microsoft holding about 27% of it, a piece worth close to $135 billion the moment the ink dried, and roughly $230 billion at the $852 billion valuation set in March 2026. The same deal sent money the other way: OpenAI contracted to buy an incremental $250 billion of Azure services, according to Microsoft's own 10-Q. Microsoft is OpenAI's cloud supplier and one of its biggest owners at once.

Amazon (AMZN) joined the loop in February 2026. Alongside its equity commitment of up to $50 billion, OpenAI expanded its existing $38 billion AWS contract by $100 billion over eight years, including capacity on Amazon's own Trainium chips. That makes Amazon an investor and a supplier to OpenAI too, and the chips in that loop are Amazon's, not Nvidia's.

AMD did it backwards

AMD (AMD) built the same kind of loop from the other direction. OpenAI is on the hook to deploy up to 6 gigawatts of AMD's Instinct chips over the next several years. What AMD handed back was the right to buy 160 million AMD shares for a penny apiece. Exercise it fully and OpenAI would own something close to a tenth of AMD. The shares only vest in pieces, tied to OpenAI actually hitting purchase targets and AMD's own stock clearing specific price levels, and the window to cash any of it in runs out in October 2030. AMD's finance chief, Jean Hu, said only that the partnership is "expected to deliver tens of billions of dollars in revenue for AMD." AMD then issued Meta an identical warrant on February 23, 2026, so two customers now hold penny warrants on 320 million AMD shares between them, worth about $152 billion at the July 31 close; as of AMD's first-quarter 10-Q, none of it had vested or touched the financial statements. In AMD's version, the buyer ends up owning a slice of the seller.

Oracle, Broadcom, and CoreWeave skipped the equity, not the risk

Oracle (ORCL), Broadcom (AVGO), and CoreWeave don't own a piece of OpenAI, and each one's exposure looks a little different.

CompanyOpenAI commitmentNotable detail
Oracle (ORCL)About $300 billion of cloud capacity, part of the Stargate buildoutOracle's quarter ended August 31 showed the cost of carrying it: remaining performance obligations of $664 billion, cloud infrastructure revenue up 121% to $7.4 billion, and free cash flow of negative $5.4 billion. In September the Financial Times reported $18 billion of loans on an Oracle-linked data center quoted at 89 to 91 cents
Broadcom (AVGO)10 gigawatts of custom AI chips, OpenAI's own designFirst chip, nicknamed Jalapeño, unveiled in June 2026, with a first physical sample delivered to OpenAI. OpenAI presented it at Hot Chips on August 25, claiming it beats Nvidia on inference efficiency
CoreWeave (CRWV)$11.9 billion to start in early 2025, expanded twice to $22.4 billion by that SeptemberNvidia holds about 9% of CoreWeave on its own, a separate stake

Their risk runs straight through the contract instead. They're financing a massive buildout on the promise that OpenAI can eventually pay for it, and a contract is only as good as the company that signed it. Our neocloud explainer gets into how CoreWeave and companies like it built an entire business model on exactly this kind of concentrated exposure to one customer.

The 1999 comparison, and where it falls apart

The comparison everyone reaches for is Cisco lending money to dot-com-era internet providers so they'd buy its routers, until the providers ran out of real customers and the stock went down with them. It's a fair comparison, and by now a familiar one.

The numbers behind it are real. OpenAI's own forecast, reported in January 2026, projected a loss of about $14 billion in 2026, while the company keeps committing to hundreds of billions more in infrastructure. Bloomberg mapped the loop in an entire graphics package, JPMorgan's asset management arm published a note asking outright whether the circularity itself is a warning sign, and Michael Burry, who's made his own bet against AI-linked chip stocks, keeps making the same point. When a chipmaker's revenue depends on a customer that is still losing money, and the chipmaker is also funding that customer's spending, some of that revenue is the chipmaker's own money coming back.

Scarcity is where the comparison breaks down. Cisco had warehouses of unsold routers by 2001, waiting on ISPs that couldn't scrounge up enough paying subscribers to justify what they'd already bought. Nvidia's chips have stayed backlogged for years running, and actual demand for compute keeps outpacing what the industry can physically build. Funding a customer that can't get enough of your product is a different bet than propping up a customer nobody wants. OpenAI isn't one fragile ISP either. Its spending is spread across Microsoft, Amazon, Google, and a long list of businesses building on top of its models, and its backers are some of the best-capitalized companies in the world. None of that erases the risk, but the base is much bigger and better funded than what came apart in 2001. We've gone deeper on that side of the argument in our bubble-or-dip breakdown and the hyperscaler capex numbers behind it.

Where it actually broke

The clearest sign that circular financing has a ceiling showed up in Broadcom's chip deal. Broadcom would only finance the first phase, roughly 1.3 gigawatts and $18 billion, if Microsoft agreed to buy around 40% of the output. Under the plan, Microsoft would install the chips in its own data centers and rent them back to OpenAI. Microsoft hadn't signed as of May 2026, and Broadcom's stock dropped on the news. As of mid-October, no Microsoft commitment had been announced, even after OpenAI took the chip to Hot Chips in August. An announced partnership is not a financed contract, and this is where that gap showed up first.

$1.4 trillion was the number Sam Altman started citing in late 2025. Then, in February 2026, OpenAI walked investors down to something smaller, somewhere near $600 billion in compute spending through 2030. Both numbers are still in use. Altman was citing about $1.4 trillion of commitments over eight years in August 2026. He also told Fortune in September that OpenAI will not go public in 2026, so the bill keeps getting financed privately for now.

What actually matters if you're holding these names

The filings say more than the press releases. A 10-K or 10-Q shows whether a company has actually booked revenue from one of these deals, not just announced one, and whether its stake in a customer still sits on the books at cost or is already carrying paper gains that could vanish just as fast as they showed up. Pay attention to the language too. Nvidia's own filing said there was "no assurance of a definitive agreement" on its $100 billion OpenAI plan, two months after the headline. And watch which buyers actually have other options. Microsoft's hesitation on the Broadcom order mattered more than it would from a smaller player, because Microsoft, unlike OpenAI, doesn't need any single vendor to keep running.

OpenAI is not the only lab wired this way. We mapped Anthropic's four money loops separately, and Nvidia hardware is named in only one of them.

Every ticker above links to its live SEC filings. Run any of these names through our stock score tool too. It grades profitability, dilution, and balance sheet strength in about ten seconds.

This is general market commentary and opinion, not investment advice. Figures reflect public disclosures as of October 15, 2026, when this piece was last updated. Markets can go down as well as up, and you can lose money. Always do your own research and consider speaking with a licensed financial professional before making any investment decision.

Frequently asked questions

What is a circular AI deal?

It is when the same company sits on both sides of an AI transaction, most often a chipmaker or cloud provider that invests in an AI company while also selling it chips or compute. Nvidia (NVDA) owning about 9% of CoreWeave (CRWV), supplying its GPUs, and agreeing to buy CoreWeave's unsold cloud capacity is a clean example.

Does Nvidia own part of OpenAI?

Yes, but not the way it was first announced. Nvidia and OpenAI announced a $100 billion, 10-gigawatt plan in September 2025 that had stalled by February 2026 and was never signed. Instead, Nvidia put $30 billion into OpenAI as part of a funding round announced at $110 billion in February 2026 and closed at $122 billion on March 31, alongside Amazon (up to $50 billion) and SoftBank ($30 billion), valuing OpenAI at $852 billion. In August 2026 Nvidia also agreed to guarantee part of the lease on an Ohio data center OpenAI will fill with Nvidia chips.

Does Amazon have a circular deal with OpenAI?

Yes. Amazon (AMZN) committed up to $50 billion to OpenAI in February 2026, with $35 billion of that tied to milestones and to OpenAI going public. In the same deal, OpenAI expanded its existing $38 billion Amazon Web Services contract by $100 billion over eight years, including capacity on Amazon's own Trainium chips.

How much has OpenAI committed to spend on AI infrastructure?

The deals with a disclosed price add up to about $710 billion: about $300 billion with Oracle (ORCL), an incremental $250 billion of Microsoft (MSFT) Azure services, $138 billion with Amazon Web Services, and $22.4 billion with CoreWeave. Broadcom (AVGO) and AMD did not put a price on theirs. Sam Altman has cited about $1.4 trillion in commitments over eight years since late 2025, while in February 2026 OpenAI told investors to plan around about $600 billion of compute spending through 2030.

Is this the same thing as the dot-com vendor financing bubble?

There are real similarities. Cisco lent money to internet providers who used it to buy Cisco gear, and the arrangement collapsed once demand did not show up. The counterargument is that AI chip demand has stayed backlogged for years, unlike the oversupplied ISPs of 2000, though analysts including JPMorgan's asset management arm still flag the structure as a real risk worth watching.

What happened with the Broadcom chip deal in 2026?

Broadcom agreed to finance the first roughly $18 billion phase of its custom chip buildout with OpenAI only if Microsoft committed to buying about 40% of the output. By May 2026, Microsoft still had not signed that commitment, and Broadcom's stock fell on the news. As of mid-October 2026, no Microsoft commitment had been announced, though OpenAI presented the chip at Hot Chips in August.

More on NVDA and MSFT

Latest stories

Dennis Singleton
Dennis Singleton

Dennis Singleton was born in Australia and later moved to the United States. He has spent years following the markets, but what keeps his attention is how AI is built. He writes about the companies behind the technology, from semiconductor designers and advanced packaging to photonics, memory, networking, and the hardware powering modern AI. His approach starts with filings, earnings, and industry research, then translates the important details into clear, straightforward analysis without unnecessary hype.