AI's trillion dollar deal wheel bubbling around Nvidia, OpenAI

How to build a trillion-dollar industry: Step 1, invest in your customers. Step 2, sell them stuff

Feature In late 2025, a series of multi-billion-dollar deals in the artificial intelligence sector is causing déjà vu among industry veterans. Money, computer chips, and cloud credits are rotating in a closed loop among a handful of companies: Nvidia, OpenAI, Microsoft, Oracle, AMD, CoreWeave, xAI, and a few others. This has fueled a trillion-dollar AI boom or bubble built on intertwined investments and contracts.

The arrangement of these deals is so circular that dollars spent by one player often return as revenue for another. This gives the impression of breakneck growth. Critics are increasingly comparing this AI money-go-round to the late 1990s dot-com bubble. They question whether today’s AI valuations are inflated by self-reinforcing deals rather than organic user demand.

A web of deals

At the heart of this circular economy is Nvidia, the world’s most valuable semiconductor company, which recently reached a $5 trillion market capitalization. Nvidia’s GPUs, moreso than any other company's, have become the critical hardware that powers advanced AI models like ChatGPT.

Surrounding Nvidia is an expanding constellation of partners and customers, many of which have also become its investors and investees. No deal epitomizes the circular AI economy better than the agreement between Nvidia and OpenAI announced in fall 2025.

In September, Nvidia agreed to invest up to $100 billion in OpenAI to finance a massive buildout of AI data centers. The plan is for OpenAI to construct at least 10 gigawatts of new data center capacity (enough electricity to power a major city). In return for Nvidia’s funding, OpenAI committed to purchasing and deploying millions of Nvidia GPUs at those facilities.

This investment-for-hardware swap means Nvidia is bankrolling its own future sales. The startup gets the cash to expand, and Nvidia is guaranteed to be the supplier for that expansion.

Critics immediately flagged the arrangement as circular, essentially moving money in circles to pad revenues. It is different from the dot-com era, where companies bought each other’s ads with no real product. Here, physical assets (AI chips and data centers) are being built. Both companies are doing what they are set up to do: OpenAI needs chips, Nvidia sells chips, just with intertwined balance sheets.

Still, the sheer scale of the deal has raised eyebrows. $100 billion is an unprecedented sum for one tech company to invest in a startup. For context, OpenAI was valued at around $30 billion in early 2023. This latest deal implied a far higher valuation and a massive bet on future AI demand. It also cemented Nvidia’s role not just as a supplier but as one of OpenAI’s most prominent backers.

Nvidia boss Jensen Huang, in a recent podcast with Bill Gurley and Brad Gerstner on Bg2, said, “I think that OpenAI is likely going to be the next multi-trillion-dollar hyperscale company. If that’s the case, the opportunity to invest before they get there, this is one of the smartest investments we can possibly imagine. And you've got to invest in things you know. The return on that money is going to be fantastic.”

OpenAI’s cloud gambit

Just weeks after the Nvidia pact, OpenAI struck another massive agreement, this time on the cloud service side. OpenAI confirmed a $300 billion deal (dubbed as Project Stargate) with Oracle Corporation to purchase cloud computing capacity over the next five years. The deal would see OpenAI use Oracle’s cloud infrastructure to run and train AI models at scale, with Oracle adding 4.5 GW of new data center capacity by 2027.

The circular twist? To fulfill OpenAI’s compute demand, Oracle must spend heavily on hardware, primarily buying Nvidia’s chips. According to the Financial Times, Oracle plans to spend around $40 billion to acquire roughly 400,000 of Nvidia's top-tier GB200 chips to power Stargate's flagship data center in Abilene, Texas.

This Stargate triangle created instantaneous value on paper. When news of the OpenAI-Oracle deal broke, Oracle stock spiked 36 percent in a day, its most significant jump in decades. Even Oracle co-founder Larry Ellison’s net worth swelled by $88 billion virtually overnight, briefly making him the world’s richest person.

Nvidia’s stock also popped 4 percent on the Oracle news to add another $170 billion to its market cap. These deals provide planning security for all parties involved and help dominate AI infrastructure. Still, they also raise important questions about self-dealing and systemic risk if one link in the chain falters.

AMD’s entry in the GPU market

Nvidia isn’t the only chipmaker in OpenAI’s orbit. In October 2025, OpenAI announced an alliance with Nvidia’s main rival, AMD. This announcement signifies the considerable demand for compute chips and the need to diversify its hardware suppliers. The terms of the deal were for OpenAI to deploy 6 gigawatts of AMD’s Instinct GPUs in its future infrastructure. This commitment could translate to tens of billions of dollars in chip purchases.

In return, AMD granted OpenAI a warrant to purchase up to 160 million AMD shares, roughly a 10 percent equity stake, vesting in tranches aligned with the GPU deployment rollout. If OpenAI follows through on using all 6 GW of AMD hardware, it gets to become one of AMD’s largest shareholders at a steep discount.

Strategically, this deal secures a marquee customer for AMD and validates its AI chip as an alternative to Nvidia’s dominant GPUs. “This partnership brings the best of AMD and OpenAI together to create a true win-win, enabling the world’s most ambitious AI buildout and advancing the entire AI ecosystem,” said Dr. Lisa Su, chair and CEO of AMD.

“Our partnership with OpenAI is expected to deliver tens of billions of dollars in revenue for AMD while accelerating OpenAI’s AI infrastructure buildout. This agreement creates significant strategic alignment and shareholder value for both AMD and OpenAI and is expected to be highly accretive to AMD's non-GAAP earnings-per-share,” said Jean Hu, EVP, CFO, and treasurer of AMD.

The agreement highlights the frenzy in AI chip demand. OpenAI is so hungry for capacity that it’s willing to tie itself financially to its suppliers to secure supply.

The cloud middleman

No discussion of the AI circular economy is complete without CoreWeave, a once-small cloud startup that has become a major infrastructure partner for both OpenAI and Nvidia.

Starting in early 2025, OpenAI began shifting some of its AI training workloads to CoreWave and signed an initial contract worth up to $11.9 billion for GPU cloud services, followed by a $4 billion expansion in May. By September 2025, a third expansion of $6.5 billion brought OpenAI’s total CoreWeave commitments to $22.4 billion.

CoreWeave’s rapid scaling was enabled by Nvidia’s backing. Nvidia took an equity stake of more than 5 percent in the company during its funding rounds. In September 2025, Nvidia agreed to purchase $6.3 billion in cloud services from CoreWeave. This unusual agreement by Nvidia to pre-purchase capacity on CoreWeave served as a guarantee. Nvidia would pay for any of CoreWeave’s GPU time that wasn’t sold to other customers.

With Nvidia’s safety net, CoreWeave would confidently buy even more Nvidia chips to build data centers. In simple terms, Nvidia was funding a customer and propping up the supplier that served that customer, thereby increasing its chip supply.

Old big tech set the stage

The initial partnership of Microsoft and OpenAI in 2019-23 helped set the stage for today’s loop. Microsoft invested $1 billion in 2019, then reportedly $10 billion in early 2023, in OpenAI, eventually obtaining a roughly 49 percent stake (with profit-sharing terms). In exchange, OpenAI agreed to use Microsoft Azure as its computing platform for several years.

Fast forward to 2025, and the landscape has shifted. OpenAI’s new mega deals with Oracle, CoreWeave, and others signaled that Microsoft’s once exclusive hold on OpenAI’s compute needs has loosened. Microsoft quietly allowed OpenAI out of certain exclusivity to pursue its own cloud buildout. Microsoft itself has diversified, integrating AI models from other startups like Anthropic and even open-source models like Meta’s Llama.

The circular deal dynamic here is subtler but present. Microsoft gave OpenAI funding and cloud credits. OpenAI returns value by driving Azure usage and requires more infrastructure, which means Microsoft buys more Nvidia GPUs for its data centers.

Uncle Sam enters the loop

It’s not just corporations that are intertwined in the AI economy. The United States government itself has become both an enabler and a gatekeeper in these dealings. As part of the federal CHIPS Act initiative, the US government took an equity stake of roughly 9.9 percent in Intel, valued at $8.9 billion, and funded an additional $2.2 billion in grants to build domestic semiconductor manufacturing capacity.

In exchange, the Commerce Department received 274.6 million Intel shares, with an option to acquire an additional 240 million shares under certain conditions. Traditionally, government incentives are not taken in the stock market, yet this shows how strategically important chip production is to national security.

In the context of the AI circular economy, the government stake is a reminder that not all money in the loop is private. Public funds are being poured into shoring up the supply side rather than directly into AI companies, but it still feeds the broader ecosystem. If Intel succeeds in making advanced AI chips domestically, it could become a counterweight buyer/seller in the loop.

Washington regulates market access (like restricting China) in ways to constrain where the loop’s money can flow. This forces Nvidia, OpenAI, and others to focus inward and on allies.

This may intensify the circular deal pattern, but it could also mitigate some bubble risk by preventing over-exuberant expansion into geopolitically risky matters. It’s a thin balance between enabling innovation and preventing policy-driven artificial inflation of demand.

All these arrangements, with OpenAI at the center, might not please accountants, who tend to view trade of this nature as deeply flawed. This is an industry sector getting fatter by eating itself. The levels of funding are unprecedented.

The web OpenAI is spinning means its own success is inextricably linked to its suppliers. It's a clever move by CEO Sam Altman. However, if it backfires, he will set the industry back decades in terms of trust.

It took years for tech to recover from the dotcom bust, and with the level of investment in AI infrastructure currently propping up the US economy, keeping it out of recession, there's a lot more at stake. ®