Nvidia $NVDA just put real money behind its favorite metaphor. After two years of calling modern data centers “AI factories,” the chipmaker is going all-in with chips — and cash — to help build them.
On Monday, Nvidia and OpenAI announced that the chipmaker plans to invest up to $100 billion in OpenAI through a staged partnership that doesn’t just guarantee the startup first dibs on the world’s most coveted GPUs — it effectively makes Nvidia a co-builder of the infrastructure that will power the next wave of artificial intelligence.
The details “of this new phase of strategic partnership” will be finalized in the coming weeks, the two companies said in the press release. Nvidia’s stock was up around 3.5% on the news.
The deal is part supplier agreement, part financing package — and has a moat big enough to make rivals sweat. Nvidia will pour in money gigawatt by gigawatt as OpenAI builds data centers stocked with millions of its chips. The first of those, running on Nvidia’s coming Vera Rubin platform, is slated to switch on in the second half of 2026. By the time the build-out hits 10 gigawatts — about the output of several nuclear reactors — OpenAI will be operating the kind of compute grid that is usually reserved for national utilities.
This isn’t a one-time wire; it’s a letter-of-intent to invest progressively in cash and equity as each gigawatt is deployed — money that shows up when concrete, power, racks, and GPUs do. And the partnership formalizes a relationship that’s been implicit for years. OpenAI runs on Nvidia’s silicon, and Nvidia’s stock has rocketed thanks to AI labs such as OpenAI driving insatiable demand.
But this deal takes things further: Nvidia isn’t just selling the chips, it’s helping pay for the factories that use them. The company becomes OpenAI’s “preferred strategic compute and networking partner,” aligning hardware and software roadmaps so that model releases and silicon arrive in lockstep. That means Nvidia’s hardware and software evolve alongside OpenAI’s models, each feeding the other’s performance.
“NVIDIA and OpenAI have pushed each other for a decade, from the first DGX supercomputer to the breakthrough of ChatGPT,” Jensen Huang, the founder and CEO of Nvidia, said in a press release. “This investment and infrastructure partnership mark the next leap forward — deploying 10 gigawatts to power the next era of intelligence.”
“Everything starts with compute”
For OpenAI, the appeal is simple. GPUs are scarce, expensive, and often monopolized by the world’s biggest cloud providers. A $100 billion backstop all but guarantees supply at the scale needed to keep training frontier models. OpenAI, by putting Nvidia in its capital stack, is effectively buying certainty at a moment when delays could derail its model pipeline.
OpenAI CEO Sam Altman said in the press release that “everything starts with compute.” He added, “Compute infrastructure will be the basis for the economy of the future, and we will utilize what we’re building with NVIDIA to both create new AI breakthroughs and empower people and businesses with them at scale.”
For Nvidia, the benefits are equally obvious. Instead of waiting for purchase orders, it now has visibility into years of demand — millions of chips, bundled networking gear, and software licenses that will keep its order book fat through multiple product generations.
Nvidia’s proposed commitment is almost twice its entire fiscal 2024 revenue of about $61 billion, and even with a market cap north of $4 trillion, the company usually spends only tens of billions a year on new factories and equipment (its most recent capital expenditures guidance was of $66–$72 billion). That means that the OpenAI deal is bigger than a year of Nvidia’s sales and on par with its entire annual investment budget — a sign of just how central Nvidia sees OpenAI to its future.
The move also changes the competitive board. Microsoft $MSFT remains OpenAI’s largest investor and its primary cloud host, even though the two companies recently redefined their cloud ties. And other partners are in the mix. The two companies said in the press release that “this partnership complements the deep work OpenAI and NVIDIA are already doing with a broad network of collaborators, including Microsoft, Oracle $ORCL, SoftBank and Stargate partners, focused on building the world’s most advanced AI infrastructure.”
Oracle stock was up about 5% on Monday around 1 p.m. ET, as investors digested the ripple effect of what will be massive new data-center demand.
The partnership might leave rivals such as Amazon $AMZN and Google $GOOGL — who also depend heavily on Nvidia chips for their own AI clouds — wondering whether OpenAI has just leapfrogged them in the allocation line. AMD $AMD and Intel $INTC, meanwhile, face the prospect of being further marginalized if OpenAI’s next-generation compute is effectively pre-committed to Rubin-era GPUs.
Questions about the partnership remain
But Nvidia’s deal creates a lane: The supplier is now also the financier, and that’s an arrangement that could draw scrutiny from regulators. Governments in the U.S. and Europe are increasingly wary of concentration in AI, and a $100 billion deal between the sector’s most influential startup and its most indispensable supplier is bound to attract scrutiny. So questions about preferential access and antitrust might not be far behind. Nvidia and OpenAI can argue that the staged, milestone-based structure keeps the relationship transactional rather than controlling. But rivals likely won’t miss the chance to suggest the deck is being stacked if OpenAI gets first crack at Rubin-era parts while other companies wait their turn.
Meanwhile, actually turning on 10 gigawatts of compute requires solving the dull but brutal bottlenecks of grid interconnection and energy sourcing — challenges that could slow the rollout no matter how much money is on the table.
OpenAI’s scale was used as evidence in the press release that it can make use of the investment, citing “over 700 million weekly active users and strong adoption across global enterprises, small businesses, and developers.” For enterprises that increasingly rely on OpenAI’s APIs, the promise of more predictable compute capacity means fewer bottlenecks. For consumers, it suggests the cadence of ChatGPT upgrades could speed up.
“This partnership will help OpenAI advance its mission to build artificial general intelligence that benefits all of humanity,” the companies wrote.
The bet is clear. The scarcest commodity in AI today isn’t clever code or new architectures — it’s compute. By investing up to $100 billion, Nvidia is locking in its place at the center of that scarcity. OpenAI gets the money it needs to scale. Nvidia gets the certainty it craves. And the rest of the industry gets the message: The next phase of AI isn’t about demos; it’s about who can light up the most gigawatts.