OpenAI and NVIDIA Strike $100 Billion Deal to Build AI Supercomputing Power
A landmark $100 billion investment will see NVIDIA deploy 10 gigawatts of systems for OpenAI, reshaping the race for AI infrastructure.
San Francisco, September 22 EST: NVIDIA is putting its money and its chips squarely behind OpenAI. The two companies on Monday announced a $100 billion partnership to build out computing power on a scale closer to a national grid than a corporate IT project.
The deal calls for deploying 10 gigawatts of NVIDIA systems dedicated to OpenAI, a figure that energy analysts say rivals the output of several nuclear reactors. The first gigawatt is expected to go live in the second half of 2026, powered by NVIDIA’s Vera Rubin platform, with the rest to follow in staged increments. NVIDIA will release its investment in phases, tied to each deployment milestone.
Why This Deal Matters
This isn’t just another tech partnership. At $100 billion, it is one of the largest single-company bets ever made on infrastructure in the private sector. For comparison, it’s about the cost of building 100 data centers the size of Google’s flagship facilities, or half of what the U.S. spends annually on roads and bridges.
The money will go into raw compute: millions of GPUs tasked with training OpenAI’s next generation of models. Both sides framed the deal less as a marketing splash than as a necessity. OpenAI needs scale to keep up with rivals; NVIDIA needs to lock in demand for its chips before competitors close the gap.
“This marks the next leap forward, deploying 10 gigawatts to power the next era of intelligence,” said NVIDIA CEO Jensen Huang, who has made similar bets on high-performance computing before but never at this order of magnitude.
OpenAI’s Sam Altman was more blunt: “Everything starts with compute. Compute infrastructure will be the basis for the economy of the future.” In other words, whoever controls the hardware controls the future business of AI.
Markets Took Notice
The ripple effect was immediate. Oracle shares rose nearly 5% in Monday trading, reflecting the company’s stake in the broader Stargate AI project, a $500 billion data center initiative that ties together OpenAI, Microsoft, and SoftBank. NVIDIA’s stock, already the most valuable semiconductor play on Wall Street, ticked higher as well, underlining investor confidence that it has secured a long-term moat around its GPU dominance.
For Oracle, the reaction wasn’t just hype. It was markets betting on the company’s central role as a landlord for these vast computing factories.
A Decade-Long Dance
The partnership has roots that go back nearly a decade. NVIDIA’s DGX supercomputers powered some of OpenAI’s earliest breakthroughs, including the training runs that made ChatGPT viable. What began as an arms-length supplier relationship has now hardened into something closer to joint infrastructure development.
In that sense, Monday’s deal looks less like a tech investment and more like a modern-day version of a steelmaker locking in rail contracts during the industrial revolution. If AI is to become the backbone of future economies, then NVIDIA and OpenAI are effectively laying down the rails.
The Bottleneck No One Can Ignore
The deal also underscores a less glamorous truth: AI’s biggest choke point isn’t talent or algorithms, it’s power, both electrical and computational. Ten gigawatts is an amount of energy that would challenge utilities, regulators, and local governments. It’s the kind of capacity usually debated in statehouses, not boardrooms.
“This is as much an energy play as it is a computing one,” noted analysts quoted by Reuters. Without reliable, cheap electricity, the entire project risks stalling.
That said, Altman and Huang appear to be betting that governments will be more willing to accommodate projects of this scale once the economic upside becomes clear.
The Competitive Backdrop
The timing is no accident. SoftBank is separately leading a $40 billion funding round for OpenAI that could push its valuation to $300 billion, and rivals from Google to Meta are racing to secure chips and grid access. Whoever solves the compute bottleneck first will be in position to dictate terms in the next wave of AI development.
For now, the OpenAI–NVIDIA agreement is still largely conceptual. Building ten gigawatts of capacity will stretch well into the late 2020s. Supply chains, permitting battles, and energy politics could all slow the rollout. But Wall Street is treating the announcement as a directional bet on where the future is heading: fewer flashy apps, more control over the pipes that power them.
In that light, Monday’s news isn’t just about a tech alliance. It’s about who owns the picks and shovels in the AI gold rush, and whether OpenAI and NVIDIA can keep the rest of Silicon Valley digging on their terms.
New Jersey Times Is Your Source: The Latest In Politics, Entertainment, Business, Breaking News, And Other News. Please Follow Us On Facebook, Instagram, And Twitter To Receive Instantaneous Updates. Also Do Checkout Our Telegram Channel @Njtdotcom For Latest Updates.
A Wall Street veteran turned investigative journalist, Marcus brings over two decades of financial insight into boardrooms, IPOs, corporate chess games, and economic undercurrents. Known for asking uncomfortable questions in comfortable suits.
