In brief
- OpenAI CEO Sam Altman calls compute power the “fuel” driving AI models, revenue, and growth.
- 10 gigawatts of AI compute equals the power use of millions of homes, sparking energy concerns.
- Experts warn AI’s electricity and water demands could strain grids worldwide.
Nvidia is betting big on the future of artificial intelligence, committing up to $100 billion to build out OpenAI’s computing muscle in what could be the largest infrastructure deal in AI history to date.
The chipmaker said Monday it had signed a letter of intent with OpenAI to deploy at least 10 gigawatts of Nvidia-powered AI systems. A gigawatt is equal to one billion watts of power—enough electricity to supply millions of homes.
According to a report by Bloomberg, the money will be provided in stages, with the first $10 billion coming when the deal is signed, citing people familiar with the matter. Nvidia will receive OpenAI equity as part of the deal.
Nvidia’s stock (NVDA) rose 4% from $177.50 to $184.16 on news of the partnership.
In an interview with CNBC, Nvidia CEO Jensen Huang called the deal the next leap forward for artificial intelligence.
“The computing demand is going through the roof,” Jensen said. “This partnership is about building an AI infrastructure that enables AI to go from the labs into the world. This is about the AI industrial revolution arriving.”
The plan calls for vast new data centers filled with millions of GPUs, including its upcoming Vera Rubin platform, designed for training and running next-generation models. The first gigawatt of capacity is expected to go online in the second half of 2026, with the rest rolling out as Nvidia ramps its investment.
Joining Huang were OpenAI CEO Sam Altman and OpenAI President Greg Brockman. Altman said the deal with Nvidia will expand the ambitions set by the Stargate project announced by President Donald Trump in January.
“This is the fuel we need to drive improvement—to build better models, generate revenue, everything,” Altman said. “This is helping us, along with our partners at Stargate, Microsoft, and Oracle, to build increasing amounts of infrastructure to deliver on what the world is demanding.”
Deploying 10 gigawatts of computing capacity won’t be simple, however. Cooling alone can consume nearly 40% of a data center’s power use, according to energy consultancy 174 Power Global.
Environmental concerns are also mounting. Deloitte estimates data centers will account for about 2% of global electricity use in 2025, or 536 terawatt-hours, but says demand from power-hungry AI could push that figure to more than 1,000 terawatt-hours by 2030.
The United Nations Environment Programme has warned of rising water consumption for cooling, while the Environmental and Energy Study Institute said data centers are already straining electric grids.
Generally Intelligent Newsletter
A weekly AI journey narrated by Gen, a generative AI model.
