SAN FRANCISCO, March 10, 2026, 07:13 PDT
- Nvidia said it made a significant investment in Thinking Machines Lab and will supply at least one gigawatt of Vera Rubin systems under a multiyear pact. 1
- Deployment is targeted for early next year on Nvidia’s next-generation Vera Rubin platform. 2
- Nvidia shares were little changed in morning U.S. trading on Tuesday.
Nvidia said on Tuesday it made a significant investment in Thinking Machines Lab and signed a multiyear agreement to supply at least one gigawatt of its next-generation Vera Rubin systems to the startup founded by former OpenAI executive Mira Murati. Financial terms were not disclosed. 1
The move gives Nvidia another flagship buyer for Vera Rubin, its next AI chip platform, just before the company’s March 16-19 GTC developer conference in San Jose. It also extends Nvidia’s strategy of backing startups that later buy its hardware, a model investors have begun to scrutinize more closely as AI spending keeps rising. 3
Thinking Machines said deployment is targeted for early next year. Reuters reported industry executives say one gigawatt of compute — a rough measure of data-center power — can cost around $50 billion and matches about the electricity use of 750,000 U.S. homes. 2
In a company post, Chief Executive Jensen Huang said Thinking Machines had “brought together a world-class team.” Murati said Nvidia’s technology was “the foundation” of the field and that the partnership would speed the startup’s build-out. 2
The startup, launched last year, raised about $2 billion in seed funding at a $12 billion valuation, a Reuters report in July said. Reuters also reported on Tuesday that several senior employees, including co-founder Barret Zoph, had recently returned to OpenAI. 4
The competitive picture is moving around Nvidia at the same time. AMD agreed last month to sell Meta up to $60 billion of AI chips over five years, while Broadcom said its business building custom AI chips for specific customers could top $100 billion next year. 5
Even so, Nvidia remains the stock the market reads first. Its shares were little changed in morning U.S. trade on Tuesday, and Jacob Bourne, an analyst at eMarketer, said after last month’s earnings that the “competitive picture is also shifting” as Meta diversifies toward AMD and big cloud groups build more of their own chips. 6
That is the risk for Nvidia. Its investments in OpenAI, Anthropic and now Thinking Machines can help lock in demand, but they also feed concerns about circular capital flows and whether the AI build-out will generate returns quickly enough. 7
For now, demand still looks hard to dent. Nvidia last month forecast fiscal first-quarter sales of $78 billion, above Wall Street estimates, and Huang said the new AI infrastructure cycle “is not going to go back.” The next readout comes at GTC next week. 8
