Mira Murati, co-founder of OpenAI, has seen her two-year-old AI research venture ink a significant agreement with semiconductor giant Nvidia.
On Tuesday, Murati’s firm, Thinking Machines Lab, revealed a multi-year strategic partnership with Nvidia, a leading AI semiconductor manufacturer. While specific financial terms were not disclosed, the agreement commits the AI research lab to deploying at least one gigawatt of Nvidia’s Vera Rubin systems, which were introduced earlier this year, starting in 2027.
Additionally, Nvidia is making a strategic investment in Thinking Machines Lab. Since its establishment in February 2025, the lab has successfully raised over $2 billion from investors including Andreessen Horowitz, Accel, Nvidia, and even the venture capital arm of rival chipmaker AMD.
This early-stage company holds a valuation exceeding $12 billion and is focused on developing AI models capable of producing consistent results. In October, the company launched its initial product, an API named Tinker.
TechCrunch sought further details on the deal terms and investment from Thinking Machines Lab and Nvidia. Thinking Machines Lab opted not to provide additional comments beyond its official announcement.
According to an Nvidia press release, the collaboration also includes a commitment to create training and serving systems specifically for Nvidia architecture.
“Nvidia’s technology forms the bedrock upon which the entire field is constructed,” Murati stated in a blog post about the partnership. “This collaboration significantly enhances our capability to develop AI that individuals can customize and integrate into their own workflows, thereby further amplifying human potential.”
Despite its relatively short existence, Thinking Machines Lab has experienced several notable departures recently. The company’s co-founder, Andrew Tulloch, transitioned from the startup to a role at Meta in October. Earlier this year, three other co-founders—Barret Zoph, Luke Metz, and Sam Schoenholz—returned to OpenAI.
This agreement highlights the ongoing demand among AI companies for computational resources. Jensen Huang, CEO of Nvidia, has projected that by the decade’s end, companies might invest $3 trillion to $4 trillion in AI infrastructure.
While the precise value of this particular deal remains undisclosed, its magnitude is plausible. In 2025, it was reported that rival OpenAI had secured a landmark $300 billion cloud computing deal with Oracle.
Updated post-publication to correct that Thinking Machines launched its Tinker product last fall.