Nvidia Partnership Hints at Big Ambitions for Former OpenAI Tech Chief’s Startup
The startup struck a “multiyear strategic partnership” with Nvidia, which also made a “significant investment,” the two announced Tuesday.

Sign up for smart news, insights, and analysis on the biggest financial stories of the day.
Thinking Machines Lab, the artificial intelligence startup led by OpenAI’s former chief technology officer Mira Murati, has just traded its ‘stealth’ status for a seat at the big table.
The startup struck a multiyear strategic partnership with Nvidia, which also made a “significant investment,” the two announced Tuesday. The terms, however vague, signal ambitions to shake up frontier AI.
You Said Watt?
Founded as a public benefit corporation in February 2025 by OpenAI veterans, Thinking Machines generated considerable buzz (and jealousy, at least from this wallet) when it raised $2 billion at a $12 billion valuation just five months later. Joining the seed round were A-list investors, including Andreessen Horowitz, Nvidia, AMD and Cisco, for whom most pitch decks serve as recycling material. The hype was on.
Thinking Machines emerged from stealth mode in October with its first and so far only product, Tinker, a tool that automates much of the work involved in fine-tuning large language models. Tinker works with Meta’s Llama and Alibaba’s Qwen, both of which are open source, allowing users to train them to perform certain specialized tasks: making their own custom versions of a frontier model, in short. Beyond that, however, Thinking Machines has kept mum about other plans. Tuesday’s deal suggests, at the very least, they aren’t small:
- The financial details of Nvidia’s “significant investment” were not disclosed, but the companies said Thinking Machines will deploy at least 1 gigawatt of the semiconductor giant’s brand new Vera Rubin hardware beginning next year.
- One gigawatt is enough to power the entire city of San Francisco, about the amount a nuclear power plant generates. Nvidia CEO Jensen Huang said in August 2025 that building a gigawatt of data-center capacity costs $50 billion to $60 billion. In other words, the resources being deployed here all but scream that Thinking Machines is planning to compete in the frontier model space against the likes of Anthropic, Google, OpenAI, Meta and Alibaba.
AI’ll See You Later: Thinking Machines hasn’t entirely avoided the kind of C-Suite drama its founder’s former company endured. (Recall Sam Altman’s ousting and nearly immediate reinstatement as OpenAI CEO in 2023.) The executive ranks at Thinking Machines have been in flux, with co-founder Andrew Tulloch leaving for Meta last year, and co-founders Barret Zopf and Luke Metz re-joining OpenAI in January.










