TheChek News Flow

The Takeaway

Thinking Machines Lab, the AI developer led by ex-OpenAI Chief Technology Officer Mira Murati, is in talks to raise between $5 billion and $6 billion at a valuation of at least $40 billion before the investment, existing investor Andreessen Horowitz has told other prospective investors.

Nvidia is expected to contribute about half of the amount, said the investment firm, which holds a board seat at Thinking Machines Lab. The move would extend a series of deals Nvidia has struck to expand its influence with open-source AI developers, including buying open-source model provider Hugging Face for $12.9 billion.

It isn’t clear whether the capital from Nvidia is part of the “significant” but undisclosed investment the chip designer said it would make in TML in March. Nvidia said it planned to make a “long-term, gigawatt-scale strategic partnership” with TML which likely refers to helping the startup get access to Nvidia’s coveted AI server chips.

Nvidia has also made large investments in open-weight model developers Reflection and Poolside. (It also agreed to pay Poolside $6 billion to access its technology and offer to hire many of its employees) Nvidia is trying to strengthen American-made open-source AI offerings as a counterweight to closed-source AI model developers such as Anthropic and OpenAI, which also depend on Nvidia hardware but are racing to develop their own chips as alternatives.

Existing investor Accel is in talks to lead the Thinking Machines round, implying the firm determined the share price, according to Andreessen’s outreach to investors and other people with knowledge of the round. The round is still open and the startup is seeking more capital, and terms of the deal could still change.

Murati has an unusual level of control at the company, which could discourage some investors. As part of Thinking Machines’ earlier fundraise, Murati gained a board vote that holds the same weight as all the other board directors’ votes plus one. That provision ensures that she will control the startup’s board votes, no matter how many other directors vote against her, on major decisions like deciding whether to accept an acquisition offer, firing executives and approving executive compensation.

The proposed valuation would be lower than the valuation the company sought late last year, when it discussed raising between $4 billion and $5 billion from investors at a $50 billion-plus valuation. But it’s still a remarkable figure, given that Thinking Machines is less than two years old.

Spokespeople for Thinking Machines, Andreessen and Nvidia did not immediately have a comment.

Thanks largely to the prominence of Murati, Thinking Machines Lab is one of the most high-profile startups known as neolabs that are taking on major AI incumbents. It previously raised billions of dollars to bankroll the costly process of renting Nvidia chips, including for the development of Inkling, an open-weight model it launched in July.

The startup competes with other open-weight model providers, in terms of trying to gain share among AI app developers that use such models, which are free to download. Rather than trying to compete directly with closed-source developers OpenAI and Anthropic, whose models are more capable than open-weight ones, TML essentially competes with cloud providers and so-called inference computing providers to help businesses customize AI models with their data. It has also released AI geared toward voice applications.

Thinking Machines said in July that its mission is to build AI that “extends human will and judgment.”

The startup makes money by selling access to tools companies can use to customize AI models—including but not limited to Inkling—using their own data.

The startup generates at least a few hundred million dollars in annualized revenue, according to a person with knowledge of the figure, which typically refers to the prior month’s revenue multiplied by 12. That would make its $40 billion valuation very high as a multiple of forward revenue, compared to most other startups in the field.

The new valuation would be a big leap from its initial fundraise in July 2025, when it raised $2 billion at a $10 billion pre-investment valuation, led by Andreessen Horowitz with participation from Nvidia, GV, Lightspeed Venture Partners, Jane Street and Factorial Funds.

Thinking Machines, like its rivals, has also experienced turnover in its top staff. Earlier this summer, co-founder Lilian Weng said she was rejoining OpenAI. Her exit marked the fourth Thinking Machines cofounder to leave the company in the last year. Murati and John Schulman, who previously co-founded OpenAI, are the two remaining cofounders at Thinking Machines.