Thinking Machines in talks to raise $1B at $40B valuation, down from $50B target

Key facts

• Accel, an existing investor, is in talks to lead a new $1 billion round for Thinking Machines at a valuation of at least $40 billion, The Information reported Thursday.
• The prospective valuation would still fall short of the $50 billion target Thinking Machines reportedly sought late last year.
• Thinking Machines' annual revenue run rate is over $100 million, making the reported valuation an extraordinarily high revenue multiple.
• The lab's prior $2 billion seed round, among the largest ever, valued it at $12 billion and was led by Andreessen Horowitz with Nvidia, GV, Lightspeed and Conviction Partners.
• Co-founders Lilian Weng and Luke Metz are among several high-profile departures who have since returned to OpenAI.

Why it matters

Thinking Machines is built around Mira Murati's record as OpenAI's former chief technology officer, and it already raised one of the largest seed rounds in history before shipping a paying product. This report is the first concrete sign of how investors value the company since then. At least $40 billion is still below the $50 billion target Thinking Machines reportedly sought late last year, so the number under discussion has moved down, not up. Against an annual revenue run rate of just over $100 million, a $40 billion-plus valuation is, in the report's own words, an extraordinarily high revenue multiple: investors are pricing the company mainly on its team and technology, not on the revenue it has generated so far.

Who it affects

Thinking Machines' own investors, old and prospective, most directly: Accel already backs the company and would lead this round, while Andreessen Horowitz, Nvidia, GV, Lightspeed and Conviction Partners hold stakes from the prior $2 billion round with no confirmation on whether they are part of this one. It also affects Mira Murati, Thinking Machines and its remaining staff, who are building revenue through the Tinker platform while some co-founders, including Lilian Weng and Luke Metz, have already left for OpenAI. More broadly, the numbers become a reference point for anyone valuing other AI labs built on founder pedigree rather than revenue.

How to use it

There is no consumer product to buy here; the closest actionable detail is how Thinking Machines actually makes money. Its Tinker platform charges usage-based compute fees for adapting models on customers' own proprietary data, and in July the company introduced Inkling, an open-weight model built for that workflow. That business, not a subscription price list, is what produced the over $100 million revenue run rate cited in this report.

How solid is it

The story rests on The Information's reporting plus TechCrunch's own unnamed source; both the person describing the funding talks and the one describing Thinking Machines' financials are anonymous. Neither Accel nor Thinking Machines confirmed or denied the report; it says only that neither immediately responded to a request for comment. The deal itself is still described as talks and discussions, not a signed agreement, so the terms and even the final valuation could still change before anything closes.

Risks and caveats

Because the round is still described as talks and discussions, everything here could shift before it closes: $40 billion is a floor, not an exact figure, and no closing date is given. Whether Nvidia, GV, Lightspeed or Conviction Partners, all investors in the prior $2 billion round, are part of this one is not stated either way. The report describes the implied revenue multiple as extraordinarily high without calculating it or citing a specific figure, and it gives no reasons or dates for the co-founder departures it mentions.