Thinking Machines, the AI lab founded by former OpenAI CTO Mira Murati, is reportedly in discussions to raise $1 billion at a valuation of at least $40 billion. Existing investor Accel is in talks to lead the financing, according to TechCrunch, citing its own source and reporting from The Information.
Neither Accel nor Thinking Machines had responded to TechCrunch’s request for comment, so the terms and completion of the round remain uncertain. But the reported deal would offer a revealing benchmark for the economics investors are willing to underwrite in the next generation of foundation-model companies.
A sharp step up from the last round
Thinking Machines previously raised $2 billion in a round valuing it at $12 billion. Andreessen Horowitz led that financing, with Nvidia, GV, Lightspeed and Conviction Partners also participating.
A $40 billion valuation would therefore represent a more than threefold jump from that prior mark. It would still be below the $50 billion valuation the company reportedly sought late last year.
The company’s reported annual revenue run rate is now above $100 million. At the proposed valuation, that implies an exceptionally high multiple of reported revenue—particularly for a company whose commercial offering is still young. For investors, the underwriting case is unlikely to be based on current revenue alone. It rests on the potential for a research lab with marquee talent to develop a durable platform, attract developers and enterprises, and capture a meaningful portion of model-related compute and software spending.
The product question is becoming central
In July, Thinking Machines introduced Inkling, an open-weight model. The company monetizes through Tinker, a platform that charges usage-based compute fees for adapting models on proprietary data.
That approach matters because it gives the company a clearer commercial mechanism than a pure research-lab narrative: customers can use proprietary data to customize models while the provider participates in the resulting compute usage. For operators, the important question is whether that workflow delivers enough performance, control and usability to become part of recurring production AI spend.
Open-weight distribution can also broaden adoption, but it does not automatically create a defensible business. Thinking Machines will need to show that Tinker—or associated tooling, support and infrastructure—remains valuable as customers gain more options to run, adapt and deploy models elsewhere.
Talent pedigree has limits
Thinking Machines’ original financing reflected confidence in Murati and a group of former OpenAI researchers. Since then, several prominent people have departed, including co-founders Lilian Weng and Luke Metz, according to TechCrunch. The company’s ability to recruit and retain senior researchers and product leaders will remain material to both its technical roadmap and investor confidence.
For founders and executives buying AI platforms, the episode is a useful reminder that vendor evaluation should not stop at model benchmarks or a founder’s résumé. Assess the product roadmap, team depth beyond a few recognizable names, deployment model, data handling, portability, pricing exposure and the likelihood that a product remains supported through rapid organizational change.
What to watch next
If a round is finalized, the key details will be the actual valuation, investor mix and whether the capital is earmarked chiefly for compute, research hiring or commercial infrastructure. More important than the headline will be evidence that the company can convert model interest into durable, usage-led revenue.
The reported financing would also signal that top AI labs can still command enormous private valuations despite demanding expectations. The next test is whether Thinking Machines can make its platform economics—and not just its research pedigree—the basis for that valuation.
