Pomegra
Mira Murati's Thinking Machines Lab is reportedly raising $5 billion led by Accel at a $40 billion pre-money valuation, a 20% cut from the $50 billion once sought.
Key Takeaways
- Business Insider reported on November 6, 2026 that Accel leads a $5B round at a $40B pre-money valuation.
- The valuation implies about $45B post-money, versus $12B at the July 2025 seed.
- The $40B figure sits 20% below the $50B level the company pursued in late 2025.
Lead
Thinking Machines Lab, the San Francisco AI startup founded by former OpenAI chief technology officer Mira Murati, is closing a $5 billion funding round led by Accel, Business Insider reported on November 6, 2026. The round values the company at $40 billion before the new money and roughly $45 billion after it, assuming the full amount is raised. The company has not confirmed the deal, and the terms could still change before it closes.
The reported figure would more than triple the $12 billion valuation set in July 2025, when the startup raised a $2 billion seed round before shipping a product. It would also be one of the largest private rounds for an AI lab that is not yet a decade, or even two years, old.
What Are the Reported Terms of the Round?
The reported terms are $5 billion in new capital, Accel as lead, and a $40 billion pre-money valuation. A full raise implies a post-money valuation of about $45 billion. Business Insider's report did not name other participants in the round.
Earlier reports in September 2026 said Nvidia was in talks to contribute between $2.5 billion and $3 billion toward a round of $5 billion to $6 billion. Nvidia has not been confirmed as a participant in the Accel-led deal. Accel already holds a position, having joined the seed round alongside Andreessen Horowitz, Nvidia, AMD, ServiceNow, Cisco and Jane Street.
Why Is the Valuation Lower Than the $50 Billion Target?
The valuation is lower because the earlier $50 billion talks, reported by Bloomberg in late 2025, did not produce a deal. The $40 billion figure is 20% below that ask, though it still represents growth of more than 230% from the seed mark.
That gap matters for how the round reads. A raise that lands below the number a company first floated suggests investors priced the business against current traction rather than the founder's reputation. Thinking Machines has been generating roughly $100 million in annual revenue as of September 2026, according to reporting at the time. At $40 billion pre-money, that implies a multiple of about 400 times annual revenue, a ratio that only holds if revenue compounds at a pace few companies have sustained.
The company also lost talent in the interval. Three co-founders left between October 2025 and January 2026, with destinations including Meta and OpenAI, according to reports. Investors are paying for the remaining team's research record and the model roadmap, not for a stable bench of founders.
What Does Thinking Machines Actually Sell?
Thinking Machines sells Tinker, a cloud service for training and customizing models, and it releases its own model family called Inkling. The first Inkling model launched on July 15. The second-generation model, Inkling 2 Small, arrived on November 5, one day before the funding report.
Inkling 2 Small has 594 billion total parameters, of which about 17 billion activate for each prompt. Its vision encoder accounts for 292 million parameters. Pre-training ran on GB300 NVL72 racks, each holding 72 Blackwell Ultra GPUs, and finished in 68 hours. A nine-day reinforcement learning phase followed.
The company says the model beat competitors on knowledge-work and financial-task benchmarks and doubled the forecasting accuracy of Qwen3.5-4B in 4.5 hours of automated research. Those are company-reported results and have not been independently reproduced. Open-source weights are due the following week.
How Does the Nvidia Relationship Fit In?
Nvidia is both a past investor and a hardware supplier, which links the round to the company's compute bill. In March 2026, the two companies announced a multi-year partnership covering deployment of Nvidia's Vera Rubin systems. A large share of any new capital is likely to return to compute vendors, a pattern common across frontier labs.
That circularity is the main structural question around the round. When a chip supplier invests in a customer that then spends on its chips, headline valuations partly reflect the supplier's own demand. The Accel lead adds an investor with no such hardware interest to the cap table, which gives the price a less conflicted anchor.
Market Context
Frontier-lab financing has moved into multibillion-dollar territory as training runs have grown. Rounds at this scale favor labs with an established model line and a paying customer base, and Thinking Machines now has both, though at a smaller scale than the largest rivals. The $5 billion figure is more than double the seed round that launched the company.
The shift from a $50 billion ask to a $40 billion deal also shows that late-stage investors in AI are negotiating on price. Rounds are still large, but terms are not automatically set by the founder.
Outlook
The round is unconfirmed until Thinking Machines or its investors announce it, and the final size, valuation and participant list may differ from the reported figures. The open-weights release of Inkling 2 Small next week will be the first public test of whether the model draws developers to Tinker. If the deal closes as reported, the company will hold $5 billion in fresh capital against roughly $100 million in annual revenue, and its next milestones will be revenue growth and retention of its remaining research leadership.
