Thinking Machines Lab, founded by former OpenAI Chief Technology Officer Mira Murati in early 2025, closed a $2 billion seed round in April 2025 at a $12 billion valuation led by Andreessen Horowitz, per Reuters and Bloomberg reporting — the largest seed round ever recorded, raised by a company that had not yet shipped a product. The round, which reportedly included Valor Equity Partners, Xavier Niel's NJP Capital, and a long list of AI researchers as angels, priced a team, not a business. Honey Badgers covers deals as information, not investment advice.
Who is on the team investors paid for?
Murati spent nearly seven years at OpenAI and served as CTO through the ChatGPT launch, making her one of the few executives with direct operating history at frontier scale. She recruited heavily from her former employer: early joiners reported in press coverage included Barret Zoph and John Schulman, researchers with deep OpenAI pedigrees, alongside alumni of Meta's AI groups and Character.AI. In a market where top researchers command seven-figure packages, the $12 billion valuation reads as a bet that the team could rebuild a frontier lab in 18 months — and the hiring velocity, more than any strategy document, is what the check paid for.
What is the company actually building?
On the record, modestly: Thinking Machines has said it works on making advanced AI systems more widely usable and customizable, with an emphasis on open research and safety. By late 2025 the company remained pre-product publicly, and press reporting described research on fine-tuning and post-training tools rather than a consumer launch. That gap — $14 billion of combined capital and valuation against zero announced revenue — is the entire risk profile, and it is not hidden; it is the terms of the bet.
How does the round compare historically?
The seed record before 2025 stood at a few hundred million dollars at most. Thinking Machines' $2 billion primary at a $12 billion valuation reset the category, and Ilya Sutskever's Safe Superintelligence followed the same template the same month — $2 billion at $32 billion with no product and an explicit no-product-near-term stance. Two rounds, $4 billion of primary capital, zero announced customers. The pattern tells you the seed label has detached from its original meaning: these are founder-track bets priced like growth equity, with the diligence concentrated entirely on personnel.
What does Murati's record support so far?
Reported, not retold: she helped lead the teams behind GPT-4 and ChatGPT at OpenAI, per her own account and extensive press record, and she briefly served as OpenAI's interim leader during the November 2023 board crisis — the shortest and most turbulent CEO tenure in the industry's history, ended by Sam Altman's return within days. What the record does not yet show: company-building at the top job, revenue generation, or product shipping under her own flag. The $12 billion prices the first list and discounts the second.
What happens next on the documented timeline?
The observable milestones are hiring headcount (reported around 60 by late 2025), compute deals, first research releases, and eventually either a product or a revenue-generating API. The comparables cut both ways: Safe Superintelligence maintained a pure research stance through 2025 without revenue pressure; Anthropic and OpenAI converted similar talent concentrations into revenue machines within two years each. The seed round's size guarantees scrutiny — at $12 billion, Thinking Machines must grow into a valuation that most Series C companies never reach, and the market will price each milestone against that bar.
The deal's real lesson is about the market, not the founder: in 2025, a documented hand on a frontier training run became a standalone asset class. Murati is the test of whether that asset converts to a company.
For more context, read The Repeat-Founder Advantage: What the Funding Data Supports.
For more context, read angel investors vs venture capital.
For more context, read When Should a Founder-CEO Hire a Replacement?.