Thinking Machines Lab closes $2B seed round at $12B valuation
On July 15, Thinking Machines Lab, the AI research startup founded by former OpenAI chief technology officer Mira Murati, closed a $2 billion seed round at a post-money valuation of $12 billion, according to filings reported by Reuters and TechCrunch. The round was led by Andreessen Horowitz with participation from Nvidia, AMD, Accel, ServiceNow, Cisco, and Jane Street.
The company was incorporated in February 2025 and remained largely out of public view for its first six months. Murati had served as OpenAI's CTO since 2018 and led the company's product organisation through the launches of GPT-4 and ChatGPT before departing in late 2024. Several core researchers from OpenAI, Anthropic, and Meta joined the lab in its early months, according to TechCrunch reporting.
In a statement on X on the day of the announcement, Murati wrote: "We're excited that in the next couple months we will be able to share our first product, which will include a significant open source component and be useful for researchers and startups developing custom models." She added in a separate post that the lab also intends to "share our best science to help the research community better understand frontier AI systems."
The deal stands out for its size relative to its stage: $2 billion at the seed level exceeds, by some published tallies, the entire 2024 seed-stage AI venture market, and the $12 billion post-money valuation is roughly 600x a typical software seed valuation. Reuters noted that the round structure resembles a Series B in everything but name, with most of the capital expected to flow into compute commitments rather than headcount.
Reaction in the industry was split. Investors interviewed by Fortune described the round as a positive signal for first-time female founders, given that Murati becomes one of only a handful of women to lead a $10B+ AI venture. Others — including more cautious infrastructure analysts — raised concerns that valuations for pre-product AI labs are decoupling from underlying revenue or technical demos. The lab had not shipped a public product as of the announcement.
For us at Enpo Sekai, the round reads less as a single fundraising story and more as a reinforcement of a structural shift: in 2025, the cost of competing at the foundation-model layer has crossed a threshold that effectively excludes everyone outside a handful of capital pools. Companies our size have to make a clear choice — and for us, that choice is to build above the model layer, in characters, voice, and persona, rather than below it.
We will be watching three things over the next twelve months: (1) whether the promised open-source component actually ships in a form that small studios can use; (2) what kind of compute relationship Thinking Machines builds with Nvidia, since the seed-stage Nvidia investment hints at a long-running supply arrangement; (3) how quickly former OpenAI alumni labs converge or diverge on product positioning. None of these change our roadmap, but all of them inform the cost structure of the field we operate in.
