Mira Murati isn't doing things by halves.
Just five months after stepping away from OpenAI, where she served as chief technology officer, Murati closed a $2 billion seed round for Thinking Machines Lab on July 15, 2025. The deal valued the San Francisco startup at $12 billion—before it had shipped a commercial product, announced major customers, or demonstrated the kind of traction that typically justifies such stratospheric figures.
According to Crunchbase, the financing represents the largest seed round in venture capital history "by a long shot." It dwarfs Mistral AI's €105 million European seed from 2023, which itself raised eyebrows at the time. Perhaps more telling: the company set a $50 million minimum investment threshold for participation, a sign that capital wasn't the constraint. Access was.
Andreessen Horowitz led the round, with a syndicate that reads like a who's who of AI infrastructure and enterprise ambitions: Nvidia, AMD, Cisco, ServiceNow, Accel, and Jane Street all joined. The composition reveals competing interests at work. Nvidia and AMD want favorable GPU allocation terms, one suspects. Cisco and ServiceNow are eyeing enterprise distribution channels. And the venture firms? They're betting on Murati herself—her network, her vision, and her years inside the machine that produced ChatGPT.
Elite Founders, Elite Capital
The Thinking Machines round reflects a narrowing funnel in AI venture capital, where mega-deals for marquee founders coexist uneasily with a drought in early-stage funding elsewhere. In the first quarter of 2025, AI deals commanded 71% of total VC value, according to PitchBook data. Yet first-time financings outside the AI superstars remained scarce.
Murati's round came in higher than initial reports suggested. Coverage from June 2025 pegged the valuation at $10 billion; the final close added another $2 billion in perceived value. Whether that premium reflects genuine momentum or frothy competition among investors remains unclear.
She's hardly alone in attracting outsize capital. Safe Superintelligence, the startup formed by former OpenAI chief scientist Ilya Sutskever, reportedly raised $2 billion at a $32 billion valuation in April 2025. These numbers reflect investor conviction—or perhaps anxiety—that the next decade's most valuable companies will emerge from this narrow cohort of AI researchers.
What Thinking Machines Is Actually Building
Details remain sparse, though Murati has offered glimpses. The company is developing multimodal AI systems focused on human-AI collaboration. She's committed publicly to releasing an open-source component as part of the first product—a nod, maybe, to the transparency debates that have roiled the AI industry.
In June, WIRED reported that Thinking Machines launched "Tinker," a fine-tuning API for open-source large language models including Meta's Llama and Alibaba's Qwen. The tool targets developers seeking to customize foundation models without building infrastructure from scratch. Whether Tinker represents the company's primary product or an early experiment isn't entirely clear.
In posts on X following the funding announcement, Murati described a mission to create AI systems "that work with how you naturally interact." She promised forthcoming research and a commitment to sharing "best science" in the months ahead. The positioning aims squarely at OpenAI, Anthropic, xAI, Google DeepMind, and Mistral—a competitive set that includes her former employer and the world's best-funded AI labs.
A Bench of Familiar Faces—and Early Exits
Murati assembled a founding team stacked with OpenAI alumni: John Schulman as chief scientist, along with Barret Zoph, Lilian Weng, Andrew Tulloch, and Luke Metz. The Financial Times reported an unusual governance wrinkle—Murati retained dominant voting rights on the board, a structure that gives her near-absolute control despite outside investors' massive capital commitments.
But the team has experienced unexpected turbulence.
In November 2025, Tulloch departed for Meta. Then in January 2026, Zoph and Metz returned to OpenAI, bringing Sam Schoenholz with them. The departures raised questions about internal dynamics at Thinking Machines. Talent retention matters especially when the primary asset is the people, not the product.
Murati responded by appointing Soumith Chintala, a well-regarded AI researcher, as the company's new CTO. Still, losing three senior engineers within six months doesn't project stability.
Already Eyeing the Next Round
By November 2025—just four months after closing the $2 billion seed—reports surfaced that Thinking Machines was in preliminary talks to raise additional capital at roughly a $50 billion valuation. No deal has been announced, and valuations at that stage tend to be aspirational. But the fact that such conversations are even happening suggests either extraordinary progress behind the scenes or a willingness among investors to keep bidding up access to elite AI talent.
The company hasn't disclosed commercial customer relationships, though the presence of ServiceNow and Cisco in the cap table hints at enterprise partnerships under discussion. Whether those translate into revenue—and at what scale—remains the open question.
The Bigger Picture
Thinking Machines Lab arrives at a peculiar moment in the AI industry. Foundation models have become table stakes. Differentiation increasingly comes from fine-tuning, deployment infrastructure, and user experience rather than raw model performance. The open-source movement, led by Meta's Llama releases, has democratized access to capable models, raising the bar for what proprietary offerings must deliver.
Murati's bet seems to be that collaboration-focused, multimodal systems—with an open-source ethos—can carve out differentiated space in an increasingly crowded market. But $12 billion is a lot to prove.
For now, investors are paying for potential, pedigree, and proximity to the AI frontier. Whether Thinking Machines can translate that into products customers actually want? That story is still being written.
---
This article has been updated to reflect the company's latest funding discussions and leadership changes through January 2026.
