Mira Murati

Who they are

Mira Murati is Co-Founder and CEO of Thinking Machines Lab — former OpenAI CTO who served as interim CEO during the Sam Altman board crisis and launched Tinker, a fine-tuning API for open-weight language models, as her startup's first product.

Person

Mira Murati studied Mechanical Engineering at Dartmouth (graduating 2012), with earlier schooling at Pearson College UWC — an international baccalaureate school that draws students from across the world. From there her career moved through Goldman Sachs as an intern, then Zodiac Aerospace, then Tesla as a senior product manager, then Leap Motion — each stop pushing her closer to the intersection of hardware, software, and applied AI. She arrived at OpenAI early and climbed to CTO, becoming one of the most visible faces of the ChatGPT era: the Time Magazine interview in 2023 where she argued AI should be regulated, the Kara Swisher podcasts where she addressed the ScarJo voice controversy head-on, and the Johns Hopkins talk where she laid out her views on OpenAI's direction. She served briefly as interim CEO during the board's November 2023 ouster of Sam Altman — a few days that put her at the center of the most-watched governance drama in recent tech history. She left OpenAI in September 2024 and unveiled Thinking Machines Lab in February 2025, structured as a public benefit corporation. She also backed Worktrace AI, an AI startup by former OpenAI PM Angela Jiang. The through-line across her career: she gravitates toward moments where technology is about to change the shape of a product category, then moves to the center of it.

Company

Thinking Machines Lab's most recent major move is a multibillion-dollar deal signed with Google Cloud in April 2026 to expand its use of Google's AI Hypercomputer infrastructure powered by Nvidia GB300 systems — the clearest signal yet that the company is betting on massive compute access as its foundation. That followed a March 2026 strategic partnership with NVIDIA involving an undisclosed investment and a multi-year agreement to deploy at least one gigawatt of Vera Rubin computing capacity starting early 2027. The company raised a $2 billion seed round in July 2025 at a $12 billion valuation, led by Andreessen Horowitz with participation from Nvidia, AMD, Cisco, and Jane Street. Its first product, Tinker — a managed fine-tuning API for open-weight language models — launched in October 2025. The company has also shown a research preview of TML-Interaction-Small, a multimodal model processing audio, video, and text simultaneously with a 0.40-second turn-taking latency. The period has not been smooth: a $50 billion funding round was reportedly in talks in late 2025 but stalled by January 2026 amid unclear revenue paths, and by mid-2026 the company had lost multiple founding members to Meta and OpenAI — including Barret Zoph, Luke Metz, and at least five others — reportedly drawn away by high compensation packages after hitting their one-year cliff.

Market

Thinking Machines Lab competes directly with OpenAI, Anthropic, Google DeepMind, and Meta AI — the incumbent frontier labs — while differentiating on its public benefit corporation structure and emphasis on customizable, fine-tunable models rather than closed, monolithic ones. The AI sector in 2026 is defined by an extreme war for technical talent, massive compute capex, and a geopolitical layer involving US export controls on advanced chips that shapes which infrastructure partners companies can rely on. At 169–183 employees and $12 billion in valuation, Thinking Machines Lab is a high-profile but early-stage player in a field where its chief rivals have years of product momentum and entrenched distribution.

Network

Murati's core team is heavily drawn from OpenAI alumni: John Schulman is Chief Scientist, Lilian Weng is a co-founder and senior researcher, and Soumith Chintala — creator of PyTorch, who left Meta in late 2025 — joined as CTO in January 2026. She has also backed Angela Jiang, a former OpenAI PM, as an investor in Jiang's startup Worktrace AI. Ilya Sutskever, another former OpenAI colleague, rounds out the extended peer network.

John Schulman· Chief Scientist, Thinking Machines Lab
Soumith Chintala· CTO, Thinking Machines Lab (creator of PyTorch, ex-Meta)
Lilian Weng· Co-founder and senior researcher, Thinking Machines Lab
Angela Jiang· Founder, Worktrace AI (Murati investor)
Ilya Sutskever· Former colleague at OpenAI

How they likely show up

Moved from mechanical engineering → Tesla → Leap Motion → OpenAI → founder: each transition traded breadth for depth in AI product, suggesting she follows conviction about where a field is going rather than optimizing for title or safety.
Served as interim CEO during one of the highest-pressure governance moments in recent tech (OpenAI board crisis, November 2023) → comfortable operating in ambiguity at the very top; doesn't rattle easily under public scrutiny.
Structured Thinking Machines Lab as a public benefit corporation → the mission framing is not just marketing; it's a structural commitment she's willing to bake into the legal entity.
Multiple on-record media appearances (Time, WIRED, Kara Swisher's podcast, Johns Hopkins) where she engaged directly with hard questions on regulation and AI ethics → she runs toward the hard conversation rather than away from it.
Possibly — the hybrid role_type_pattern and medium tenure_shape suggest she moves on when the core problem is solved or the org has shifted away from where she can have the most impact, rather than staying for institutional reasons.
Invested in Angela Jiang's Worktrace AI while building her own company → angel activity alongside company-building signals high agency and a desire to stay connected to the broader field, not just her own lane.

Conversation tips

→ Reference the public benefit corporation structure specifically — she chose that deliberately, and asking what it concretely changes about decision-making will get a more interesting answer than asking about 'mission'.
→ She has spoken plainly on the record about AI regulation and the ScarJo controversy — don't treat these as sensitive topics; she engages directly and will respect a specific, well-formed question over a careful one.
→ Ask about Tinker and the fine-tuning thesis — it's the product bet that distinguishes Thinking Machines Lab from closed-model competitors, and it's where her strongest specific views likely live.
→ Don't bring up the November 2023 interim CEO period as drama — she's been matter-of-fact about it publicly; frame it as a governance and leadership question if you go there at all.
→ She came through hardware (Zodiac, Tesla, Leap Motion) before AI — references to embodied systems or the hardware-software interface will resonate more than pure software framings.

Toolbox

Openers

Open on the Google Cloud deal signed in April 2026 — pairing that with the NVIDIA gigawatt compute agreement suggests a specific infrastructure thesis is emerging at Thinking Machines Lab, and asking what that thesis is gets straight to the company's direction.
Lead with TML-Interaction-Small: a multimodal model with 0.40-second turn-taking latency is a very specific technical bet — it's a good way to open on the product roadmap without starting with a generic 'what are you building' question.
Reference her 2023 Time interview where she called for AI regulation — she staked out that position early, and asking how her views have evolved now that she's running her own lab (not a research org) is a genuinely live question.

Discovery questions

Tinker is focused on fine-tuning open-weight models — how does that product thesis hold up as the frontier labs keep closing their best models and pulling the ladder up behind them?
You structured Thinking Machines Lab as a public benefit corporation. What's a decision you've already made differently because of that structure — something a standard Delaware C-corp board would have pushed back on?
The NVIDIA and Google Cloud compute deals are significant for a 169-person company. How do you think about the risk of being deeply dependent on two infrastructure partners that are also investors or partners in your direct competitors?

Avoid

Don't treat the founding-team departures to Meta and OpenAI as a gotcha — she has publicly navigated hard organizational moments before and will engage on the substance, but leading with 'your people are leaving' as an opener signals you're looking for conflict, not insight.

Make it yours