AI these days isn’t really held back by ambition anymore — it’s held back by architecture. Today’s AI models are bigger and more powerful than ever, but they’re also more secretive and mostly controlled by just a handful of major labs. For developers, businesses, and researchers, the real challenge isn’t getting access to powerful AI anymore — it’s actually being able to understand, control, and work with it. Thinking Machines Lab was created specifically to tackle that problem.
A Founder Who’s Seen the Industry From the Inside
In February 2025, Mira Murati — the former Chief Technology Officer of OpenAI and one of the key people behind some of the biggest generative AI systems in the world — founded Thinking Machines Lab. Having spent years working at the cutting edge of AI research, Murati understood firsthand both what these systems could do and where they were falling short. Big AI labs are great at scaling up, but they often lack transparency, personalisation, and real, meaningful interaction with the people using them.
When Murati started Thinking Machines Lab, the goal wasn’t to build a bigger model than everyone else. It was about rethinking how AI is actually designed, built, and understood in the first place. The company is based in San Francisco and has brought together top researchers and engineers who previously worked at places like OpenAI, Meta, and Mistral — a clear sign that it’s not just following the AI industry; it’s aiming to lead it.
A Record-Breaking Funding Round
In July 2025, Thinking Machines Lab raised $2 billion in seed funding — one of the largest early-stage funding rounds any tech startup has ever raised. The round was led by Andreessen Horowitz, with Nvidia, AMD, Accel, Cisco, and Jane Street also joining in, putting the company’s valuation at around $12 billion.
For investors, this wasn’t a bet on one specific product — it was a bet on an idea: that the future of AI depends on systems that are modular and easy to understand, rather than massive, all-in-one black boxes. As businesses increasingly need AI that’s explainable, regulation-ready, and tailored to specific industries, Thinking Machines positioned itself right at that gap in the market.
Moving Past the “One Giant Model” Approach
Unlike traditional AI labs that build massive, all-purpose models, Thinking Machines Lab focuses on making AI more customizable and modular. Its early work is centred around helping developers fine-tune and optimise powerful AI models without needing deep technical infrastructure knowledge — something that’s historically kept AI adoption limited to only the most advanced tech teams.
One of its early projects, internally called “Tinker,” is designed to make adapting AI models much simpler for developers. The bigger vision here is to turn AI from something static and closed-off into something that actually collaborates with people, rather than acting like an unpredictable black box.
This fits into a broader shift happening across the AI industry — moving away from one-size-fits-all models, toward systems that can actually reason within specific industries, workflows, and contexts.
Talent, Turbulence, and the Reality of Building Frontier AI
With that much capital and ambition also came real pressure. In early 2026, Thinking Machines Lab went through some significant leadership and talent turnover. Co-founder and CTO Barret Zoph was let go following internal disagreements, and several early researchers — including some well-known AI scientists — left shortly after.
This kind of turbulence isn’t unusual for cutting-edge AI startups, especially ones trying to balance pure research with actually building a real, scalable company. In an industry where top AI talent is scarce, and competition is fierce, innovation often moves faster than internal stability can keep up with. For Thinking Machines, these departures highlighted just how hard it is to go from being a research-first vision to running an operationally solid company.
Even so, industry observers say the company’s core direction and mission haven’t really changed — Murati is still leading, and the vision remains intact.
A Different Kind of Competitive Edge
Thinking Machines Lab isn’t really trying to win by being the fastest to market. Instead, it’s focused on building AI that can genuinely work alongside regulation, corporate accountability, and human judgment — not just raw scale. Its systems are designed to be understandable, auditable, and adaptable to real-world use.
FAQs:
Q1: Who founded Thinking Machines Lab?
Mira Murati, former Chief Technology Officer of OpenAI, founded the company in February 2025 in San Francisco.
Q2: How much funding has Thinking Machines Lab raised?
The company raised $2 billion in seed funding in July 2025, valuing it at around $12 billion.
Q3: How is Thinking Machines Lab different from other AI labs?
Instead of building one giant model, it focuses on modular, customizable AI systems that are easier to understand and adapt.
Q4: What is “Tinker,” Thinking Machines Lab’s early project?
Tinker helps developers fine-tune and adapt AI models without needing deep technical infrastructure knowledge.
