The Rise
When Mira Murati, formerly chief technology officer at OpenAI announced in early 2025 that she was founding a new company, the tech world pricked up its ears. Within months, Thinking Machines Lab raised a staggering US$2 billion in its seed round, backed by heavyweight investors like Andreessen Horowitz, Nvidia, AMD, Jane Street and Cisco, valuing the startup at roughly US$12 billion.
What makes this story more than another generative-AI play is the pedigree. Murati assembled a founding team that reads like the Avengers of AI: John Schulman, a co-founder of OpenAI; Barret Zoph, former VP of Research at OpenAI; and Lilian Weng, another OpenAI alum. Their shared history working on cutting-edge reinforcement learning, large-scale model training and safety research is now being leveraged into something far more ambitious.
Seeing a Global Moment in AI
The market opportunity for Thinking Machines couldn’t be more timely. As generative AI matures, the demand is shifting from chat and content to reasoning, planning and long-term problem solving. Businesses and governments are hunting for more intelligent, autonomous systems that can handle complex, multi-step tasks, not just spit out text. Investors are pouring capital into this shift.
Moreover, with regulation tightening around how LLMs are used, there is room for a public-benefit corporation like Thinking Machines to carve out a reputation for building safe, mission-driven AI. Its structure with Murati having a weighted deciding vote signals a long-term view.
What Sets Thinking Machines Apart
The core of Thinking Machines Lab’s value proposition is scientific-level AI reasoning combined with real-world action. Rather than merely training large language models on internet text, the company seeks to develop models that can plan, execute, experiment and self-improve essentially thinking systems that act on their insights. Because its founding team has deep experience in reinforcement learning and architecture design, they are well positioned to deliver on that vision.
This is not just about making better chatbots. It’s about building autonomous agents for science, robotics, enterprise workflows, and decision support. In a way, Thinking Machines Lab is aiming to be the “general-purpose mind” behind future industry-scale transformation.
Business Model & Revenue Potential
Thinking Machines Lab plans to monetize via a mix of enterprise licensing and research partnerships. It will offer its AI through APIs, customized agents, and collaborative research contracts. Given its lab-grade ambitions, it’s likely to partner with corporations in sectors like biotech, energy, aerospace, and robotics industries that can pay a premium for deeply capable AI agents.
Because its founding investors include hardware players like Nvidia and AMD, there is a tight alignment to deliver value by optimizing both compute and software. The dual revenue streams (software + research) give it flexibility and a higher ceiling as scaling begins.
Early Traction & Scalability
Though fresh, the startup is not starting from zero. By July 2025, it had already hired about 30 researchers and engineers many from OpenAI, Meta AI and Mistral AI. Its structure as a public-benefit corporation gives it credibility, attracting talent who want to work on purpose-driven high-risk, high-reward AI research.
Given its capital base, the company can scale aggressively: build more internal labs, fund long-duration experiments, and deliver commercial-grade solutions internationally. Its model is scalable across geographies and industry verticals because reasoning-based AI can be applied in so many domains.
Standing Among Rivals
In a crowded AI space, Thinking Machines Lab competes with both entrenched players (OpenAI, Anthropic) and emerging labs. But its difference lies in its deep reasoning-first approach, not just producing text or images, but executing. It may also appeal more to mission-driven organisations and governments because of its public-benefit structure.
On the other hand, incumbents are already building agentic systems, and new labs are constantly popping up. The challenge will be delivering usable, generalizable agents that justify the lofty valuation.
Exit Scenarios & Investor Pathways
Possible exits for Thinking Machines Lab might include a future IPO, given its size and capital intensity. Alternatively, strategic acquisition by a major tech or engineering firm (say, Nvidia, Microsoft or a robotics giant) is plausible, especially if its agents become central to core B2B workflows. Because of its public-benefit charter, a full acquisition would likely be structured carefully to preserve the mission.
Vision That Cuts Across Borders
Murati’s vision for Thinking Machines goes beyond profit. She sees a future where AI doesn’t just mimic intelligence, but augments scientific creativity, elevates human reasoning, and accelerates breakthroughs that no single lab could achieve. The weighted voting structure in the company also suggests she’s building something built to last, not a quick flip.
Financial Discipline & Governance
From the start, Thinking Machines Lab has struck a balance. Raising a huge $2 billion gave it fuel, but its public-benefit incorporation sends a signal: profits matter, but purpose matters too. The governance structure ensures that founding researchers retain control, reducing short-term pressure from investors. This governance model may be rare among high-valuation startups but it’s crucial for long-term scientific AI bets.
Looking Ahead: A New Era of Intelligence
As Thinking Machines Lab takes shape, its potential impact is enormous. If it succeeds, it could redefine how organisations think about AI: not as a tool, but as a thinking partner. It might power scientific labs that design new materials, or autonomy agents that plan entire projects, or decision systems that reason ethically.
At a time when AI is maturing, Thinking Machines Lab is offering something more than scale. And if their journey continues, they might well become the backbone of a smarter, more capable generation of AI.
