The stage was set in San Francisco on Thursday for the Bloomberg Technology Summit, and the seat at the front of the room belonged to someone the AI industry had not heard from in a long time. Mira Murati, the former chief technology officer of OpenAI, sat down with Bloomberg’s Emily Chang for her first extended on-record interview since resigning in September 2024. TechCrunch described the appearance as stepping back into the spotlight, carefully, and the description matched the performance: she did not relitigate the OpenAI chapter, and she did not speculate about her former employer. She talked about what she is building now.
Murati spent eighteen months in near-silence after leaving OpenAI, an eternity in an industry where founders give interviews weekly and product launches are measured in weeks. She resigned in September 2024 amid a broader executive exodus that saw chief research officer Bob McGrew and research vice president Barret Zoph depart within days of her announcement, and she spent the following months building in private. In February 2025 she launched Thinking Machines Lab, a public benefit corporation with a mission statement about making AI systems more widely understood, customizable and generally capable.
The company has raised a reported $2 billion at a valuation around $12 billion, a figure that would make it one of the largest AI startups in the world before shipping a single major product. Its first offering, a tool called Tinker, arrived in October 2025 and lets developers create custom frontier models. In May 2026, the lab published a research preview of what it calls interaction models, a class of AI system designed for real-time interaction with the physical world, sensing and responding as events unfold, rather than the text-in, text-out pattern of conventional chatbots.
That architecture difference is the core of her argument. On Thursday she laid out the case that the next generation of AI will be judged not by how well it answers questions but by how well it operates in real time, and that the industry’s current models, built around conversation, are not structured for that shift. She positioned interaction models as a new category rather than an incremental improvement, and she said the company would build tooling and models around it rather than bolt it onto an existing chatbot. She did not name competitors, and she did not promise release dates.
The timing of her return matters as much as the content. In the current AI market, attention is a form of capital: funding follows founders who are visible, and talent follows momentum. Months of silence had left the field to a crowded set of rivals, and the appearance signals that Thinking Machines Lab is ready to be measured in public. Industry analysts said the return also serves a recruiting purpose, reminding researchers that a well-funded alternative to the established labs exists, and that it is led by one of the few executives in the industry with a track record of shipping products at scale.
The context is a strained moment for her former employer. OpenAI has been navigating internal turbulence and intensifying competition, and the alumni network it produced now runs some of the most valuable AI companies in the world, with former executives leading labs whose combined valuations run into the trillions. Murati’s own trajectory is part of that pattern: a senior executive with deep product experience, a public benefit structure, and a checkbook large enough to fund frontier research.
Her next move is being watched as a barometer of the AI talent market. When a former OpenAI CTO chooses between a quiet research lab and a public launch, or between staying independent and taking on partners, the industry reads it as a signal about where the value is moving. Thursday’s appearance suggests she is choosing the public path, and that the company’s product cycle is approaching.
Murati’s biography is part of the reason her return draws attention. A mechanical engineer educated at Dartmouth, she joined OpenAI after leading product engineering at Leap Motion and Tesla, where she oversaw the Model X program, and she rose to run the company’s research and product work through the launches of GPT-4 and ChatGPT. She served briefly as interim chief executive during the boardroom crisis of November 2023, a tenure that tested her under the most public pressure the industry has produced. That record, and the network it built, is now the core asset of Thinking Machines Lab.
For the market, the question is whether Thinking Machines Lab can convert its architecture thesis into shipped products that enterprises will pay for. The company’s valuation already prices in significant success, and the open-weight movement, led by rivals with more resources, is crowding the same territory. Murati’s answer, delivered from a San Francisco stage on Thursday, is that the interaction model is a new lane, and that the industry is about to find out who can build it first.
