Team Startupcity

·  Thinking Machines Lab raises $2B in funding at a $12B valuation

·  Founded by ex-OpenAI exec Mira Murati; backed by Nvidia, Andreessen Horowitz, and others

·  First open-source product to launch soon, focused on safer, reliable AI systems

Thinking Machines Lab, a startup artificial intelligence company led by former OpenAI executive Mira Murati, has raised about $2 billion in a historic funding round headed by Andreessen Horowitz, valuing the company at a mind-boggling $12 billion.

Launched just in February 2025, Thinking Machines has yet to launch a product or generate revenue, but it has already attracted strong backing from industry giants including Nvidia, Accel, Cisco, AMD, ServiceNow, and Jane Street. The raise underscores Murati’s influence in the AI ecosystem and the investor appetite for visionary leadership in the booming deep tech space.

Also Read- GetWhy Raises $20 Million to Expand AI-Driven Consumer Insights Platform

We look forward to launching our first product later this year," said CEO Mira Murati, "which will feature a significant open-source element designed to enable researchers and startups developing bespoke AI models.

Thinking Machines seeks to develop more secure and trustworthy AI systems with wider real-world use than current models. Significantly, almost two-thirds of the startup's founding team are former OpenAI staff, making it one of several spinout startups from the ChatGPT creator's alumni community.

Murati’s venture joins other high-profile AI startups founded by OpenAI alumni, including Dario Amodei’s Anthropic and Ilya Sutskever’s Safe Superintelligence, both of which have also raised billions amid a hot funding climate for next-gen AI companies.

Reports state that AI now represents more than 64% of the value of U.S. startup deals, driving a 76% year-over-year increase in overall startup funding in the first half of 2025 to $162.8 billion. Investor interest in core AI businesses shows no signs of slowing, despite caution across the broader tech sector.