Mira Murati, who used to be OpenAI’s Chief Technology Officer and is now its interim CEO, has said that her AI startup, Thinking Machines Lab, has gotten $2 billion in seed funding. Andreessen Horowitz (a16z) led the round, which also included well-known investors like NVIDIA, AMD, Accel, ServiceNow, CISCO, and Jane Street. The company is now worth $12 billion.
This is one of the biggest early-stage investments in AI history, and it shows how much faith investors have in Murati’s leadership and long-term vision.
AI Leadership is Growing Quickly
Mira Murati became well-known in late 2023 when she was named interim CEO of OpenAI. The company’s board suddenly fired CEO Sam Altman, which led to a short but widely publicized leadership crisis. Murati’s short time as CEO, even though Altman was quickly reinstated, showed how powerful and important she was.
Murati left OpenAI for good in September 2024. She started Thinking Machines Lab in February 2025, which was the start of a new chapter in her career as an AI leader. Since then, the startup has stayed out of the public eye. That changed in July 2025, when she announced on X (formerly Twitter) that the company had raised $2 billion and that its first product would come out in a few months.
What is the Thinking Machines Lab
Thinking Machines Lab is a company that does research and makes products in the field of artificial intelligence. Its goal is clear: to create AI systems that can work together, be customized, and use multiple modes of communication to meet the needs of people in many different fields.
Even though AI has come a long way, the company thinks there are still some important gaps:
- It’s still hard to customize or align most AI systems with your own or your organization’s goals.
- People don’t always understand how systems work and how they are trained.
- Most of the time, only top research labs have access to advanced AI.
Thinking Machines Lab wants to fill in these gaps by making systems that are:
- Easier to read and understand.
- More customizable and safer.
- Made to work together in the real world across many fields.
The people who started it are the same ones who made ChatGPT, Character.ai, PyTorch, OpenAI Gym, Fairseq, and Segment Anything, which are all tools and platforms that changed the game.
What the Thinking Machines Lab Is Making
The company hasn’t given out any specific product information, but Murati has shared a general idea:
- The startup’s main goal is to create multimodal AI systems that can work with text, images, audio, and maybe even video.
- The company is focusing on human-AI collaboration instead of fully autonomous AI agents. This makes it easier and more productive for people to work with AI.
- A big part of the first product will be open-source, so smaller companies, researchers, and independent developers will be able to build on its models.
This open, collaborative method is very different from the more closed ecosystems of rivals like OpenAI, Google DeepMind, and Anthropic.
Murati and her team want AI that can learn from a wide range of human skills, not just math and software development. These skills include science, engineering, healthcare, education, and creative work.
A Group of the Best AI Experts
Thinking Machines Lab’s people are just as important as its money. Almost two-thirds of the people who started it used to work for OpenAI. Important members of the team are:
John Schulman – is one of the people who started OpenAI.
Barret Zoph – used to be the VP at OpenAI.
Lilian Weng – is an expert on AI alignment and safety.
In addition to working on OpenAI, team members have also helped build important open-source AI projects like PyTorch, OpenAI Gym, and Segment Anything.
This gathering of top-tier talent is part of a larger trend in AI: top researchers and engineers leaving established labs to do work that is more independent and aligned with their goals, like Anthropic (Dario Amodei) and Safe Superintelligence (Ilya Sutskever).
The Investment Scene
A recent Pitchbook report said that AI startups generated more than 64% of the $162.8 billion in investment in the United States. startups in the first half of 2025. There are no signs that investors are losing interest, especially in areas that focus on AI safety, open-source access, and making products that work.
NVIDIA and AMD, two of the biggest hardware companies, are some of the people who support Thinking Machines. Their chips power most large-scale AI models. Their involvement suggests that the industry is closely aligned and that there may be models in the future that are better suited for hardware.
What Comes Next
Thinking Machines Lab is going into a very competitive field where big companies like OpenAI, Google DeepMind, and Anthropic have already spent billions on infrastructure and research.
Thinking Machines Lab is in a great position to make a difference, though, because it has a world-class team, strong funding, and a commitment to openness, personalization, and collaboration.
The startup’s first public demonstration of this new vision of AI will be their first product release, which is expected to happen in the next few months.
Conclusion
In just a few months, Thinking Machines Lab has gone from being a secret startup to one of the most closely watched AI companies. With $2 billion in funding, a $12 billion valuation, and a strong leadership team, the company is ready to play a big role in the future of artificial intelligence.
Mira Murati’s new business is not only making advanced models, but it’s also rethinking how AI should help people by putting human-AI collaboration, open research, and responsible innovation first.
One thing is clear to everyone in the industry: Thinking Machines Lab isn’t just thinking about the future of AI; it’s making it happen.
