Thore Graepel Seeks Tens of Millions for Metis Reasoning
The former Google DeepMind researcher is developing AI systems designed to plan and act under uncertainty. No financing round has been publicly completed.
Thore Graepel, a researcher associated with Google DeepMind’s AlphaGo project, is seeking tens of millions of dollars for a startup designed to help AI systems reason, plan and act under uncertainty.
The startup, Metis Reasoning, is discussing an initial financing with a small group of backers. The fundraising remains preliminary, and no financing round has been publicly completed.
Metis Reasoning may later seek hundreds of millions of dollars at a higher valuation. That potential financing also has not been completed, and the company has not publicly disclosed investors, a valuation or a product launch.
Graepel wrote that the project aims to bring “AlphaGo-style reasoning” to frontier AI systems capable of planning and acting under uncertainty.
“I build machines that reason and act — from weighing the odds at web scale, through AlphaGo, to a new project that brings AlphaGo-style reasoning to frontier AI,” Graepel wrote.
The project aims to extend search-and-learning methods beyond games to AI systems that must plan and act under uncertainty. Possible applications include robotics, science and engineering.
AlphaGo combined deep neural networks, search algorithms and reinforcement learning. It defeated European Go champion Fan Hui five games to zero and later beat Lee Sedol 4-1 in a 2016 match.
Graepel’s career history includes work at Google DeepMind from 2015 to 2021, followed by Altos Labs from 2021 to 2025 and Google DeepMind’s Post-AGI team from 2025 to 2026.
For now, Metis Reasoning remains focused on developing its AI project while its initial fundraising discussions continue.
