AlphaGo Co-Creator Graepel Launches London AI Firm Metis, Seeking Tens of Millions to Challenge Large Model Paradigm
Former Google DeepMind research lead and AlphaGo co-creator Thore Graepel has launched Metis Reasoning in London. The company advocates for verifiable reasoning over current large models and is seeking tens of millions in initial funding, targeting robotics and engineering sectors.
ImportanceMaterialEvidenceE2 unreplicatedWrite-upQuick
According to tech.eu, Thore Graepel, former research lead at Google DeepMind and co-creator of AlphaGo, launched the AI company Metis Reasoning in London on October 20, 2026. The firm aims to develop a "compute-efficient" alternative path, distinguishing itself from frontier large model providers like Anthropic and OpenAI.
Metis was co-founded by Graepel, serial entrepreneur Piers Ingram, robotics expert Michael Warren, and David Berman, head of AI at Cambridge Consultants. The team consists of approximately 12 people. The company argues that while current Large Language Models (LLMs) generate fluent answers, they lack an audit trail in their reasoning process, making them difficult to fully trust in high-stakes fields such as medicine, law, or engineering.
Reports indicate Metis is seeking tens of millions of dollars in initial funding. Co-founder Ingram stated the company has been "more than adequately" funded to date and claimed its architecture is "much much more" compute-efficient than frontier models, capable of solving advanced reasoning problems at a fraction of the cost. It remains unclear when Metis will publicly release its first AI model.