Key Takeaways
- • David Silver, architect of AlphaGo and former DeepMind lead researcher, has departed Google DeepMind to launch Ineffable Intelligence, a London-based AI startup focused on building superintelligence through reinforcement learning rather than large language models.
- • Ineffable Intelligence secured a $1 billion seed round led by Sequoia Capital at a $4 billion pre-money valuation, representing Europe's largest seed funding round, with Nvidia, Google, and Microsoft reportedly in talks to invest.
- • Silver's startup is pursuing an "Era of Experience" model where AI agents learn predominantly from direct interaction with environments rather than human-generated data, positioning this approach as a path beyond the limitations of current large language model architectures.
- • Silver, age 49 and a professor at University College London, was instrumental in DeepMind's landmark breakthroughs including the 2016 AlphaGo victory over world Go champions, and has been with DeepMind since its 2010 founding when he knew cofounder Demis Hassabis from university.
- • The funding and strategic positioning reflect a directional bet by major technology investors that reinforcement learning and experience-based learning architectures represent the next frontier in AI development, potentially superseding the current dominance of large language model approaches.
Summary
David Silver, one of artificial intelligence's most accomplished researchers, has left Google DeepMind to establish Ineffable Intelligence, a London-based startup explicitly designed to develop superintelligence through reinforcement learning methodologies. Silver's departure from DeepMind, where he had been on sabbatical before formally leaving, marks a significant institutional shift in the AI research landscape. Google DeepMind confirmed his departure, with a spokesperson acknowledging that "Dave's contributions have been invaluable, and we're grateful for the impact he's had on our work at Google DeepMind." Silver maintains his professorship at University College London while leading the new venture.
The funding achievement underscores institutional conviction in Silver's strategic direction. Ineffable Intelligence raised $1 billion in a seed round led by Sequoia Capital, valuing the company at approximately $4 billion pre-investment. This represents Europe's largest seed-stage funding round. Notably, U.S. technology giants Nvidia, Google, and Microsoft are reportedly in active discussions to participate in the round, indicating broad acceptance of the startup's fundamental thesis among major technology investors. The company was formally established in November 2025, with Silver appointed as director in January 2026.
Silver's research trajectory positions him as a credible architect for this endeavor. He was among DeepMind's first employees when the organization was founded in 2010 and played an instrumental role in the company's most celebrated achievements, including the 2016 AlphaGo breakthrough that demonstrated an AI system could defeat world-class human competitors in the ancient game Go. He also contributed to subsequent innovations including AlphaZero and MuZero, all of which leveraged reinforcement learning principles.
The strategic differentiation of Ineffable Intelligence rests on a fundamental premise about the trajectory of AI development. In an academic paper co-authored by Silver, he argued that while artificial intelligence has achieved significant advances through training on large volumes of human-generated data, this approach faces inherent limitations in domains such as mathematics, coding, and scientific discovery where "the knowledge extracted from human data is rapidly approaching a limit." His framework, termed the "Era of Experience," proposes that a new generation of AI agents will acquire superhuman capabilities by learning predominantly from direct experience and interaction with environments rather than relying exclusively on human-derived training data. This positioning implicitly challenges the dominance of large language model architectures that have defined the current competitive landscape in AI development.
The funding achievement underscores institutional conviction in Silver's strategic direction. Ineffable Intelligence raised $1 billion in a seed round led by Sequoia Capital, valuing the company at approximately $4 billion pre-investment. This represents Europe's largest seed-stage funding round. Notably, U.S. technology giants Nvidia, Google, and Microsoft are reportedly in active discussions to participate in the round, indicating broad acceptance of the startup's fundamental thesis among major technology investors. The company was formally established in November 2025, with Silver appointed as director in January 2026.
Silver's research trajectory positions him as a credible architect for this endeavor. He was among DeepMind's first employees when the organization was founded in 2010 and played an instrumental role in the company's most celebrated achievements, including the 2016 AlphaGo breakthrough that demonstrated an AI system could defeat world-class human competitors in the ancient game Go. He also contributed to subsequent innovations including AlphaZero and MuZero, all of which leveraged reinforcement learning principles.
The strategic differentiation of Ineffable Intelligence rests on a fundamental premise about the trajectory of AI development. In an academic paper co-authored by Silver, he argued that while artificial intelligence has achieved significant advances through training on large volumes of human-generated data, this approach faces inherent limitations in domains such as mathematics, coding, and scientific discovery where "the knowledge extracted from human data is rapidly approaching a limit." His framework, termed the "Era of Experience," proposes that a new generation of AI agents will acquire superhuman capabilities by learning predominantly from direct experience and interaction with environments rather than relying exclusively on human-derived training data. This positioning implicitly challenges the dominance of large language model architectures that have defined the current competitive landscape in AI development.
Go deeper than the headline
Research every firm in this story, track the ones that matter and get the daily briefing — free for 7 days.