Sequoia and Nvidia back David Silver’s Ineffable Intelligence at $5. 1B

London-based artificial intelligence startup Ineffable Intelligence has secured backing from Sequoia Capital and Nvidia at a valuation of $5.1 billion. The financing round, led by Sequoia managing partner Alfred Lin and partner Sonya Huang, values the company despite its lack of a product, revenue, or public roadmap. Nvidia’s venture arm contributed at least $250 million to the deal, which was confirmed following reports in February 2026. The startup was founded by David Silver, a British researcher who spent over a decade at Google DeepMind leading the creation of AlphaGo, AlphaZero, and AlphaStar. Silver left DeepMind in late 2025 and incorporated Ineffable Intelligence in November 2025, later being appointed director in January 2026. He remains a professor at University College London and chose London deliberately for its academic pipeline and proximity to DeepMind headquarters. Ineffable Intelligence’s core thesis challenges the prevailing paradigm of large language models. Silver argues alongside Richard Sutton that LLMs are limited because they learn exclusively from human-generated data. Instead, the company plans to scale reinforcement learning, allowing AI to learn from interaction with its environment through trial and error. This approach aims to produce strategies and insights that no human has conceived, citing AlphaGo’s famous Move 37 as evidence of machine reasoning beyond human intuition. This investment reflects a broader trend in the 2025-2026 AI market where valuations prioritize researcher credibility over current capabilities. Similar patterns were observed with Ilya Sutskever’s Safe Superintelligence, Mira Murati’s Thinking Machines Lab, and Yann LeCun’s AMI Labs. By this metric, Silver’s track record commands a premium, making Ineffable Intelligence one of the most valuable pre-product AI startups in Europe. Critics note that reinforcement learning has historically struggled in open-ended real-world environments where reward signals are ambiguous. Defining winning conditions for general intelligence remains a central unsolved problem in AI safety. Investors are effectively paying $5.1 billion for the possibility that scaling this approach unlocks qualitatively different capabilities, though the claim remains untested.