David Silver, the researcher behind AlphaGo, has raised a $1.1 billion seed round on the thesis that the AI industry’s dominant approach is wrong.
In a direct challenge to the large language model strategy championed by OpenAI and Google, former DeepMind researcher David Silver has secured $1.1 billion in seed funding for his new startup, Ineffable Intelligence. The London-based company, which aims to build self-learning AI through reinforcement learning, reached a $5.1 billion valuation in the largest-ever seed round for a European startup.
“There’s only a very, very small number—less than a handful of people—who have done truly foundational work,” Sonya Huang, a partner at lead investor Sequoia Capital, said. “Dave is one of them. I fundamentally agree with his thesis on where we’re going to find the next big breakthroughs.”
The round was co-led by Sequoia Capital and Lightspeed Venture Partners, with significant participation from Nvidia, Google, Index Ventures, and the UK’s Sovereign AI fund. The massive financing for a company formally established in January 2026 with no product underscores intense investor competition to back elite AI researchers.
The investment represents a multi-billion dollar bet that reinforcement learning (RL), the technology that powered AlphaGo’s success in the game of Go, is a more viable path to superintelligence than scaling LLMs on human-generated data. If successful, Ineffable’s approach could disrupt the current market dominated by LLMs and create a new paradigm for AI development, though first model benchmarks are not expected until late 2026.
The Bet Against LLMs
Silver’s thesis is a direct counterpoint to the prevailing industry method of achieving artificial intelligence by training models on vast quantities of human-generated text and images. He argues that while LLMs are powerful, they are fundamentally limited by the "fossil fuel" of human data. In contrast, he describes his RL-focused approach as a "renewable fuel" capable of learning "forever, without limit."
The vision for Ineffable Intelligence is to create a "superlearner" that discovers knowledge from its own experience within complex simulations, rather than simply mastering information created by humans. Silver uses a thought experiment to illustrate the limits of LLMs: an AI trained on historical data from a time when the world was believed to be flat would remain a "flat-earther," unable to discover the truth for itself. An RL agent, however, could potentially run its own experiments within a simulation to arrive at new scientific discoveries.
This approach is significantly more challenging than training on static datasets, which partly explains the need for such a large capital injection to build simulated environments and compute infrastructure that rival the largest LLM training runs.
A Watershed for European AI
The scale of the funding round marks a watershed moment for the European AI scene, which has historically operated in the shadow of San Francisco-based labs. It is part of a broader trend of prominent researchers leaving established giants to found their own companies, including Mistral AI’s Arthur Mensch, also formerly of DeepMind.
For the UK, the investment is a symbolic victory, suggesting that top-tier talent with a compelling thesis can attract massive Silicon Valley funding without relocating. However, the move is not without its skeptics. Google DeepMind CEO Demis Hassabis has previously voiced concerns about startups raising enormous seed rounds at multi-billion dollar valuations before developing a product or generating revenue.
Ineffable Intelligence fits this description, but investors are betting on the founder’s track record. Silver’s work at DeepMind, from leading the AlphaGo team to developing its successors AlphaZero and MuZero, provides a powerful, coherent argument for his ability to scale intelligence without relying on pre-existing human knowledge. With over a billion dollars in capital, the question for investors is whether he can translate that success from the closed world of board games to the complexity of reality.
This article is for informational purposes only and does not constitute investment advice.
