DeepMind's David Silver just raised $1.1B to build an AI that learns without human data

David Silver, the DeepMind research scientist who pioneered reinforcement learning breakthroughs like AlphaGo and AlphaZero, has raised a massive $1.1B funding round for his new stealth AI startup. The company's core thesis: build AI systems that learn entirely from self-play and synthetic data, without ingesting any human-generated content. This marks a radical departure from current LLMs (like GPT-4o or Claude) that rely on trillions of tokens scraped from the internet.

Silver's approach leverages reinforcement learning at scale, where agents learn by interacting with simulated environments and optimizing reward functions. The $1.1B warchest—one of the largest seed rounds in AI history—will fund massive compute clusters for self-play training. If successful, the models could exhibit emergent reasoning capabilities unconstrained by human data biases, potentially achieving AGI (artificial general intelligence) through pure environmental interaction rather than mimicking human text.