David Silver's Startup Raises $1.1B for AI Without Human Data

British lab Ineffable Intelligence, launched just months ago by former DeepMind researcher David Silver, has raised $1.1 billion at a $5.1 billion valuation.

The very fact of such a round looks like a huge bet not just on a new company, but on a new technical idea: creating AI that acquires knowledge and skills without relying on massive arrays of human data. For the market, this is an important signal. In recent years, the main race in AI has revolved around increasingly large models trained on texts, images, code, and other data created by humans.

This approach has already yielded impressive results, but at the same time has surfaced limitations: there is not an infinite supply of quality data, pressure around copyright is growing, and gains from simple scaling are becoming increasingly expensive. Against this backdrop, the thesis about a system that learns not from a human archive but from its own interaction with tasks and environment sounds like a claim to the next stage of industry development. The name David Silver makes this bet particularly notable.

He is known as one of DeepMind's key researchers and one of the main popularizers of approaches in which an agent learns through experience, search, and self-play. This line of research underlay DeepMind's high-profile breakthroughs in game environments, where systems not only replicated human examples but found their own strategies and often went beyond human intuition. Therefore, Silver's new company is perceived not as another trendy AI startup, but as an attempt to scale reinforcement learning and autonomous search ideas to a broader class of problems.

The formulation about learning without human data does not mean that humans completely disappear from the process. Rather, it's about a shift in emphasis: less dependence on manually created corpora and annotation, more reliance on simulations, task generation, verifiable environments, feedback from results, and internal improvement loops. If such an architecture works, it could potentially allow creating systems that not only retell what humanity has already accumulated, but develop new strategies and knowledge in the process of action.

This is especially important where historical data is scarce, of poor quality, or too limiting to the model within the frames of past experience. The scale of financing deserves special attention. A round of $1.

1 billion for a lab that appeared just months ago shows how aggressively capital continues to flow into AI infrastructure and fundamental research. Investors in such a case are not buying revenue and not a confirmed product market, but a combination of founder reputation, scientific school, and the chance to be first to occupy a position in the next big wave. The $5.

1 billion valuation underscores that the market is willing to pay dearly for teams that offer an alternative to the current LLM development path and promise a more general way of training machines. But alongside ambitions, there is serious risk here. Building a system that truly learns without a human dataset is much harder than formulating an idea in a headline.

It needs quality training environments, self-verification mechanisms, resistance to error accumulation, and ways to transfer skills from artificial scenarios to the real world. Moreover, even very strong results in games and simulations do not guarantee the same progress in open and chaotic tasks, where there is too much ambiguity, hidden variables, and safety requirements. What this means in practice: the AI market is again betting not just on scale, but on a paradigm shift.

If Ineffable Intelligence can prove that autonomous learning can be reliably transferred beyond gaming and laboratory environments, the industry will gain a powerful argument for models that learn by doing rather than just reading human internet. If not, this round will remain an example of how expensively today the mere possibility of going beyond the current generation of AI is valued.