Beyond the Limits of Human Data: Google DeepMind Veteran David Silver Founds $1.1 Billion AI Startup 'Ineffable Intelligence'

On Monday, April 27, 2026, a major announcement was made that will change the landscape of the artificial intelligence (AI) industry. Ineffable Intelligence, a startup founded by David Silver, a key researcher at Google DeepMind, has successfully raised a record-breaking $1.1 billion in seed funding.

Valued at $5.1 billion just months after its founding, the London-based company does not aim to develop simple chatbots. Instead of relying on human-authored datasets, they intend to break through the AI 'data wall' by developing 'superlearners' that generate knowledge themselves through experience.

This investment round was co-led by Sequoia Capital and Lightspeed Venture Partners, marking it as the largest seed financing in the history of European technology companies. Strategic investors such as Nvidia and Google also participated in large numbers, lending weight to David Silver's new vision.

Ineffable is building systems designed to generate knowledge from their own experience. This is a system that learns without relying on human records, presenting a new path to true superintelligence.

Founder David Silver was the lead architect of 'AlphaGo,' one of the most iconic events in AI history. Based on his experience creating the first AI to defeat a human world champion, he now seeks to solve the limitations faced by Large Language Models (LLMs) through Reinforcement Learning.

The Philosophy of 'Superlearners' Beyond Human Data

LLMs that currently lead the AI industry derive results by learning from vast amounts of human data on the internet, but this has the disadvantage of being trapped within the scope of human knowledge. The superlearners pursued by Ineffable Intelligence aim to independently create new knowledge and strategies that humans have yet to discover through countless trials and errors in simulated environments.

• Independent learning structure that eliminates dependence on human-authored data
• Knowledge expansion through self-play based on reinforcement learning
• Infinite data generation and learning mechanisms in simulated environments

Concerns about the so-called 'Data Wall' are deepening in the industry. As predictions emerge that high-quality human-generated data for AI learning will soon be exhausted, Ineffable Intelligence's non-human data-based learning method is emerging as an essential alternative for next-generation AI.

The fact that Nvidia and Google invested in this startup is particularly noteworthy. Nvidia, a hardware leader, and Google, Silver's former employer, are seeking to gain an advantage in the competition for next-generation AI hegemony by preemptively investing in technology that has the potential to disrupt the LLM paradigm they currently lead.

This large-scale investment also holds significant meaning for the European technology ecosystem. The rise of London-based Ineffable Intelligence is diversifying the axis of AI innovation that was concentrated in Silicon Valley and is impressing upon the market that reinforcement learning is the key technology to overcome the stagnation of LLMs.

Risks of Self-Generated Intelligence and the Journey to Superintelligence

However, technical and safety concerns regarding self-generated intelligence also exist. Experts warn that if AI models learn only from data they generate themselves without the anchor of human truth, 'model collapse' may occur, where the model loses real-world variability or becomes trapped in specific errors.

Ineffable Intelligence's challenge is expected to be a testing ground to prove that reinforcement learning is the path to true superintelligence beyond the limits of LLMs. If David Silver's hypothesis is proven, humanity will enter a new era of intelligence that transcends human limits, moving beyond the stage of merely mimicking human intelligence.