DeepMind's Former CEO David Silver Raised $1 Billion to Create Superintelligence Without LLMs

David Silver, one of the pioneering figures at DeepMind whose work in artificial intelligence has had a global impact, has successfully closed a $1 billion seed round for a new research initiative aimed at creating superintelligence without relying on large language models (LLMs). This investment funds a newly established laboratory in London, named “Ineffable Intelligence,” which was founded in early 2026 and has generated significant buzz in the AI world. Silver is renowned for his role as the architect of AlphaGo and AlphaZero at DeepMind from 2016 to 2024, projects that demonstrated AI’s superhuman performance in games.

Superintelligence Beyond LLMs: A New Paradigm

Silver’s new project does not rely on large language models (LLMs), which form the foundation of the current AI industry. Instead, it is built upon classical reinforcement learning, simulation-based learning, and neural architectures inspired by biological intelligence. Its goal is to develop a system that understands real-world dynamics and possesses general-purpose decision-making capabilities—not merely making data-driven predictions. This approach represents a complete departure from the LLM-centric strategies of companies like OpenAI, Google DeepMind, and Anthropic.

Investment and Team Contributions

The $1 billion in capital was raised by a consortium of seven global venture investors, including Andreessen Horowitz from the U.S., Speedinvest from Europe, and a new fund affiliated with Sequoia Capital from Asia. After leaving DeepMind, Silver assembled a team of over 40 researchers in London in 2025. The majority of the team consists of experts from DeepMind, Google Brain, and MIT. Additionally, a partnership has been established with the University of Cambridge to integrate neuroscience with AI in the pursuit of understanding the fundamental mechanisms of intelligence.

2026 Goals: The First Test System

In the second quarter of 2026, Ineffable Intelligence plans to unveil its first superintelligence prototype. This system aims to demonstrate superhuman performance in real-time strategic games, complex physical simulations, and multi-agent environments. The project is not merely a technological development effort—it also seeks to redefine the ethical boundaries of artificial intelligence. Silver asserts, “Superintelligence is born not from data, but from understanding,” offering an alternative vision to the data-intensive paradigm of LLMs.

Impact on the Industry

This investment is reshaping the debate over “LLM dominance” in the AI industry. Experts believe Silver’s approach could revolutionize fields such as scientific discovery, medical diagnostics, and autonomous systems. Investors anticipate that this model will prove more efficient in the long term by addressing challenges such as the high costs of LLMs, data exhaustion, and generalization limits. By 2026, the successes of this project could become a pivotal turning point in determining the future of artificial intelligence.

• Project name: Ineffable Intelligence
• Location: London, United Kingdom
• Investment amount: $1 billion (seed round)
• Lead: David Silver (former principal researcher at DeepMind)
• Core technologies: Reinforcement learning, neuroscientific architectures, simulation-based learning
• LLM usage: None
• First prototype target: 2026 Q2