David Silver's Startup Ineffable Intelligence Hires Six DeepMind Veterans

David Silver, the creator of AlphaGo and longtime head of reinforcement learning research at Google DeepMind, has expanded the leadership of his new startup Ineffable Intelligence with six experienced researchers and executives. All six received the title of cofounder, even though the company has existed for nearly a year and its funding round already closed in the spring.

Who Joined Silver

Four of the six new cofounders are former DeepMind colleagues of Silver's. Chris Apps, formerly a leader of technical programs and research delivery at DeepMind, will be responsible for 'accelerating the mission' of the company. Wojciech Czarnecki, a specialist in multi-agent systems, co-created AlphaStar and AI projects that played capture-the-flag, and now oversees the science team. Junhyuk Oh, who led reinforcement learning research on AlphaStar at DeepMind and pioneered using AI to discover new RL algorithms, now manages reinforcement learning operations.

The fifth cofounder, Lasse Espeholt, was responsible at Google for building the MetNet weather forecasting models and now heads compute, infrastructure and engineering strategy at Ineffable. Alexandre Laterre, formerly head of research at InstaDeep, the company acquired by BioNTech in 2023, will oversee research engineering. The last of the six, Heather Gorham, until recently a partner at the venture capital fund Flying Fish, will handle finance, compute and operations; it was reportedly she who, in 2024, persuaded Silver into a conversation about reinforcement learning that gave rise to the idea for the company.

A Bet on Reinforcement Learning

Ineffable Intelligence is distancing itself from today's dominant approach based on large language models trained on human-written text. The team is betting on reinforcement learning, in which a system learns through trial and error while maximizing a reward, an approach closer to the conditioning mechanisms known from behavioral psychology, rather than on replicating patterns from human-generated data.

The company's goal is to build a so-called 'superlearner,' a system capable of independently discovering knowledge and skills without relying on human training data. This is a direct callback to Silver's own record: as head of the reinforcement learning team at DeepMind, he was behind AlphaGo, the first program to defeat world champions at Go, and later developed further systems that learned through self-play and experimentation.

A Wave of DeepMind Departures

Silver formally left Google DeepMind and took the role of director at Ineffable Intelligence in January 2026, after more than a decade at the lab that co-created the groundbreaking AlphaGo and AlphaZero systems. His new company fits into a broader trend industry observers call 'neolabs,' a wave of new, well-funded AI labs founded by former top researchers from Google DeepMind, OpenAI and Anthropic who want to pursue their own visions of AI development outside the structures of the big tech companies.

Silver's departure, and now that of six more experienced DeepMind researchers, shows just how intense the competitive pressure is in the AI talent market. Companies like Ineffable offer not only high salaries but also, as in this case, the title of cofounder and a real equity stake in a company valued at billions of dollars, even though these hires are joining after the main funding round has already closed.

Record Funding

The $1.1 billion seed round, closed in April 2026, was the largest of its kind in the history of European tech startups. It was led by Sequoia and Lightspeed, with Nvidia and Google also among the investors; Google additionally signed a compute partnership with Ineffable in July 2026 based on Google Cloud infrastructure. The company's first check came from Flying Fish, the fund one of the new cofounders now hails from.

The current $5.1 billion valuation places Ineffable Intelligence among the best-funded European AI labs, alongside companies like Mistral AI. For the Polish market, the most relevant angle is the competition for talent: one of the new cofounders, Wojciech Czarnecki, has Polish roots and previously worked at Google DeepMind on multi-agent systems.

What's Next

Ineffable Intelligence has not yet announced a specific timeline for releasing its technology or the first results of its 'superlearner' research. The company remains at an early, pre-product stage of development, and its value rests on the team's reputation and investors' conviction that reinforcement learning, rather than ever-larger language models, could be the path to systems that surpass human capabilities.

For the AI industry, the expansion of Ineffable's leadership team signals that the race over the architecture of future AI systems is still wide open. While OpenAI, Anthropic and Google DeepMind pour billions into scaling language models, competing approaches based on reinforcement learning and learning without human supervision are attracting increasingly serious financial backing, along with more experienced researchers willing to stake their careers on them.