Summary: On April 27, 2026, Bloomberg reported that David Silver’s Ineffable Intelligence raised $1.1 billion at a $5.1 billion valuation. Sequoia and Lightspeed led the round for the British lab’s “superlearner” plan, which aims to learn through reinforcement learning without human-generated training data. Google, Nvidia, the British Business Bank, and the UK sovereign AI venture fund also participated.
The news broke on April 27, 2026: Bloomberg had the story first, and TechCrunch quickly confirmed it. Ineffable Intelligence — a British AI lab that David Silver set up just months earlier — has closed a $1.1 billion round at a $5.1 billion valuation, stepping straight over the “pentacorn” line of $5 billion and up. Silver, formerly the head of DeepMind’s reinforcement learning team and a University College London professor, now counts Sequoia and Lightspeed as lead investors, with Index Ventures, Google, Nvidia, the British Business Bank, and the UK’s newly created sovereign AI venture fund also on the list.
The strangest part of the deal is what the money actually buys: neither a product nor any revenue — a research direction instead. The plan at Ineffable is a “superlearner,” a system that gains knowledge and skills through reinforcement learning’s trial and error alone, never drawing on human-generated training data of any kind. With high-quality human data running short and frontier models feeling the squeeze, no bet on a post-human-data era has been placed this bluntly.
The Superlearner’s Lineage Begins With AlphaZero
No record sells the route better than Silver’s own. He ran DeepMind’s reinforcement learning team for over a decade, and the headline result is AlphaZero: a system that never read a single human game, began from self-play alone, picked up chess and then Go, and finished ahead of the strongest specialized engines of its era.
What Ineffable wants is to carry the same self-learning principle beyond the closed board, into the open world. The company’s website invokes Darwin for the comparison: “his law explained all Life” and “ours will explain and build all Intelligence.” In a Wired interview, Silver called the venture an extension of his life’s work and pledged that money earned from Ineffable will flow to high-impact charities whose aim is saving as many lives as possible.
Reading the Fine Print of the Round
- $1.1 billion raised against a $5.1 billion post-money valuation — a company only months old walks straight into the $5B-plus “pentacorn” club
- Two investors share the lead: Sequoia Capital and Lightspeed Venture Partners
- Also taking part: Index Ventures, Google, and Nvidia, alongside the state-backed British Business Bank and the UK’s sovereign AI venture fund
- Ineffable’s executive team already includes several former DeepMind staff
The official money is worth a second glance. The sovereign AI fund is participating directly, which makes this more than a venture deal — it is Britain betting national capital on a star researcher of its own.
Star Researchers, Inflated Seeds: Inside the Coconut-Round Era
Ineffable is hardly alone. Bloomberg has a name for the 2026 pattern, the “coconut rounds”: when elite researchers strike out on their own, their first round now lands Series C-scale money. Cases in point:
- AMI Labs — the company Yann LeCun, Turing Award winner and ex-Meta chief AI scientist, co-founded — raised $1.03 billion in March 2026, with a $3.5 billion pre-money valuation
- At Recursive Superintelligence, Tim Rocktäschel, a former DeepMind principal scientist, has reportedly sealed $500 million as co-founder, and demand at one point could have carried the round to $1 billion
Put Ineffable’s $1.1 billion next to those and it marks yet another high-water point. Money crowding into a handful of people whose names are themselves technical roadmaps tells you what is really being bought: options on a paradigm shift.
Why the Third AI Hub Is London
The round adds fresh confirmation of London’s weight. Since Google acquired DeepMind in 2014, the city has been cultivated as a core AI base; Jeff Bezos’s Project Prometheus is reportedly office-hunting around Google’s London AI hub; and Ineffable now houses a group of former DeepMind executives within a single startup. When talent, private capital, and state backing pile up inside one city, the combination is hardly found anywhere outside the San Francisco Bay Area and Beijing.
Three Things Builders and Researchers Should Watch
Three developments merit close attention. One, human data’s monopoly on the moat may be ending: if the superlearner route proves out, data licensing and web crawling lose weight, while the capacity to generate experience yourself gains it. Two, evaluation has to be rebuilt — for a model trained without human data, which benchmark even applies? Most current benchmarks silently bake in the edges of human knowledge. Three, pulling back for the wider view, this wave of capital — star researchers paired with giant seeds — lines up with the 2026 opening outlook we wrote at the start of the year: the year’s story has shifted from scaling the current paradigm to hunting down the next one.