Ineffable Intelligence, the London startup founded by the DeepMind researcher behind AlphaGo, has named Google Cloud as the infrastructure partner for its frontier AI lab. The deal, announced at Google Cloud’s London summit on 16 June, gives a company with grand ambitions and no product the computing scale its founder says the work demands.
The founder is the reason the announcement carries weight. David Silver led the research behind AlphaGo, the program that beat the world champion at Go, and AlphaZero, which taught itself chess, shogi, and Go from nothing but the rules.
He spent more than a decade at DeepMind running its reinforcement-learning group before leaving to start Ineffable. The new company is a bet that the self-play idea behind AlphaZero can be turned on the world itself.
That bet has a name: a “superlearner.” Ineffable describes a system that discovers knowledge from its own experience, from basic motor skills up to scientific breakthroughs, without leaning on human-generated data.
Where today’s large models train on static datasets scraped from the internet, a superlearner would generate, evaluate, and learn from experience in real time. The company frames the goal in cosmic terms, making “first contact” with superintelligence.
Whatever one makes of the framing, the computing requirement is real and unusual. Experience-based learning, the company says, places different demands on hardware than training on fixed data, needing tightly coupled training and inference, high-performance networking, and enormous scale.
To meet it, Ineffable will deploy one of the largest clusters of Google’s A5X instances, powered by Nvidia’s forthcoming Vera Rubin NVL72 GPUs.
The choice of partner is a pointed one. Silver said the lab evaluated the market and picked Google Cloud for its integrated AI Hypercomputer architecture rather than a plain “box of chips” GPU rental, the kind of systems-level setup that bundles processors with networking and storage.
“Training frontier models requires more than just raw compute,” he said, casting the decision as one about orchestration rather than horsepower alone.
For Google Cloud, the win is a marquee name in a fiercely contested market for AI workloads. Thomas Kurian, its chief executive, said Ineffable was using “our full-stack AI Hypercomputer, from Jupiter networking to our optimized storage,” and pitched the deal as evidence that frontier labs are choosing Google to move faster. Landing a lab of Ineffable’s profile is the kind of reference customer that cloud providers compete hard to claim.
It also extends a deepening tie between Google Cloud and Nvidia, whose Vera Rubin platform anchors the deal, at a moment when the two are jointly pushing their hardware into ever-larger training clusters.
There is a geographic angle the companies were keen to stress. By basing the lab in London and building on Google’s infrastructure, Ineffable casts itself as a cornerstone of Europe’s AI ambitions and a magnet for engineering talent that might otherwise drift to the United States.
For a continent anxious about its dependence on American AI, a homegrown frontier lab of this profile is the kind of project policymakers like to point to, even if the compute underneath it is American-supplied.
The deal follows a startup origin story that is striking even by current standards. Ineffable raised a $1.1bn seed round, billed as the largest in European history, and was later backed by Sequoia and Nvidia at a reported $5.1bn valuation, all for a company with no product, no revenue, and no public roadmap. The market is pricing the founder’s track record and the tractability of the thesis, not anything shipped.
That is the wager in plain terms. A researcher with a genuine claim to having built systems that taught themselves, now armed with a billion dollars and a vast cluster, is trying to do it again at the scale of all human knowledge. The infrastructure is now arranged, but if the superlearner arrives is the part no amount of compute can guarantee.
Stephan is the sports journalist for the Maple Grove Report.
