TL;DR

Rene Haas, who runs Britain’s largest home-grown technology company, believes AI will find a cancer cure human researchers could not reach unaided, and that self-teaching robots will be commonplace inside five years. He also said the thing standing in the way is mundane: there are not enough chips. And he does not think Britain should try to make them.

The claim, and the pushback

Haas’s argument on cancer is about scale of computation. Simulating a cell, or how a DNA marker behaves under cancer, defeats both people and today’s machines — but feed the models enough and the machines get there. Arm designs the processors found in phones, cars and gadgets by the hundred billion, and its shares peaked this summer at a level no British company had reached before in cash terms.

Not everyone frames the problem as compute. Chris Bakal, a professor at London’s Institute of Cancer Research who also runs Sentinal4D, said the useful question is what the models are trained on rather than whether they are used at all. His lab generates its data from patient samples instead of scraping it off the internet, and needs no enormous facility to run it. Medical AI, on his account, will be won by whoever holds the right measurements — not the biggest computer.

Robots, and the standards to go with them

The five-year robot forecast is not just talk from an interview. Arm used the same day to extend its Total Design programme to physical AI, pulling in more than 80 partners — Siemens, Hugging Face and Unitree Robotics among them — plus a Robotics Capability Framework drafted by Richard Grisenthwaite, Arm’s chief architect. The framework tries to give the industry shared language for grading what a robot can do, pinned to requirements like latency, memory, power draw and safety. Standards work of that sort usually arrives when a market is being readied for volume.

Looking forward

The constraint is fabrication. Haas pointed at multi-gigawatt sites planned in France and the United States, and at proposals to put data centres in orbit, before observing that more fabs come first. On whether Britain should build any, he was flatly negative — they cost too much, need scarce specialists and heavy resources, and depend on an ecosystem that is not here. That is an awkward answer for ministers who have been discussing exactly that with the industry, and it comes from the company they most often cite as proof British silicon matters. Half of Arm’s staff are still in the UK, where it remains Cambridge’s biggest employer.