Arm chief: AI will help cure cancer — chip shortage is slowing progress

Arm chief executive Rene Haas told the BBC Big Boss Interview podcast that AI will eventually help find a cure for cancer, but that modelling a cell and how a DNA marker is impacted by cancer is currently ‘‘too complex’’ for humans and present computers.

Haas said as more models are fed into computers and those computers become more sophisticated, they will solve those problems. He also predicted widespread humanoid robots within five years and said Arm’s chips will significantly power self-learning robots in manufacturing, cleaning, security, construction and repairs within a decade.

Haas warned that a shortage of various types of chips is holding back the roll-out of AI and the construction of data centres, pointing to planned ‘‘multi-gigawatt’’ facilities in France and the US and proposals to put data centres in space. ‘‘We are absolutely in a supply-constrained environment. Can you get enough chips? We need more fabs,’’ he said.

He expressed scepticism that the UK needs to build fabs, arguing they are very expensive, require specialised workers and large natural-resource inputs and a broad ecosystem.

Haas noted Arm designs processor architectures used in hundreds of billions of devices and that its power-efficient technology is now used in half of AI data centres worldwide. He said demand for the Arm AGI chip, launched in March at Meta’s request, has exceeded $2bn.

The report also quotes Prof Chris Bakal of the Institute of Cancer Research, London, and CEO of Sentinal4D, who said the critical issue is what data is used to train AI. He said labs train AI on data generated from patient samples rather than scraped from the internet, and that the future of medical AI will favour those with the right measurements rather than simply the largest computer.