Arm chief Rene Haas predicts AI will cure cancer within this generation despite the chip shortage

Rene Haas, chief executive of Arm Holdings —the largest UK-based technology company, designer of the processors found in hundreds of billions of phones, cars, smartwatches and other devices worldwide— has said that artificial intelligence will manage to cure cancer within our generation, something he believes humans would not achieve on their own. In an interview for the BBC podcast Big Boss Interview, Haas explained that modelling a cell, a complete human being or the way a DNA marker is affected by cancer is currently "too complex a problem", not only for people but also for the computers running AI systems. He argues, however, that as more models are fed to these machines and they become more sophisticated, "they're going to solve it".

The claim carries weight coming from someone who also holds a key post at SoftBank, the Japanese group that is Arm's majority owner and which also has investments in companies such as OpenAI. Haas recently left the board of British pharmaceutical company AstraZeneca, in April, which adds context to his stated interest in the intersection of AI and medicine.

Not every voice in the sector shares the same view on how to reach that goal. Professor Chris Bakal, of the Institute of Cancer Research in London and chief executive of the company Sentinal4D, told the BBC that the relevant question is no longer whether AI should be used, but what data it is given. As he explained, labs like his train models on data generated from real patient samples, not pulled from the internet, and without needing huge data centres to run. For Bakal, the future of medical AI will not belong to whoever builds the biggest computer, but to whoever has the right measurements, an approach that —he says— can indeed cut years off the development of new treatments and deliver real benefits to patients.

Beyond medicine, Haas predicted that the chips Arm designs will drive, over the next decade, a "large-scale" expansion of humanoid robots capable of learning on their own, with applications in manufacturing, cleaning, security, bridge building and repair work. On his account, thanks to AI these robots will be able to see, learn and reprogram themselves for new tasks: a robot programmed to make beds, for example, could also learn to put out towels or empty wastebaskets.

On the impact on jobs, Haas played down fears of mass employment destruction, calling estimates of workers being entirely replaced by machines "somewhat exaggerated", and arguing that the effect will be offset by the emergence of new opportunities. In the same optimistic vein, he downplayed fears of a sharp correction in the stock market valuation of AI companies, arguing that there is long-term demand for AI.

On the business front, Haas stressed that Arm's low-power technology is already used in half the world's AI data centres, which has led the company to move towards designing its own chips. Meta commissioned Arm to develop the so-called Arm AGI chip, launched in March, demand for which —according to Haas— has been "off the charts", with orders worth more than $2 billion since it reached the market.

That growth runs up, however, against a bottleneck that Haas himself identifies as the main brake on AI expansion today: the shortage of chips needed to build data centres. He mentioned plans to erect "multi-gigawatt" data centres in France and the United States, as well as projects to install them even in space, and warned that the sector is "absolutely" in a supply-constrained environment: more semiconductor plants (fabs) will be needed before it is viable, for example, to put a data centre in orbit.

Asked whether the United Kingdom could manufacture chips on its own soil —a possibility the British government has discussed with the industry— Haas was sceptical. In his judgement, the country does not need to build its own fabs, since these plants are very costly, require highly specialised labour and large natural resources, and a broad ecosystem already exists for that purpose, an implicit reference to the current concentration of manufacturing around Taiwan's TSMC.

The interview also addresses Arm's relationship with the United Kingdom, a sensitive matter since the company was sold to Japanese investors in 2016 and floated partially on New York's Nasdaq rather than in London, a decision several British government officials have publicly lamented. Haas argued that half of Arm's workforce is still based in the United Kingdom and that the company remains by far the largest employer in Cambridge, the city where it is headquartered.

Taken together, Haas's remarks combine far-reaching medical optimism —an AI-driven cure for cancer within our lifetime— with a pragmatic reading of the industrial bottlenecks that shape that future: chip availability, where they are manufactured and the geographic split of Arm's business between the United Kingdom, the United States and Japan. The scientific counterpoint offered by Professor Bakal is a reminder, however, that real progress in oncology will depend less on raw computing power and more on the quality and specificity of the clinical data used to train the models, a nuance that tempers the scope of a prediction that was otherwise not accompanied by timelines, concrete mechanisms or additional clinical evidence in the interview.