Arm CEO Rene Haas believes AI could help cure cancer within our lifetimes, but in an interview with the BBC released on Tuesday he was clear that the semiconductor industry faces a supply crisis severe enough to delay that ambition. The Arm CEO chip supply problem, as Haas described it, is an industry-wide constraint that no single company can resolve quickly.
‘I’ve always thought that the killer app for AI is health,’ Haas told the BBC. ‘AI is going to not only shorten the amount of time that those drugs can be invented, it’s going to shorten the amount of time that you test them.’ He went further: ‘I believe in our lifetime, AI will help cure cancer.’
Why the Arm CEO chip supply crunch is so hard to fix
Haas has led Arm since February 2022, having joined the British chip designer in 2013. Before that, he spent seven years at Nvidia as vice president of its computing products business. That background gives him an unusually detailed view of where the bottlenecks sit. His core argument is straightforward: the computing power needed to run frontier AI models, whether for drug discovery or humanoid robotics, is being absorbed by data centre buildouts before it can reach research applications.
‘Right now, it’s quite constrained,’ Haas said. ‘We need more fabs before we can put a data center in space, I’ll tell you that much.’ New semiconductor fabrication plants, or fabs, can cost tens of billions of dollars and take two or three years to construct. Haas described the current environment as ‘absolutely supply-constrained’ and expected tightness to persist for some time.
The scale of demand is striking in context. According to the BBC, more than 350 billion chips using Arm technology have shipped worldwide, underlining how deeply the company’s architecture is embedded across the global semiconductor ecosystem. Yet even that installed base offers little relief when each new wave of AI infrastructure requires chips and memory at a scale the industry has not previously encountered.
Meta’s AGI chip and the demand signal Arm is already navigating
One concrete illustration of that demand pressure comes from the BBC’s reporting that Meta asked Arm to develop the Arm AGI chip. Demand has been substantial: more than $2bn worth of orders came in since the chip launched in March, according to that reporting. That single product’s commercial trajectory captures the tension Haas was describing. The appetite for AI-specific silicon is real and growing, but manufacturing capacity cannot keep pace.
Haas is not alone in raising the alarm. Manish Bhatia, Micron’s chief operating officer, described the memory shortage in a January interview as ‘really unprecedented’, saying that high-bandwidth memory for AI accelerators was consuming so much manufacturing capacity that it was leaving a ‘tremendous shortage’ for conventional products such as phones and personal computers. On a February earnings call, Qualcomm CEO Cristiano Amon said ‘industry-wide memory shortage and price increases’ would determine the size of the handset market through the year, adding that ‘the whole sector is impacted by memory.’
Data centres, public opposition and the limits of AI optimism
Even setting aside the Arm CEO chip supply constraints, the case for AI curing cancer near-term is complicated by the biology. Medical researchers note that cancer is not a single disease but a collection of over 100 diseases that vary significantly between patients. While AI can help identify drug targets and model how compounds might behave, clinically relevant results remain limited. The wide range of variables in human physiology makes it difficult for models to predict reliably how a candidate drug will perform in trials.
The data centre infrastructure being built to feed AI ambitions is also generating community resistance. By the end of last year, more than 1,400 data centres had been built or approved for build-out in the United States. A Gallup survey of 1,000 adults in the US, published in May, found that seven in 10 respondents opposed constructing data centres for AI in their local area, with nearly half indicating they were ‘strongly opposed’. Concerns centred on the impact of data centre operations on air and water quality in surrounding neighbourhoods.
Haas acknowledged the longer horizon candidly. The BBC interview in which he made his cancer comments was released on Tuesday, and his position is essentially that the potential is real but the preconditions, more fabs, more memory capacity, and resolved energy constraints, must come first. Until supply catches up with demand, the computing headroom needed to run the kind of AI that might meaningfully accelerate drug discovery remains, by his own assessment, out of reach.