He’s not the only tech executive making this exact prediction this month. Anthropic’s CEO said the same thing days later, revealing he survived cancer himself. A cancer researcher’s response cuts through both: the breakthrough won’t come from whoever builds the biggest computer.
Arm CEO René Haas told the BBC Tuesday that AI will help cure cancer within our lifetimes — but once the chip industry solves a supply bottleneck that’s already constraining the research. “I’ve always thought that the killer app for AI is health,” Haas said. “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. I believe in our lifetime, AI will help cure cancer.” Haas has led Arm since 2022, following seven years at Nvidia as vice president of its computing products business, and served on the board of British pharmaceutical giant AstraZeneca.
What’s Happening & Why It Matters
It’s the Chips
Haas’s optimism came with a specific, unglamorous caveat. Reaching a cancer cure will require enormous amounts of computing power, and the chips needed for both cancer research and humanoid robot development are being consumed by AI companies building data centres measured in multiple gigawatts. Haas expects memory supply to hold fast for some time — a claim that lines up with the RAMageddon crunch TF has tracked throughout 2026, including in RAMageddon Hits MacBook Air Supply as Chip Shortage Deepens.
That’s an uncomfortable perspective for Arm’s business, even if Haas didn’t state it. The chips consumer devices need, and the chips cancer research needs, are drawing from the identical constrained supply — meaning every MacBook or Xbox price increase TF has covered this year represents compute capacity that isn’t going toward the medical research Haas is describing. Arm’s architecture is underneath much of that AI infrastructure, giving Haas a specific vantage point on where the bottleneck lives.
Amodei’s Claim Days Later
Haas wasn’t making this prediction in isolation. Days after his BBC interview, Anthropic CEO Dario Amodei published the essay TF covered in Amodei Says Rogue AI Swarms Could Take Over the Internet in Six Months, and within that same essay, Amodei made a specific medical prediction of his: AI could cure most major diseases within five to ten years — a timeline that places the breakthrough between 2031 and 2036. Amodei also revealed, for what appears to be the first time, that he’d survived early-stage cancer himself. “I feel the urgency personally,” he wrote, explaining that his father died from a disease that was cured only a few years after his death.
That’s two frontier-tech CEOs, within the same week, making identical predictions about AI curing cancer — one from a chip designer whose hardware powers the industry, one from an AI lab CEO who’s calling for that same industry to slow down. Amodei’s earlier 2024 essay had already laid the groundwork for this claim, arguing powerful AI could compress “50 to 100 years of biological progress” into five to ten years by accelerating drug discovery and genetic disease research.
A Cancer Researcher Pushes Back
Not everyone in the field shares the optimism, and the pushback comes with real domain expertise. Professor Chris Bakal, of the Institute of Cancer Research in London and CEO of Sentinel4D, restated the question Haas and Amodei are answering. “The real question was no longer whether we use AI; it’s what we feed it,” Bakal said. His lab trains AI on data generated from patient samples — not scraped from the internet — and his conclusion cuts against the scale-first logic behind both CEOs’ predictions: “It does not need a giant data centre to run. The future of medical AI will not belong to whoever builds the biggest computer. It will belong to whoever has the right measurements.”
That’s an insightful theory of the bottleneck compared to Haas’s chip-supply view. Haas argues the constraint is compute capacity — more chips, faster progress. Bakal argues the constraint is data quality — the right patient measurements, not raw processing power. Both can’t be fully right, and the gap between them is where AI medical research is being fought out, outside public view.
TF Summary: What’s Next
Neither Haas nor Amodei has offered a specific technical roadmap beyond their stated timelines. Amodei’s essay calls for slowing the pace of AI development. This position is somewhat at odds with his medical-progress predictions, which require rapid capability advances to materialise on schedule. Arm has not detailed specific initiatives tying its chip architecture to cancer research applications. Bakal’s lab continues developing patient-data-trained AI models independent of the large-scale infrastructure debate.
MY FORECAST: Expect Bakal’s data-quality argument to gain more traction among practising researchers than the compute-scale predictions dominating headlines, given how his point addresses a documented weakness in large, internet-scraped training datasets for specialised medical applications. Haas’s chip-shortage claim will prove the more durable near-term constraint regardless of which theory proves correct—a genuine breakthrough requiring either massive compute or curated patient data, both of which depend on infrastructure and institutional access that are scarce today. Watch whether Amodei’s specific 2031-2036 window is a benchmark critics cite if AI-driven medical breakthroughs fail to materialise on that timeline, as earlier AGI predictions have been revisited and challenged throughout 2026.
Related Stories
- Amodei Says Rogue AI Swarms Could Take Over the Internet in Six Months
- RAMageddon Hits MacBook Air Supply as Chip Shortage Deepens
- AI in Medicine: Two Seconds to Diagnose, Wrong Drug Name to Prescribe