Cancer is not a single disease. It is a catch-all term for the uncontrollable growth of abnormal cells across more than 100 tumour types in the brain alone, each underpinned by distinct genetic mutations and shaped by patient-specific biology. This is the foundational reality that big tech's cure-cancer rhetoric consistently elides. Dario Amodei of Anthropic suggests cancer could be cured within a decade. Sam Altman of OpenAI muses that "maybe with 10 gigawatts of compute, AI can figure out how to cure cancer." Google DeepMind spin-off IsoLabs declares its mission to "solve" all disease. The framing converts an impossibly complex biological problem into an engineering bottleneck — just add compute. The clinical reality is more specific and more useful. At the Mayo Clinic, radiologist Ajit Goenka has been training machine-learning models on CT scans to detect pancreatic cancer before clinical diagnosis — identifying microscopic changes invisible to human eyes in scans previously read as normal. A 2022 proof-of-concept paper confirmed the approach works. At the University Hospital Heidelberg, Felix Sahm leads the EU-funded EUcanAI collaboration using agentic AI to accelerate the brain tumour treatment pipeline, delivering tumour sequencing results during surgery rather than days later. These are real, bounded gains — not cures, but meaningful reductions in diagnostic delay and treatment uncertainty. The equity argument is the strongest case for AI in oncology. Sherene Loi, a breast cancer specialist at the Peter MacCallum Cancer Centre in Melbourne, points out that machine-learning models could deliver specialist-level diagnostic capability to rural communities and under-resourced facilities. This is AI as infrastructure — extending existing medical knowledge to places that lack it, rather than generating fundamentally new knowledge about disease. But physician and scientist Emilia Javorsky of the Future of Life Institute delivers the sharpest counterpoint: "We cannot compute the fundamental truth of biology, we can only measure it." Thirteen years into the AI drug discovery movement, not a single FDA-approved drug has cleared the full regulatory bar through AI-driven discovery. The pipeline from algorithmic insight to approved treatment remains brutally long, expensive, and governed by biological complexity that does not yield to scale. The tension here is structural. Big tech needs the cancer narrative to justify extraordinary capital expenditure on datacentres and compute infrastructure. The environmental backlash against AI's energy consumption is growing, and "we might cure cancer" is the most emotionally compelling counter-argument available. This creates a feedback loop: grander claims justify larger investments, which require grander claims to sustain. The actual scientists doing the work speak in terms of efficiency, equity, and incremental diagnostic improvement — language that does not move stock prices. The pancreatic cancer numbers frame the stakes. It is the 12th most common cancer worldwide and the sixth most common cause of cancer death. The WHO describes its prognosis as among the least favourable. If AI-driven early detection tools like Goenka's can shift diagnosis even months earlier, survival outcomes improve materially. That is not a cure. It is something arguably more valuable: a tool that works within biological reality rather than promising to transcend it. The honest trajectory is clear. AI will make cancer detection faster and more equitable. It will personalise treatment plans and reduce diagnostic uncertainty. It will not cure cancer in any timeframe that fits a product roadmap. The question is whether society can invest in the real, bounded gains without requiring the fictional moonshot narrative to justify the spending.
