2026 → 2037 · A path to EOD 2037
A decade to change
a lifetime.
Dario Amodei argues that AI could help defeat most diseases within 5–10 years of powerful AI arriving. Machines of Loving Grace ↗ We call that goal End Of Disease (EOD) 2037: a median lifespan of 100 by 2037.
Reaching EOD 2037 would require a 20× acceleration in mortality improvement compared to today. Here is a scenario for how that could happen.
USA · Median lifespan
The scenario
One path to EOD 2037, in five stages.
The stages run from discovery in 2026 to longer lives in 2037, with three milestones each. This scenario is hypothetical, based on the mortality projections and analysis in Forecasts & analysis ↗.
AI finds cracks in biology’s hardest problems.
In the next two years, we predict that AI will become a scientist, moving from answering questions to generating testable biological insight. Three main inflection points will drive this change.
AI makes its first fundamental discoveries
AI-driven findings appear in Cell, Nature, and Science, with important results independently replicated.
Automated research becomes the norm
Research systems connect biomedical datasets, build biological models, and use experimental results to choose the next hypothesis.
Early AI-designed drugs outperform
Matched comparisons show AI-designed programs reaching validated candidates or first human trials significantly faster, without sacrificing candidate quality.
The AI-driven research loop
Every result changes the next question.
Connect molecular measurements, cell behavior, and patient outcomes. Each experiment informs the next prediction.
Research investment doubles. Promising ideas scale up.
By 2029–2030, we predict that AI discoveries will become a predictable output of data generation. That will make basic biology far more investable, much as scaling laws made it easy to justify scaling language models. Three shifts will follow.
2× funding with shared infrastructure.
Total biomedical research funding doubles from 2026. Government, academia, pharma, and AI companies establish large public data-generation consortia.
New drug submissions soar to record highs
Applications for human testing and marketing approval reach record levels, and the FDA implements AI-driven review to accelerate development.
Digital organisms begin to show promise
Models linking molecules, cells, tissues, and organs begin predicting unseen experimental results and treatment responses, guiding the next generation of therapies.
Chart sources
Research!America / TEConomy (2026), Table 1 ↗. All-sector U.S. medical and health R&D, not just NIH; includes pandemic funding. Adjusted using BLS CPI-U ↗. 2026 holds the 2025 estimate flat in real terms; the dashed curve doubles it by 2030.
U.S. biomedical research funding
The research engine, doubled by 2030.
+18.9% per year · 2026–2030
Annual funding · billions of 2025 dollars
One shared research infrastructure
AI becomes part of everyday healthcare.
In 2031–2032, we predict that frontier AI labs will go full-stack, bringing healthcare directly to the consumer. AI will diagnose disease and prescribe drugs, managing each patient’s personalized healthcare journey. To keep up, the FDA will revolutionize how drugs are tested and approved.
Breakthrough cancer drugs approved at phase 2
Selected AI-designed therapies gain accelerated or conditional approval after compelling phase II results, with continuing safety and confirmatory studies.
Frontier AI labs open personalized healthcare clinics
Licensed clinicians and AI systems work together on personalized prevention, diagnosis, treatment, and follow-up. Each clinic runs its own experimental lab, imaging, genomics, and blood testing, and stays connected to patients through wearables and on-demand check-ins.
A single-trial pathway from preclinical work to approval
For validated applications, in silico models outperform animal models. Human-cell and tissue experiments lead directly into monitored human testing; selected therapies qualify for approval on one rigorous trial.
AI moves to the center of care.
Diagnosis, prescriptions, and follow-up in one place, informed by the data patients choose to share.
Diseases become chronic conditions to be managed, rather than a source of early death.
By 2033–2034, we predict that better models and adaptable treatments will begin changing what is treatable. Three advances will mark the turn.
Full-stack AI clinics reach half of America as a subscription service
50% of Americans subscribe directly to AI clinics run by frontier AI labs, rather than paying for insurance. The clinics now provide the whole stack: they design tailored drugs, manufacture them, and deliver the care.
Cancer becomes a chronic condition
Early diagnosis catches cancers before they turn invasive. Repeated diagnosis and simultaneous targeting of multiple vulnerabilities then allow treatments to adapt before options run out.
Replacement organs address loss of organ function
Rather than treating the disease, we replace the organ. Breakthroughs in surgical delivery and in stem-cell manufacturing and growth enable lab-grown hearts, livers, and other organs to address irreversible organ failure.
Personalized cancer interception.
From each tumor, tailored drugs are synthesized to intercept it.
Early diagnosis enables continuous, real-time monitoring. Each time the tumor changes, new drugs are designed and made to match it.
A new frontier of health
From 2035 to 2037, we predict that the ambition will move beyond treating individual diseases to extending healthy life. Three milestones will close the decade.
Anti-aging therapies target late-life frailty and comorbidity
Therapies reduce frailty and disability, extending healthy life into the oldest ages.
A median lifespan of 100
Anti-aging therapies and other advances lower late-life mortality enough to bring period median lifespan to 100.
Health becomes a human right
Health becomes a guarantee, not just access to treatment. When someone falls ill with a condition that has no good treatment, AI spins up the lab research and drug development needed to treat them. Society commits to that work for every person, whatever the illness and whoever the patient.
Health is a human right.
If you get sick, AI spins up the research to treat you.