SubscribeLists
As models become a commodity, the moat moves to data. These are the labs sitting on the datasets that will shape the next decade of medicine.
https://cdn.prod.website-files.com/6a494c345c90bdaf071b56ce/6a4e5e030e04e4635ea6017b_hts-audio-best-health-ai-research-labs-to-watch-v3-felix.mp3
The most valuable data in the world right now is not on the open internet. It is locked inside health research labs.
As foundation models turn intelligence into a commodity, the scarce thing becomes proprietary data and the labs that generate it. These are the health AI research labs I watch most closely, because what they build today becomes the medicine of the next decade.
The DeepMind spinout using AI to design drugs, led by Nobel laureate Demis Hassabis and built on the legacy of AlphaFold. I explained why its data partnerships matter more than its $2.1 billion raise.
Why watch it: If AI compresses drug discovery from a decade to a fraction of that, it starts here.
The parent of the AlphaFold breakthrough that cracked protein folding after 50 years of stalled progress, still pushing the frontier of biology and medicine.
Why watch it: DeepMind repeatedly solves problems the field considered permanently stuck. Betting against it has been a mistake.
Behind the reasoning model that helped solve 18 rare disease cases doctors could not crack, plus a wave of clinician and patient products I covered in my launch analysis.
Why watch it: When a general reasoning model can diagnose rare disease, the definition of a medical tool changes.
Its Claude models now power enterprise healthcare deployments, including Banner Health's rollout to more than 55,000 employees, with a stated focus on safety in high-stakes settings.
Why watch it: In medicine, a model you can trust to refuse and stay safe matters as much as a model that is clever.
The combination behind Dragon Copilot, the ambient documentation tool deployed across hundreds of health systems and folded deep into the Epic workflow.
Why watch it: Microsoft turned research into the single most widely deployed clinical AI tool in the world.
A health system running as a research lab, co-developing AI on one of the richest de-identified clinical datasets anywhere, including palliative care identification and nurse-focused documentation with Abridge.
Why watch it: Mayo has the data, the clinicians and the trust. That combination is almost impossible to replicate.
Mount Sinai built a model that recommends against blood thinners in up to half of certain atrial fibrillation patients, and signed the first enterprise OpenEvidence deal. Academic centers like it are quietly producing clinical AI that changes real decisions.
Why watch it: Academic labs sit on decades of outcomes data and the clinical rigor to validate what the models claim.
Every lab on this list wins on the same thing, and it is not model size. It is proprietary data plus clinical validation. As I argued in my Boring Moat piece, when intelligence becomes rentable by the token, the moat moves to the data and the workflow around it.
These labs own datasets and clinical relationships that no amount of compute can conjure from scratch. That is why they will shape the next decade of medicine, whoever wins the model race.
Which lab are you most excited or most nervous about? Subscribe free to The HealthTech Signal and I will keep decoding what they ship.
HealthTech news, AI, funding and research. Straight to your inbox. Free.