BioStack raises $9.5m seed to sell hospital data as training sets for medical AI

BioStack, a San Francisco health data start-up that came through Y Combinator, announced $9.5 million in seed funding on 22 October 2026 ¹. The investors named are Peak XV, Afore Capital, DCVC, Emerson Collective, SV Angel, Verdict Capital, Streamlined Ventures and Y Combinator ¹.

What BioStack is building

The company's model has two sides. For hospitals, it builds workflows that pull together notes, labs, imaging, medications, treatments and outcomes held in separate systems, set up around each provider's own protocols ¹. From the de-identified data it collects from each provider, it then assembles proprietary datasets that it offers to AI developers and biotech companies for model training and evaluation, research and drug development ¹.

BioStack says its network now covers more than 100 million de-identified patient records across the US, India, Brazil and Argentina ¹. It reports having combined several hundred thousand X-rays, MRIs and CT scans with pathology and clinical records into one imaging dataset, which it pitches for training AI agents that help clinicians weigh treatment options and likely outcomes ¹. The release also describes reinforcement learning environments for medical AI agents, though it gives no detail on them ¹. In India, the company says it is helping link cancer patients to an investigational cancer vaccine programme ¹.

The seed money is earmarked for expansion in the US and abroad, platform development and hiring in engineering and research ¹.

What it means for the market

The investment case is about data supply rather than models. Pushpak Kedia, vice president at Peak XV, said the next phase of healthcare AI “will depend not only on better models, but on access to high-quality, longitudinal data that reflects how patients are treated over time.” ¹

For buyers of training data, including AI developers and drug developers, BioStack adds a supplier whose records come largely from outside the US and Europe. The company presents the spread of its hospitals as access to a diverse patient population ¹. For hospitals, the offer is integration work in exchange for de-identified data flowing to commercial datasets. The release does not say how hospitals are paid, how consent is handled or what share of value they keep ¹.

The list of target uses is broad: model development and evaluation, clinical research, disease modelling, biomarker discovery, patient stratification and treatment-response analysis ¹.

Caveats

All of the figures come from the company's own announcement and have not been independently verified. BioStack does not name its hospital partners or customers, give revenue, or say how many of its 100 million records are linked across imaging, labs and outcomes rather than held as single entries ¹. The release does not set out which data protection regimes apply in each of the four countries or how the de-identification is checked ¹. At $9.5 million, this is an early round, and the company says the money will fund expansion across the US and international markets ¹.