Stealth Startup
AI science startup aiming to gather scientific data to train models and accelerate scientific discovery
The founders
KW
The company
This stealth startup is building AI for science — specifically, gathering scientific data at scale to train models aimed at accelerating discovery across health and enterprise. Founded in 2026 and based in Portola Valley, it's tackling one of the harder problems in applied AI: the data itself.
General-purpose models have raced ahead, but scientific progress is still bottlenecked by the scarcity and quality of usable data. The bet here is that whoever owns the data pipeline for science shapes what AI can actually discover.
Details remain thin — the company is in stealth — but its stated aim is to close the gap between raw scientific data and models capable of doing useful research. The company says the focus is on gathering data to train models rather than just fine-tuning existing ones.
The startup is founded by Kevin Weil and has already secured $150M in funding, a sizeable war chest for a company this young. That level of backing suggests investors are betting on the founding team's ability to execute on an ambitious, capital-intensive plan.
Who else is building in this space
Companies working on the same thing, at every stage.
Unicorn
Partners with healthcare organizations to provide generative AI agents that answer phone calls, reducing costs and boosting appointment book
Developing AI models for drug discovery, including tools to identify new uses for existing and failed medicines
Unicorn
AI models for biology and materials science research
Scale-up
Tennr enables healthcare practices of all sizes to optimize their back-office operations with enterprise quality automations
Scale-up
Secure enterprise AI infrastructure for transforming healthcare systems.
Scale-up
Ambient AI assistant for automated clinical documentation
Thousands more founded every month.
Dealroom tracks new companies as they appear — founders, funding and signals, over the API or in the platform.