UFO Capital backs pre-seed founders building the picks-and-shovels of the AI era — compute, data, retrieval, eval, and the boring layers that compound.
Every generational platform shift is won at the substrate. The browser made HTTP boring; the cloud made racks invisible. The AI shift will be no different — the moats are forming underneath the chat windows.
We write first checks of $250K–$2M into pre-seed teams building the runtimes, retrieval engines, evaluation harnesses, data pipelines, and inference economics that the next thousand AI companies will quietly depend on.
We are concentrated, technical, and unhurried. We take three years to write thirty checks. We answer pitches in 48 hours. We do not lead by FOMO.
We meet you before the deck is finished. No bridge rounds, no pro-ratas we won’t earn — we lead the round that gets the first hire on payroll.
Our partners shipped systems at OpenAI, Modal, Databricks, and CoreWeave. We read your benchmarks. We notice the FLOP per dollar.
Infrastructure compounds slowly, then suddenly. Our LPs are endowments and family offices that measure in decades, not quarters.
UFO led our pre-seed before we had a domain. They had read my dissertation. They asked sharper questions than my advisor.
The memo they wrote on inference economics is required reading on our team. It’s why we took their check over a fund three times the size.
A small fund, a long horizon, and the technical taste of a research lab. That’s a rare combination at pre-seed.
A working prototype. A benchmark we’ve never seen. A weird research direction. We answer every pitch within 48 hours — yes or no, with a reason.
