Three questions we want founders to answer.
Each one maps to companies we already back, and each is still wide open. If you’re chasing one, we want to talk.
Physical AI
Why can AI reason about the world but still not load a dishwasher?
A robot learns only as fast as something can grade it. We back the layer that closes that gap: world models and simulators that score a policy before it touches hardware, hands dexterous enough to act on the score, data that transfers across bodies, and AI that runs heavy industry. We want the shortest path from a capable model to a machine that ships.
AI Infrastructure
What gets built when tokens are cheap and the user is a machine?
Every agent, model, and robot is priced in tokens, and every token is priced in silicon and power. We back the teams bending that cost curve and the stack above it: inference on any chip, power for private data centers, models with memory, and the backend, context, runtime, simulation, and trust layer an autonomous workforce runs on.
Science & Health
What compounds when every experiment is also a training run?
The largest prizes in AI sit in biology and health. We want enzymes that turn waste into feedstock, human tissue printed for drug testing, research packaged so a machine can rerun and verify it, and care agents for people the system underserves. Show us the system where doing the work is how it learns.
The next billion users
may not be people.