The Y Combinator Request for Startups, from Summer 2026: 16 ideas YC’s partners want founders to build. Each links to a problem page with more detail, related problems and people working on it.
Build industrial capacity on the Moon and in space, from extracting silicon, aluminum, iron and titanium to 3D printing structures from lunar regolith.
Coding agents could let every user reshape the software they use, the way forward-deployed engineers customize enterprise tools. YC wants the new delivery stack that makes this possible.
Reusable rockets will put far more in orbit, creating demand for compute in space. The ask is inference chips tuned for mass, heat and radiation.
A Shenzhen team can turn a design into a physical part within a day; US teams often wait weeks. YC wants startups that make US hardware iteration an order of magnitude faster.
Agents loop through tool calls, branches and long contexts, and GPUs sit 30–40% utilized on that work. YC wants silicon and compilers designed for the agent loop.
AI has cut the cost of building software 10–100x, eroding legacy SaaS's code moats. YC wants challengers, from cheaper clones to AI-native suites aimed at categories that seemed untouchable.
AI agents are becoming major users of the internet but run on software built for people clicking buttons. YC wants agent-first software: APIs, MCPs, CLIs and docs agents can use without a human.
AI has made the world's largest companies reachable, eager buyers for small teams. YC wants more startups going after Fortune 100 deals from the start.
An advanced AI chip passes through about 1,400 steps across a dozen countries, managed with spreadsheets and phone calls. YC wants real-time supply chain software built for chips.
The best AI-native teams make their whole company queryable so an intelligence layer can learn from it. YC wants the product that connects all that context by default.
Instead of selling software or copilots, AI-native companies can sell the service itself, doing the work in areas like insurance brokerage, accounting, compliance and healthcare admin.
Cheap drone swarms give attackers the cost advantage, and today's patchwork defenses aren't built for them. YC wants a counter-swarm stack that works like a real-time distributed system.
AI agents can't automate a company whose know-how lives in people's heads and old Slack threads. YC wants a 'company brain' that turns scattered knowledge into skills AI can execute.
Cheaper diagnostics, AI agents that read personal health data and falling costs for one-patient genetic therapies could make truly personalized care possible.
Science still runs on a slow hypothesize-experiment-interpret loop. YC wants AI-native systems that run that loop end to end alongside researchers, not just research copilots.
Farmers are spraying more chemicals for diminishing returns as weeds and pests adapt. AI vision, cheap sensors, precision robotics and new biology could cut pesticide use while raising yields.