YC signals where to build next

Y Combinator just did one of its oldest signaling tricks again. It updated its Requests for Startups for Summer 2026 right as the current batch application deadline approaches — May 4 at 8 p.m. PT, with decisions for on-time applicants due by June 5. That matters because founders treat these lists less like commandments and more like a map of what YC partners are itching to fund next. And this year the map is pretty clear: AI is no longer the feature. It’s the company, the workflow, and increasingly the physical system around it. (ycombinator.com) ### What did YC actually publish? The Summer 2026 RFS page is YC’s official list of startup ideas it wants to see. The intro says AI has “stopped being a feature and started being the foundation,” and frames the opportunity as rebuilding software, services, and silicon while pushing AI into the physical world. That wording is the headline, really —(ycombinator.com)wants companies built around the assumption that capable models are the starting point. (ycombinator.com) ### Why do founders care so much? Because RFS pages are one of the cleanest public windows into partner taste. YC has been doing this since 2009, and it has long described the lists as ideas it wants made real, not a box founders must squeeze into. Sam Altman made that explicit years ago too — these prompts are there to get people thinking, and YC (ycombinator.com)if you are deciding between three adjacent ideas, the RFS tells you where the internal excitement probably is. (ycombinator.com) ### So what themes jump out? First, AI-native service companies. Gustaf Alströmer’s section says the last wave of startups mostly built copilots that helped people do jobs, while the next wave should just do the work and sell the service itself. That is a huge shift. It pushes founders away from seat-based SaaS and toward (ycombinator.com)olved. YC’s point is simple: services spending is much bigger than software spending, and outsourced work is easier to replace than deeply embedded internal workflows. (ycombinator.com) ### Why is physical-world AI all over this? Because YC thinks models have gotten good enough to leave the browser. The clearest example on the page is low-pesticide agriculture. Garry Tan’s prompt argues that cheap sensors, better vision models, robotics, and newer biological tools can finally let startups treat individual plants and pests precisel(ycombinator.com)tandout number is a 90% reduction in pesticide use while increasing yields — the kind of claim YC thinks could create a “generational company.” That is not a casual niche bet. It is AI applied to one of the world’s biggest real-economy markets. (ycombinator.com) ### Where does infrastructure fit? Right underneath the AI talk, YC is still leaning into hard tech. Its separate techno-industrialist push calls for startups applying modern software to energy, materials, CAD, manufacturing, and biotech. In other words, YC is not choosing between software and industry. It is betting the next strong companies sit a(ycombinator.com) systems get designed and run. (ycombinator.com) ### Is this only an AI story? Mostly, but not entirely. Earlier 2025 and Spring 2026 YC requests also pushed into onchain finance, modern industrial systems, and new financial primitives. That broader pattern matters because it shows the firm is still interested in platform shifts beyond chat interfaces. But even there, the center (ycombinator.com)ct building, services, and industrial operations. (ycombinator.com) ### What should founders take from it? Don’t copy the list literally. That usually produces dead-on-arrival “YC bait.” The better read is narrower: YC is telling founders where it believes new technical leverage has appeared. If your idea turns recent model gains into a full product or service, especially in a giant market with ugly manual wor(ycombinator.com)YC wants to hear right now. (ycombinator.com) ### Bottom line The Summer 2026 RFS is basically YC saying the easy copilot era is crowded. The next opening is bigger — AI-run services, AI-driven industrial systems, and software that escapes the screen. (ycombinator.com)