YC maps Summer 2026 requests

Y Combinator’s Summer 2026 Requests for Startups is basically a map of where one of Silicon Valley’s most influential seed investors thinks the next wave of companies will come from. Not a prediction. Not a rulebook. More like a shopping list with attitude. The big change this time is that YC is no longer mostly pointing founders toward pure software. It’s pointing them toward software plus services, software plus chips, and software plus the physical world. (ycombinator.com) ### What did YC actually publish? YC’s RFS page is a recurring feature where partners lay out startup ideas they want to fund. The Summer 2026 edition 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 real world. YC also makes the caveat explicit — these are only a fraction of what it funds, and founders do not need to follow the list to apply. (ycombinator.com) ### What feels new this time? The tone is much more industrial. Earlier AI-era startup lists were heavy on copilots, workflow tools, and software you could build from a laptop. This one still has software categories, but it also reaches into agriculture, medicine, semiconductors, defense, and space. That shift matters because it suggests YC thinks the easy “AI wrapper” phase is fading, while harder markets are opening up as models get better and implementation costs fall. (ycombinator.com) ### Which categories stand out most? A few are classic software with a twist. YC calls for AI-native service companies that sell completed work instead of tools, “Company Brain” systems that turn internal company knowledge into something agents can use, software built for agents rather than humans, SaaS challengers aimed at big incumbents, and dynamic interfaces that adapt to each user or agent. But the eye-catching part is the harder-tech set — low-pestici(ycombinator.com) defense, semiconductor supply chains, space electronics, and even lunar manufacturing. (ycombinator.com) ### Why are services such a big deal here? Because YC is arguing that AI can now do more than assist workers — it can replace chunks of outsourced labor. That changes the business model. Instead of selling software seats, a startup can sell outcomes: insurance processing, accounting work, tax prep, compliance, healthcare admin. The market for services is far larger than the market for software, and outsourced categories are especially exposed because buyers(ycombinator.com) the job. (ycombinator.com) ### Why the push into physical industries? Turns out AI is getting cheap enough and capable enough to combine with sensors, robotics, and specialized hardware. YC’s agriculture example makes the logic very clear — cameras and models can identify weeds in real time, robotics can target individual plants, and new biological tools can replace some chemical treatments. That is the broader thesis in miniature: AI is no longer just a chat box on top of (ycombinator.com)-world systems. (ycombinator.com) ### Is this a prescription for founders? Not really. It’s better read as a market map. YC itself says the list is only a slice of what it funds. So the useful move is not “pick one of the 15 and obey.” The useful move is to notice the pattern: YC wants startups where AI is the engine, but the moat comes from distribution, workflow ownership, regulated-market knowledge, or deep integration with the physical world. (ycombinator.com), the message is that “AI startup” is no longer a category by itself. If everyone has model access, the winning companies will be the ones that use AI to crack open giant markets that old SaaS barely touched — enterprise operations, industrial systems, healthcare, defense, chips, agriculture, and space. That is a much harder game. But it is also where YC seems to think the billion-dollar outcomes are moving. (super([ycombinator.com)/yc-summer-2026-rfs-hard-tech-pivot)) ### Bottom line Summer 2026’s RFS reads like YC declaring the post-copilot phase. The easy demos are not enough anymore. The interesting startups now are the ones that use AI to do real work in hard markets — and then build the company around that wedge. (ycombinator.com)