Back

Y-Combinator’s Fall 2025 “Requests for Startups”: Where They Want to Fund Next YC just dropped its latest list of high-impact startup ideas they’re eager to back. If you’re building in any of these spaces, YC wants to hear from you 👇 1️⃣ AI-Powered Vocational Training • Shortage of skilled tradespeople to build AI infra (data centers, fabs, HVAC). • Use multimodal AI (voice, vision, AR/VR) to train workers in welding, electrical, etc. 💡 “If you can build one great AI teacher, you can train thousands.” – Harj Taggar 2️⃣ Video Generation as a Primitive • Tools like Google Veo can already generate 8s photorealistic clips at near-zero cost. • Treat generative video as infrastructure: → Personalized entertainment (1:1 TikTok, family cartoons) → Virtual commerce (AI try-ons) → Gaming/training (AI for synthetic data) 🎯 The next TikTok or Netflix may be built without cameras. 3️⃣ The 10-Person, $100B Company • AI slashes headcount needs. Small, high-agency teams can now build unicorns. • Key metric: Revenue per employee. 📌 “The next unicorns will be built by 10 people using $500K and AI.” – Aaron Epstein 4️⃣ Infrastructure for Multi-Agent Systems • Single AI agents → fleets collaborating on distributed tasks. • Platforms for managing agent fleets, abstractions, prompt chaining, trust & cost. 💡 “We want to fund the Kubernetes for AI agents.” – Pete Koomen 5️⃣ AI-Native Enterprise Software • SaaS 2.0: Systems like Salesforce will be rebuilt natively on AI. • AI-first sales, HR, finance, and accounting—deeply embedded, not bolted on. 6️⃣ Replacing Government Consulting with LLMs • $100B+ market currently dominated by Deloitte/Accenture. • LLMs can automate compliance, policy checks, and regulatory workflows. 💡 “LLMs can do what consultants do, better, faster, cheaper.” – Gustaf Alstromer 🔥 YC is basically saying: If you’re building at the intersection of AI x infra, enterprise, or government, they want in. Which of these areas would you bet on?