Y Combinator's 15 Most Wanted Startup Ideas for Summer 2026

What is Y Combinator's Request for Startups?

Y Combinator's Request for Startups (RFS) is their annual list of startup ideas they're actively looking to fund. When YC partners publish an RFS, it means they've identified massive market opportunities where they believe extraordinary companies can be built.

Unlike most accelerator programs that passively wait for applications, YC's RFS signals: "If you're working on this, we want to talk to you."

1. AI for Low-Pesticide Agriculture

Partner: Garry Tan

Modern agriculture relies heavily on synthetic pesticides and herbicides. YC wants founders building AI vision systems that can identify weeds and pests in real-time, combined with precision robotics to eliminate them without chemicals.

The Opportunity:

• Computer vision to detect weeds pixel-by-pixel in crop fields
• Precision robotics for targeted elimination (lasers, mechanical removal)
• Biological alternatives to synthetic pesticides
• Goal: Cut pesticide use by 90% while increasing yields

Why now? Advances in edge AI, computer vision, and robotics make precision agriculture economically viable. Regulatory pressure on chemical pesticides creates strong market pull.

2. AI-Native Discovery Engines

Partner: Jon Xu

YC wants closed-loop discovery systems in drug development, materials science, and protein engineering. Not research copilots—full autonomous systems.

How It Works:

1. AI proposes candidate molecules based on desired properties
2. Automated labs synthesize and test the candidates
3. Results feed back into the model, which learns and proposes better candidates
4. Repeat at 100x the speed of human-led research

3. AI-Native Service Companies

Partner: Gustaf Alströmer

Instead of building software to help accountants work faster, replace accountants entirely with AI agents. YC calls this the biggest opportunity in startups today.

Target Markets:

• Insurance brokerage: AI agents quote, compare, and purchase policies
• Accounting & tax: Automated bookkeeping, tax filing, and audit preparation
• Legal compliance: Regulatory filing and corporate governance
• Healthcare administration: Medical billing, prior authorizations, and claims

Market size: Service industries are 10-100x larger than software markets. A B2B SaaS tool for accountants might reach $100M ARR. An AI service that replaces accountants could reach $10B.

4. AI Personalized Medicine

Partner: Ankit Gupta

Genomic sequencing costs have plummeted from $3 billion to under $100. YC wants startups using this data with AI agents to provide personalized health recommendations—and even n-of-1 genetic therapies tailored to individual patients.

The Vision:

• AI analyzes your genome, microbiome, and health records
• Provides lifestyle and medication recommendations specific to you
• Designs mRNA therapies targeting your specific genetic variants
• Continuous monitoring adjusts recommendations as new data arrives

5. Company Brain & AI Operating Systems

Partners: Tom Blomfield, Diana Hu

Enterprise knowledge is scattered across Slack, email, Linear, GitHub, Google Docs, and tribal knowledge. YC wants systems that extract this into "an executable skills file for AI"—making companies queryable and automatable.

Two Approaches:

6. Counter-Swarm Defense

Partner: Tyler Bosmeny

Coordinated drone swarms represent the next generation of warfare. YC wants defenses that can handle hundreds of simultaneous threats.

Defense Strategies:

• High-capacity interceptors: Take down multiple drones per defensive unit
• Sensor fusion: Track and predict swarm behavior in real-time
• Non-kinetic defenses: Jamming, spoofing, directed energy
• Autonomy attacks: Disrupt swarm coordination algorithms

Think: "Cloudflare rather than Raytheon"—software-defined defenses that scale infinitely.

7. Space Electronics & Moon Manufacturing

Partners: Philip Johnston, Adi Oltean

Electronics in Space:

Design inference chips optimized for space deployment—radiation-hardened, thermally efficient, and mass-optimized for launch costs.

Industrial Manufacturing on the Moon:

Extract silicon, aluminum, iron, and titanium from lunar regolith using electrolysis. 3D print structures and components directly on the moon, eliminating launch costs.

8. Inference Chips for AI Agents

Partner: Diana Hu

Current GPUs are designed for training, not agentic workflows. YC wants purpose-built silicon for AI agents that need:

• Fast context switching between multiple models
• Native speculative decoding for faster inference
• Memory architectures optimized for KV caches
• Multi-model orchestration on a single chip

9. AI-Native SaaS Challengers

Partner: Jared Friedman

Rebuild legacy software from the ground up with AI. YC specifically calls out:

• Chip design software (EDA tools)
• Enterprise Resource Planning (SAP, Oracle)
• Industrial control systems (SCADA, PLCs)
• Software for agents with machine-readable APIs instead of visual UIs

Key insight: Legacy software has visual UIs because humans use it. AI agents need APIs, MCPs, and CLIs—enabling them to discover and use new tools instantly.

10. How to Apply to Y Combinator

If you're building in one of these areas, here's how to maximize your chances:

Before You Apply:

1. Build a prototype: Even a rough demo shows execution ability
2. Talk to 10+ potential customers: Validate the problem exists
3. Find a co-founder: YC rarely funds solo founders
4. Study the RFS partner: Read their essays and understand their thesis

The Application:

• Be specific: "AI for agriculture" is too broad. "Computer vision for weed detection in corn fields" is better
• Show traction: Even 5 paying beta customers matter
• Explain why you: Domain expertise, technical skills, or unique insights
• Reference the RFS: Explicitly mention you're building in a requested category

The Common Thread

Looking across all 15 RFS categories, a clear pattern emerges: YC wants founders replacing entire industries with AI, not just making them more efficient.

Whether it's agriculture without pesticides, drug discovery without human researchers, or companies without traditional org charts, the opportunity is to rebuild from first principles now that AI makes the impossible possible.

The best time to start was yesterday. The second best time is today.