YC data: Hard tech, solo founders, and the new revenue speed limit.

The data. YC's latest batch analysis reveals three seismic shifts: hard tech companies now make up 20% of accepted startups, up from 8%. Solo founder acceptance has jumped from 5% to nearly 20%. Median batch revenue at demo day has climbed from $8K to $20K MRR, with some companies hitting seven figures within three months—a pace that previously took 18 months. These aren't hypotheticals; they're the actual numbers from the last 12-18 months of YC batches.

The hard tech surge. Robotics, defense, semiconductor, and power infrastructure companies now dominate. Robotics went from 1% to 7% of the batch, industrial manufacturing from 4% to 10%, defense from 1.5% to 5%. One in six founders in the current batch holds a PhD. The driver: AI coding agents let small teams build full-stack hardware-software systems that previously required hundreds of engineers. Companies like Exosat (sovereign Starlink) and Nine Mothers (anti-drone defense) are securing seven-figure contracts during the batch.

The solo founder shift. The biggest surprise is solo founder acceptance quadrupling. The partners argue that AI tools have collapsed the need for a technical co-founder—one person can now prompt, build, and sell. But they caution that successful solo founders almost always add co-founders after getting traction. The trend isn't that co-founders don't matter; it's that you can get to product-market fit before needing one.

The revenue acceleration. The median company now generates $20K MRR by demo day versus $8K historically. The cause: AI agents let startups build end-to-end products that automate entire jobs, not just point solutions. The batch share of companies doing full workflow automation went from 10% to 25%. Examples include Juicebox (AI recruiting) evolving from candidate search to full outreach and scheduling, tripling revenue per customer. These aren't AI hype purchases—customers pay because the product actually does the job.

The unsung data economy. A hidden mega-trend: YC has funded over a dozen companies doing $10M+ annually selling data or RL environments to AI labs, with some reaching hundreds of millions. The big labs reportedly spend over a billion dollars on this. Companies like Scale and Mosaic paved the way, but now a whole ecosystem serves model training across robotics, finance, and defense. These companies stay stealthy—they don't want competitors knowing their goldmine.