Most investors wait for Demo Day. We read the batch weeks earlier — before pitches, before the hype, before the pricing event. The batch-level statistics aren't just trivia. They're the first signal of where the best companies are hiding.
Here's how we read the W26 numbers — and what they told us about where to focus.
38% Repeat Founders: The Quality Signal
When 38% of a YC batch are repeat founders, that tells you something important before you've spoken to a single company: the batch quality ceiling is unusually high.
Repeat founders know what they're getting into. They've built teams before. They've navigated board dynamics, hiring misfires, customer churn, and pivots. More importantly, they come with enterprise relationships that first-time founders spend years building. When a repeat founder says 'I have three enterprises lined up for pilots,' it's usually true.
Our W26 portfolio reflects this. Cignara's founder is a two-time YC founder — he knew exactly how to sign Fortune 500 clients (LG was already on board before the raise). Standout's Alexis Aftalion had already scaled Zealy to 1.5 million users. These aren't first-time founders learning on the job.
For a pre-Demo Day investor, a high repeat-founder rate is a prioritization signal: spend more time on these companies first, because their hit rate is meaningfully higher.
65% B2B: The Enterprise AI Wave Is Here
Two-thirds of the W26 batch is B2B. That's not a trend — it's a structural shift. The consumer AI wave produced ChatGPT, Midjourney, and a handful of breakouts. The enterprise AI wave is still in the early innings, and that's where W26 is playing.
What does 65% B2B look like in practice? It looks like eight years of SaaS tooling being rebuilt as AI agents. It looks like enterprise workflows — clinical trials, manufacturing validation, QA testing, ERP, hiring — that have been underserved by software for decades suddenly getting their first genuinely intelligent layer.
The opportunity in B2B AI isn't the same as the opportunity in consumer AI. Enterprise buyers are slower, but they're also stickier. Once a hospital system adopts Astraea for clinical trial data, they don't switch. Once a manufacturer adopts Arzana for ERP automation, the switching cost is enormous. Enterprise revenue compounds in a way that consumer revenue doesn't.
When we see 65% B2B, we know to look hard at the enterprise validation for each company. Not 'are they targeting enterprise?' but 'have enterprise customers already paid?' That's the filter that separates the real B2B companies from the ones with enterprise aspirations.
14% Industrial: The Surprise Signal
This one surprised us. 14% of the W26 batch is building for physical-world industries — manufacturing, construction, agriculture, logistics, energy. That's not a historically YC-heavy sector.
The reason it's happening now: the cost curve for deploying AI into physical environments has finally crossed a threshold. Sensors are cheap. Vision models are good enough. Edge compute is accessible. The combination has opened a window for software to enter industrial operations in a way that wasn't possible three years ago.
Our W26 industrial picks — Prototyping.io (autonomous manufacturing validation) and Arzana (autonomous ERP for manufacturers) — both came from this signal. We saw 14% industrial in the batch composition and immediately asked: which of these companies has already landed the hardest enterprise customers in the world? Prototyping.io had Tesla. That was enough.
Industrial is a space where founder-market fit matters more than almost anywhere else. You can't fake your way through a conversation with a manufacturing plant manager. Either you understand the floor or you don't. When we see industrial companies from YC, we look hard at whether the founders have actually spent time in the environments they're building for.
How We Use Batch Stats to Prioritize
Batch statistics don't tell us which companies to invest in. They tell us where to focus our time first. A batch with 38% repeat founders means we spend the first two weeks on repeat founders. A 65% B2B mix means we screen for enterprise validation before anything else. A 14% industrial signal means we look for the one or two industrial companies that have already cracked their hardest customer.
The goal is to arrive at Demo Day having already done the work. By the time the crowd shows up, we've already spoken to 60+ companies, run deep diligence on 25, and written checks to the ones that passed every test. The batch stats are the map. The companies are the territory.
