I pulled the full YC 2026 list — 478 companies across the Winter, Spring, Summer, and Fall batches — from public data (ExploreYC plus the YC Startup Directory) and lined it up against every batch since 2021. The headline isn't "AI is hot." It's where the money quietly moved.
Six years of YC, by industry
| Year | Companies | B2B | Consumer | Industrials | Fintech | Healthcare |
|---|---|---|---|---|---|---|
| 2021 | 727 | 47% | 11% | 6% | 16% | 15% |
| 2022 | 633 | 47% | 9% | 6% | 21% | 11% |
| 2023 | 495 | 68% | 6% | 2% | 10% | 10% |
| 2024 | 592 | 63% | 10% | 7% | 7% | 10% |
| 2025 | 628 | 66% | 7% | 10% | 5% | 7% |
| 2026 | 478 | 61% | 4% | 14% | 9% | 8% |
Three lines tell the story: B2B peaked in 2023 and is slowly drifting down, Consumer has been bleeding out for six years, and Industrials climbed from a 2% trough in 2023 to 14% — the highest in the dataset.
1. Consumer AI is quietly disappearing
Only 20 of the 478 companies in YC 2026 are Consumer — 4%, down from 11% (roughly 80 companies) in 2021. It's not that consumer is dead as a market; it's that early-stage capital is buying certainty. Enterprises pay directly for labor they no longer need to hire. Consumer attention keeps getting more expensive.
2. The median team is 2 people
Team size across the batch: 34 solo founders, 205 two-person teams, 97 with three, 54 with four, and only 67 with five or more. Median: 2. Mean: ~3.1.
Exactly half the batch is two people or fewer, and the two-person team alone is 43% of it. Two founders plus coding agents now cover what used to take a ten-person team — and YC is effectively pricing that in.
3. Nearly half the batch is building AI employees
A rough keyword screen (agent / agentic / copilot / operator / assistant in name or description) matches about 226 companies — 47% of the batch. These aren't tools for people to use; they're digital workers meant to do the job.
Two sub-clusters stand out:
- Voice: ~65 companies (13.6%) matching voice / phone / call. For support, sales, scheduling, and collections, the real interface was never a chat box — it's a phone line. The next big AI entry point may not look like ChatGPT at all.
- Knowledge work: ~71 companies (14.9%) matching email / calendar / docs. If your job lives in an inbox and a calendar, 71 funded startups are working on doing it for you.
4. AI is leaving SaaS for the physical world
By official category, Industrials is 68 companies (14%). Widen the lens with a keyword screen — robot / drone / autonomous / hardware — and you get ~103 companies, 21.6% of the batch. One in five YC 2026 companies touches the physical world: robotics, defense, energy, manufacturing, farms.
The logic is simple: pure-software margins on ideas are thinning after twenty years, while the "hard" sectors are exactly where AI capability just crossed the usefulness threshold and competition is still sparse.
One more echo: Fintech recovered from 5% in 2025 to 9%, with ~37 companies (7.8%) touching stablecoins or crypto. Payment rails are being rebuilt.
Methodology and caveats
- The 478 total covers Winter/Spring/Summer/Fall 2026; the Summer and Fall batches may not be fully published yet, so percentages will shift.
- Data comes from ExploreYC and the public YC Startup Directory; there can be sync lag versus YC's live directory.
- Keyword screens are deliberately coarse — good for direction, not for precise counts. Expect both false positives and misses.
How I ran this
Every number above came from the ExploreYC Startup Research Agent, which queries and aggregates ExploreYC's public dataset. There's a longer writeup of how it works on the ecosystem page.
If you want to slice the data differently — one vertical, one batch, one team size — the agent runs in a ready-to-use agent environment such as ClawMama, so you can query the same dataset from Telegram or WhatsApp without setting anything up. The data is public; the conclusions are yours to break.
Top comments (0)