What Y Combinator funds in 2026: data from every recent batch

1,378 companies across eight batches, compared with 1,357 from 2021 and 2022: what share mention AI, which industries grew, and where the companies are.

By , Founder, SproutPublished 8 min read

Key takeaways

  • 68% of the 1,378 companies in YC's Winter 2025 to Fall 2026 batches describe their product using AI, against 30% of the 1,357 companies in the 2021 and 2022 batches.
  • 30% of 2025 and 2026 YC companies describe an AI agent or agentic product, up from 7% in 2021 and 2022.
  • Industrials (robotics, manufacturing, hard tech) rose from 6% to 15% of YC companies between 2021-22 and 2025-26, while fintech fell from 18% to 7%.
  • 68% of 2025 and 2026 YC companies list San Francisco as a location, compared with 28% of the 2021 and 2022 batches.

Across the eight Y Combinator batches from Winter 2025 to Fall 2026, 68% of the 1,378 listed companies describe themselves using AI, against 30% in the 2021 and 2022 batches. 61% are B2B, 68% list San Francisco as a location, and industrial and robotics companies have more than doubled their share while fintech and consumer have shrunk.

Those numbers come from our own count of YC's public company directory, taken on 14 October 2026. Below are the per-batch figures, the industry shift since 2021, the caveats that matter, and what any of it means if you are deciding what to build. You can browse the same companies in our free YC directory.

Every batch from Winter 2025 to Fall 2026

BatchCompaniesMention AIMention agentsB2BIndustrialsLists San Francisco
Winter 202516666%25%63%11%62%
Spring 202514467%33%68%10%58%
Summer 202516679%33%68%9%68%
Fall 202514574%34%62%10%63%
Winter 202619869%30%64%14%67%
Spring 202619367%35%58%13%69%
Summer 202623064%26%51%25%80%
Fall 202613653%27%54%24%69%
All eight batches1,37868%30%61%15%68%

Snapshot of 14 October 2026. "Mention AI" and "Mention agents" are text matches on each company's own one-liner and description; see the methodology below for the exact definitions. Industry is the single primary industry YC lists for each company.

AI is the default, not a category

In the 2021 and 2022 batches, 30% of companies described their product in AI terms. In 2025 and 2026 it is 68%, and no recent batch is below 53%. The rise was already under way inside the earlier period: 20% of Winter 2021 companies mentioned AI, rising to 46% by Summer 2022.

The more telling number is agents. 30% of recent companies describe an AI agent, an agentic workflow, or a voice or coding agent, against 7% in 2021 and 2022. Agents for sales, support, coding, finance operations and healthcare administration now sit across every industry, which is why the AI industry page overlaps heavily with the B2B and developer tools pages.

For a founder, the practical reading is that "uses AI" no longer differentiates an idea in front of YC or anyone else. What is being judged is the problem, the customer, and why this team gets there first.

How the industry mix has shifted since 2021

Primary industry2021 and 2022 batches2025 and 2026 batchesChange
B2B47%61%+14 points
Industrials6%15%+9 points
Healthcare13%8%-5 points
Fintech18%7%-11 points
Consumer10%6%-4 points
Real estate and construction3%2%-1 point
Education3%1%-2 points

2021-22 is Winter 2021, Summer 2021, Winter 2022 and Summer 2022 (1,357 companies). 2025-26 is the eight batches above (1,378 companies). Shares are of each period's total.

What grew, and what shrank

Industrials is the fastest riser. It went from 6% to 15% of companies overall, and to a quarter of the two most recent batches (25% of Summer 2026, 24% of Fall 2026). Companies in the manufacturing and robotics subindustry, or tagged robotics, went from 1.8% to 8.6%, nearly five times the share.

Fintech fell hardest. It was 18% of the 2021 and 2022 batches, peaking at 22% in Winter 2022, and is 7% now. That is a shift in mix, not a closed door: there are still 100 fintech companies in the recent batches, and the fintech directory lists them.

Consumer and healthcare shrank as a share. Consumer went from 10% to 6% and healthcare from 13% to 8%. Both still have funded companies in every recent batch. See the consumer and healthcare lists for what did get in.

B2B is now three companies in five. Within it, the largest named subindustries are engineering, product and design tools (11% of all recent companies) and infrastructure (8%).

Where the companies are, and how big

Measure2021 and 2022 batches2025 and 2026 batches
Lists San Francisco as a location28%68%
Lists the United States as a region57%85%
Median team size today103
Still active74%98%
Acquired10%1%
Inactive16%1%

Team size and status are as of the snapshot date, not at the time of the batch, so older companies have had four or more years to grow, sell or shut down. Compare the recent column with itself, not with the older one, for those three rows.

The location shift is the largest single change in the data. In 2021 and 2022, 28% of companies listed San Francisco. In 2025 and 2026 it is 68%, and 80% of the Summer 2026 batch. 85% of recent companies list the United States as a region, against 57% before.

The median recent company has three people. That is a reminder that a team of two or three is the norm in a batch rather than a weakness.

What this means if you are choosing an idea

A category being popular in recent batches is not a reason to avoid it. Several funded companies working on the same problem is evidence that investors believe the budget is real. The useful question is whether you have a narrower wedge than the ones already there, which is what competitor analysis for a startup idea is for.

A category shrinking is not a reason to avoid it either. Fintech at 7% still means about one company in fourteen, and a smaller share means fewer funded companies chasing the same customers. What matters is whether your idea answers the question every category gets asked: why now, and why you.

The most direct use of this data is to find your closest comparables and read how they describe themselves. Benchmarking a startup idea against YC companies walks through doing that properly, including how to read a comparable that failed.

Methodology

Source. The yc-oss API, an open mirror of the Y Combinator company directory, downloaded on 14 October 2026. It held 6,297 companies. We counted every company listed in the Winter, Spring, Summer and Fall 2025 and 2026 batches (1,378) and, for comparison, the Winter 2021, Summer 2021, Winter 2022 and Summer 2022 batches (1,357).

Mention AI means the company's one-liner or long description contains "AI" as a word (case-sensitive, so "said" or "aim" do not count), "artificial intelligence", "LLM", "large language model", "machine learning" or "GPT". Mention agents means it contains "AI agent", "agentic", "autonomous agent", "voice agent", "coding agent", or "agent" followed by "that", "which", "for" or "to". We used description text rather than YC's tags because the tags are missing for most of the newest companies: 146 of 198 Winter 2026 companies and 137 of 193 Spring 2026 companies had no tags at the snapshot date. Among companies that do have tags, 74% of 2025 and 2026 companies carry an AI-related tag, against 29% in 2021 and 2022, which agrees with the text measure.

Industry is the single primary industry the directory gives each company. Location matches "San Francisco" in the listed locations; region uses the directory's region field. Team size is the directory's current figure; companies with no team size were left out of the median.

Limitations. Descriptions and tags are written by the companies themselves, so "mentions AI" measures how a company presents itself, which may overstate how central AI is to the product. The directory lists funded companies, not applications, so nothing here says anything about acceptance rates or what YC turns down. The newest batches are the least complete: listings, tags and descriptions fill in over the months after a batch, so expect their figures to move.

Frequently asked questions

What percentage of Y Combinator companies are AI companies?
68% of the 1,378 companies in YC's batches from Winter 2025 to Fall 2026 describe their product using AI, by a text match on their own directory descriptions taken on 14 October 2026. In the 2021 and 2022 batches the same measure was 30%.
How many companies are in a Y Combinator batch?
Between 136 and 230 in the eight batches from Winter 2025 to Fall 2026, as listed in YC's directory on 14 October 2026. Summer 2026 was the largest at 230 and Fall 2026 the smallest at 136. The 2021 and 2022 batches were larger, up to 398 in Winter 2022.
How many Y Combinator batches are there a year?
YC's directory lists four batches in each of 2025 and 2026: Winter, Spring, Summer and Fall. Together the eight batches from Winter 2025 to Fall 2026 hold 1,378 companies.
What industries does Y Combinator fund most in 2026?
B2B software is 61% of the companies in YC's 2025 and 2026 batches, followed by industrials at 15%, healthcare at 8%, fintech at 7% and consumer at 6%. Industrials, which includes robotics and manufacturing, rose from 6% in 2021 and 2022 and reached about a quarter of the Summer and Fall 2026 batches.
Does Y Combinator still fund fintech and consumer startups?
Yes. Fintech is 7% of 2025 and 2026 YC companies, 100 companies, and consumer is 6%. Both are smaller shares than in 2021 and 2022, when fintech was 18% and consumer 10%, but neither category has disappeared.
Do most Y Combinator startups move to San Francisco?
68% of companies in YC's 2025 and 2026 batches list San Francisco as a location, and 85% list the United States as a region. In the 2021 and 2022 batches those figures were 28% and 57%.
How big is a typical Y Combinator startup team?
The median company in YC's 2025 and 2026 batches had three people as of 14 October 2026. A team of two or three is the norm in a recent batch.

Sources

Written by

Usman founded Sprout, the product studio that builds Truewick, and leads its engineering as CTO. Usman is also a founding engineer at Kollegio AI, a Stanford-backed college counseling platform serving more than 300,000 students, and studied Computer Science at LUMS.