Y Combinator W26 Batch: The Best Startups From the Latest Class

Fourteen of the nearly 190 startups in Y Combinator's W26 batch had already passed $1 million in annualized revenue before Demo Day on March 25, 2026 — three times the prior batch and the most in YC history. That's the short answer for why people are calling W26 the strongest class YC has ever run. The longer answer is more interesting.

I've been investing in early-stage companies for over a decade and have written more than 65 checks, and I read every YC batch the same way: ignore the hype reel and look at what the best companies actually shipped before Demo Day. W26 is the rare batch where the data backs the hype. Average week-on-week revenue growth across the cohort hit 14% — the fastest YC has ever measured — and Rebel Fund's machine-learning batch score put 35% of W26 companies in the top 20% of all YC startups ever evaluated. No previous batch has come close. Here are the best startups in it and why this class is different.

The Best Startups From the Y Combinator W26 Batch

The best startups in Y Combinator's W26 batch include Asimov (real-world movement video to train humanoid robots), Condor Energy (an operating system for reliable, sustainable electricity), Hex Security (an agentic offensive-security platform), Sazabi (log analysis that auto-generates and ships production fixes), and Silmaril (prompt-injection defense for AI agents). Out of nearly 190 companies, 14 had already crossed $1M in annualized revenue before Demo Day on March 25, 2026.

TechCrunch named 16 of the most-interesting W26 companies, and a separate VC-sourced list later crowned a top eight. The overlap is telling: the companies that show up on both lists are overwhelmingly technical, B2B, and selling into hard, regulated, or infrastructure-heavy markets. This is the table I keep open when founders ask me what "the best of W26" actually means.

Company descriptions compiled from TechCrunch's "16 most interesting startups from YC W26 Demo Day" (March 26, 2026), TechCrunch's VC-sourced standout list (June 18, 2026), and the official YC W26 launch directory. Sectors are author classifications.

How the W26 Batch Became the Strongest in YC History

YC president Garry Tan called W26 the fastest-growing cohort the firm has ever run, and the metrics support it rather than the other way around. Three independent signals — revenue, growth rate, and an outside quality model — all moved in the same direction at once. That alignment is what makes this batch worth paying attention to beyond the usual Demo Day noise.

Figures are 2026 reporting blended from Garry Tan / YC Demo Day disclosures, TechCrunch, and Rebel Fund's published batch-scoring model (which has rated every YC Demo Day since 2013). Revenue figures are annualized run-rate self-reported at Demo Day.

The number I weight most is Rebel Fund's 35%. Revenue at Demo Day is partly a function of how YC coaches founders to present run-rate, and a $1M ARR figure built on three enterprise pilots can evaporate. But a third-party model trained on thirteen years of outcomes flagging a third of the batch as top-quintile is a harder signal to game. It says the underlying company quality — team, market, traction shape — is genuinely higher, not just better packaged.

Y Combinator W26 Startups by Sector: Where the Best Companies Cluster

The best Y Combinator W26 startups cluster in four places, and none of them is the consumer app. At 64% B2B and roughly 5% consumer, W26 is the most enterprise-weighted batch I can remember. The standouts split cleanly into agentic developer tooling, AI security, physical-world AI, and infrastructure.

Agentic dev tooling and ops is the densest cluster — Sazabi auto-fixing production incidents, Polymath automating reinforcement-learning environment generation, Jinba running enterprise workflows through plain-language chat. These sell into budgets that already exist, which is why several hit $1M ARR fastest. AI security is the second wave: Silmaril and Hex Security exist only because agents are now in production and getting attacked. That's a market that did not meaningfully exist two batches ago.

The third cluster is physical-world AI — Asimov on humanoid training data, Milliray on drone-detection radar, Avoice automating contract and drawing review. The fourth is hard infrastructure: Condor Energy on the grid and GRU Space on lunar logistics. Roughly 60% of the batch builds AI, but the winners apply it to a specific, expensive, previously-manual workflow rather than shipping another general chat wrapper. If you track where billion-dollar outcomes come from, our unicorn tracker shows the same pattern: durable companies own a workflow, not a model.

How to Evaluate the Best YC W26 Startups as an Investor

A record batch is also a dangerous batch, because price follows quality. At least two W26 companies raised at $175M+ post-money and the top reached $200M — for a company a few months out of a $125K check. When the median seed across the broader market sits closer to $15–25M post-money, those W26 rounds are priced for a level of certainty no Demo Day deck can earn. The batch being strong does not make every round in it a good entry point.

The way I'd underwrite a W26 standout is to separate the three signals the batch is famous for. Revenue run-rate tells you a customer paid once, not that they'll renew — ask for net revenue retention and the count of logos behind the ARR. The 14% weekly growth is a batch average over a short window; the question is whether it survives the post-Demo-Day cliff once YC's hothouse cadence ends. And the Rebel Fund score is a probability, not a guarantee — top-quintile still means most of those companies will not return a fund.

The practical filter: back the W26 companies selling into a budget that already exists (security, dev tooling, legal, energy) over the ones inventing a category, and discount any valuation that assumes the growth rate is permanent. You can pressure-test any of these against current seed and Series A benchmarks on our funding dashboard and AI-specific pricing on the AI valuations page. A strong batch rewards discipline more than enthusiasm — the best entry on the best company at the wrong price is still a bad investment.