YC’s Record Breaking W26 Demo Day Recap
Why Investors Are Paying Double
Fourteen YC companies walked into Demo Day on Tuesday at $1M+ ARR. That’s 3x more than the previous batch 3 months ago, and they did it before raising capital beyond YC’s initial check.
While you’re in YC, you’re typically not hiring because you haven’t raised yet. That happens after Demo Day. Many of these companies hit $1M annualized revenue with founding team resources and 3 months of focus.
One company is at $27M ARR.
Another has $175M in letters of intent.
The average weekly growth across ~200 companies is 14%, the fastest in YC’s history.
Something fundamental has changed about what’s possible in 90 days, and understanding what changed matters because it’s reshaping how early-stage investing works.
The composition of the batch explains part of the velocity shift.
W26 saw 24 repeat founders, including 15 who’ve been through YC before.
Matt Riley from Fixture, Louis Zardi and Sam Alba from Mendral, Joseph McAllister from MouseCat.
Experienced operators who came in knowing exactly what to build and how to move fast, which compresses the learning curve that usually eats up the first month of most batches.
The technology stack explains another part.
88% of the batch is AI-first, and 56 out of 198 companies are building fully autonomous agents that do jobs without human intervention.
Three separate startups literally call themselves “Claude Code for X” because that’s the clearest explanation of what they do. This matters because AI tools collapsed the time from idea to working product. What used to take 6 months of engineering now takes 6 weeks, which means teams can get to revenue faster because they’re not stuck in build mode for the entire batch.
Likewise, even hardware can get to market quicker with 13 robotics companies deploying actual systems in 90 days because the cost of robotics parts dropped enough to make hardware iteration compatible with YC’s three-month timeline. GrazeMate has robot cowboys herding cattle with AI drones. RoboDock runs autonomous depots. Not just prototypes, but deployed products generating revenue.
The category selection explains the rest.
Healthcare is the largest vertical at 22 companies, which makes sense when you realize healthcare is 18% of the economy but only 1% of AI agent usage according to recent Anthropic data. The gap between market size and AI penetration creates obvious opportunities for teams that can navigate medical workflows and compliance.
But zoom out and you see the deeper pattern running through much of the batch: vertical AI coding agents for specific industries. Many Companies are building full-stack automation for categories where the software is clunky and/or the labor is expensive.
This is the most reliable company-building pattern in the batch, and it works for three reasons.
Vertical agents can own the full workflow, which means they capture more value per customer and create stronger lock-in.
Customers will pay for outcomes rather than tech access, which means pricing isn’t tied to AI API costs but to the value of replacing manual work or improving speed.
Domain expertise creates moats that pure AI labs won’t touch because they can’t justify building deeply vertical tools when they’re optimizing for horizontal platform scale.
The pattern scales across any category where automation was previously impossible because the judgment calls were too complex for traditional software.
However, most of the traction we’re seeing at Demo Day got built during the three months inside YC, which means execution velocity during the batch matters more than the idea you showed up with.
So investors need to build early relationships to measure how fast teams can identify the right workflow to automate, build the agent, get it in front of customers, and iterate based on what’s working.
This also means that if you are just investing post-Demo Day, you may not have the full picture on the team’s capabilities and upside.
Entry Prices are Matching The Pace
This level of execution velocity changes the valuation conversation whether investors want it to or not. The default YC round this batch is $4M on $40M post-money. Three years ago in W23, it was $2M on $20M and investors complained loudly about YC pushing founders to take those terms.
The entry price ~doubled in three years.
Last batch, one startup raised at $100M post-money.
Unsurprisingly, that caused an uproar:
However, in this batch, W26 has roughly 10 companies raising at $100M post (!!)
Brian Chesky pointed out this week that Airbnb raised $615k at $3M post-money after YC, which was the highest valuation in their batch.
While a stark contrast, it’s worth noting, capital was scarce post-recession and valuations reflected that scarcity - YC’s deal was also $20k for 6% back then with no MFN check.
The landscape today is unrecognizable compared to 2009, and complaining about current pricing without acknowledging what changed misses the point entirely.
Here’s why the pricing makes sense…
Companies are creating value faster. When teams can go from zero to $1M ARR in three months instead of twelve, when 14% weekly growth becomes the average instead of an outlier, when AI tools compress six months of engineering into six weeks, the math changes. Higher entry prices aren’t irrational exuberance. They’re the market repricing based on evidence that velocity increased.
Higher valuations create real risk for founders still searching for product-market fit. Optimizing for valuation this early usually shrinks outcome options because you box yourself into needing massive exits to make the cap table work. But if companies are capital-efficient enough that this round is the last expensive one they need, the pricing works. Several companies in our YC portfolio are already profitable and don’t need to raise again, though most will for growth acceleration. They’re raising strategically rather than out of necessity, which is fundamentally different than the capital-intensive growth model that defined the last decade.
The bears will say valuations are unsustainable and these companies are overpriced relative to fundamentals. That argument assumes linear growth trajectories and ignores what happens when AI scaling laws and Moore’s-law-like progress in model capabilities compound.
Exponential tech curves blow up linear forecasts every time, and the companies hitting $1M ARR in three months with founding teams are operating on exponential curves.
But ultimately, for Lobster Capital, this means being extra ruthless about finding the top 2% of each batch.
The premium is real and the entry price doubled, which means we can’t afford to be wrong as often.
We are not in the business of trying to get good businesses at fair prices. We’re trying to be in the best businesses that can produce generational outcomes, and those businesses can absorb higher entry prices if execution matches ambition.
The key stats from W26:
14% average weekly revenue growth across the batch
14 companies at $1M+ ARR
23,000 applications, less than 1% accepted
70% of the batch applied with zero revenue
P26 starts next week. Application deadline for S26 is May 4th.
Visit lobstercap.com for more.
Good take! Check this out- and lets stay in touch https://keithnewman.substack.com/?r=3d4ef&utm_medium=ios