Analysis
I have been underwriting seed and Series A rounds for long enough to know what a 10x markup in five months usually means, and it is almost never that the business got ten times better. By the start of September, AfterQuery was reportedly valued at $3.2 billion, which Y Combinator partner Gustaf Alstromer called the fastest run from launch to unicorn in the accelerator's history. Pulse covered the round on Monday.
- Series A (April 2026) -- $30 million raised at a $300 million post-money valuation
- Current valuation (September 2026) -- reportedly $3.2 billion, per YC
My read is that this price is not being set by AfterQuery's revenue. It is being set by the scarcity of a specific input: credentialed human judgment. Frontier labs have exhausted the easy internet, and the remaining training signal that improves a model on medicine, law and finance has to be produced by people who hold those credentials. There are a finite number of them, and whoever has them under contract has pricing power for as long as the labs keep buying.
“Mercor was in talks at a $20 billion valuation in July, and Scale AI has been the incumbent since 2016.”
That is why the comparable is not a data-services company. It is a staffing firm with a scarce union. Mercor was in talks at a $20 billion valuation in July, and Scale AI has been the incumbent since 2016. Three companies are bidding for the same specialists, and the labs are bidding for all three.
- AfterQuery run rate (April 2026) -- $100 million annualized against a $300 million valuation, roughly 3x revenue
- AfterQuery valuation (September 2026) -- $3.2 billion; the multiple only holds if the run rate has moved substantially, which nobody has confirmed
What I would actually diligence, if I could get into this round: contract duration and concentration. Expert data supply is not a moat if the underlying doctors and lawyers are contractors who can work for Mercor next quarter, and it is not durable revenue if two frontier labs are most of the book. Ask for the weighted average remaining contract term and the top-two customer concentration. If the answer is short contracts and concentrated buyers, the $3.2 billion is a scarcity premium on a market that resolves the moment labs decide synthetic data is good enough for the medical vertical.
Room for disagreement: the strongest counter-argument is that this is genuinely a new category and I am pricing it with old comparables. If expert data becomes a permanent line item in every frontier lab's budget -- the way cloud spend became permanent -- then the right analogue is not staffing but a marketplace with network effects on both sides, and marketplaces do trade at these multiples when they win. Founders aged 22 and 23 who went through YC's Winter 2025 batch and built a nine-figure run rate in 18 months have already outperformed anything my mental model predicted, which is a reason to hold the view loosely.
The number I am watching is whether the next expert-data round prices above $3.2 billion or below it. One more markup and the category has a floor. A flat round from any of the three and this was the top.
