AfterQuery Hits $3.2B, YC's Fastest Unicorn Ever

AfterQuery, a San Francisco startup that builds training data and reinforcement-learning environments for AI labs, has raised at a $3.2 billion valuation. The round was first reported by Forbes and picked up by TechCrunch on 1 September. The company was unavailable for comment, and the size of the round itself has not been disclosed.

The number that made it news is the interval. In April the company announced a $30 million Series A at a $300 million valuation. Five months later the mark is more than 10x higher. Y Combinator partner Gustaf Alströmer called it the fastest any startup has gone from launch to unicorn in the accelerator’s history — AfterQuery went through the Winter 2025 batch roughly 18 months ago, and its two founders, high-school friends, are 22 and 23 years old.

What the company sells is people, packaged. Rather than scraping the open web, AfterQuery pays working specialists — doctors, lawyers and other practitioners — to encode the patterns, decisions and reasoning behind how they actually complete tasks, then turns that into curated datasets and RL environments a lab can train against. As of April it reported a $100 million annualized revenue run rate, with Nvidia, the legal-AI firm Legora and the Korean lab Motif Technologies among named customers.

That business exists because of a specific bottleneck. Frontier pretraining has largely worked through cheap public text, and the remaining gains in agentic work come from environments where a model can be scored on doing a real job end to end — which is exactly what is hardest to scrape and easiest to commission. The result is a data layer that has quietly become one of the better-funded corners of the industry, alongside Mercor, Scale AI and smaller specialists such as Wirestock.

Some proportion is worth keeping, though. The $3.2 billion figure is a private mark set by investors competing for allocation in a category everyone agrees is scarce, and the last public revenue number is five months stale. A 32x multiple on a $100 million run rate is priced for that run rate to keep moving. The thing to watch is not the valuation but whether lab spending on commissioned expert data holds once the labs start building those pipelines in-house — which is the move every large buyer in this market has eventually made.