AI training data startup AfterQuery has reached a $3.2 billion valuation in a new funding round, Forbes reported Monday. The figure is more than 10 times the $300 million valuation the company carried five months earlier, when it closed a $30 million Series A.
Y Combinator partner Gustaf Alströmer said the jump makes AfterQuery the fastest startup in YC’s history to go from founding to unicorn status. AfterQuery declined to comment on the report. One Forbes source said the company is already profitable and has lined up a lead investor for the round.
Founded in early 2025 by two longtime friends
Spencer Mateega and Carlos Georgescu founded AfterQuery in February 2025. According to the report, the pair entered Y Combinator’s Winter 2025 batch as two high school friends with no product and no fixed idea.
Mateega wrote on X in July that annual recurring revenue had climbed into the “hundreds of millions”, up from $100 million in April. In that post, he said the company was rapidly closing the gap with larger players and had already grown several multiples beyond the previously cited $100 million revenue run rate.
From finance agents to reasoning data
The founders first wanted to build AI agents for finance. Testing pushed them in a different direction. Leading models, they found, kept missing on nuanced professional-grade work. The issue, as described in the report, was not raw model capability but the absence of training that shows how experts actually reason through hard decisions.
AfterQuery then pivoted into reasoning data. The company pays specialists to produce written, step-by-step records of how a professional works through a problem. That material is used to train AI systems on judgment rather than simple factual recall.
Per Forbes, Nvidia has used that data to train its open-source Nemotron models. AfterQuery’s clients also include Thinking Machines Lab, the company founded by former OpenAI CTO Mira Murati, and legal AI firm Legora.
Demand extends well beyond one startup
The shift reflects a broader industry need. Companies building frontier AI systems have already worked through much of the usable text available on the open web, while synthetic data can only do so much. What remains scarce is judgment drawn from real experts with actual credentials.
Other firms are tackling that shortage in different ways. The report pointed to South Korean fintech company Toss, which recently opened its 30 million users to the AI data economy through a partnership with data infrastructure firm Poseidon. The arrangement pays ordinary users to record real-world data that models cannot easily find online.
Competition in the data pipeline is heating up
AfterQuery is not alone in turning the shortage into a business. Scale AI founder Alexandr Wang became the industry’s first data-labeling billionaire in 2021 at age 24. Meta later paid $14.3 billion for a 49% stake and placed Wang at the top of its own AI lab.
Rival Mercor also surfaced in the report. Its founders became billionaires at 22 last October, surpassing Wang’s earlier age record.
Mateega has said AfterQuery’s advantage over Mercor lies in custom software that screens submissions for a “Goldilocks” level of difficulty: hard enough to challenge a frontier model, but not so difficult that the model cannot learn from the answer. The company also trains its own models on the data before selling it, aiming to show labs that the material improves performance instead of asking buyers to trust the claim.
Mercor, meanwhile, is in talks with Nvidia on a funding round that would value it at $20 billion, doubling its $10 billion valuation from last October.
AfterQuery’s own round has not yet closed, and the company has not identified its lead investor.