AfterQuery becomes Y Combinator's fastest-ever unicorn at $3.2 billion valuation

AI data startup AfterQuery has reached a $3.2 billion valuation, making it the fastest company in Y Combinator history to hit unicorn status. Founded in early 2025 by Spencer Mateega and Carlos Georgescu, the company pivoted from building financial agents to providing expert-generated reasoning data. Utilizing a network of 100,000 professionals, AfterQuery serves major clients like OpenAI, Google DeepMind, and Microsoft AI, capitalizing on an industry-wide shortage of high-quality training data.

AfterQuery unicorn valuation jump

• ▪AfterQuery is profitable and has lined up a lead investor for its new funding round, according to sources with direct knowledge of the matter
• ▪AfterQuery's $3.2 billion valuation is a ten-fold increase from the $300 million valuation it received during a $30 million Series A funding round in April 2026
• ▪San Francisco-based AI data startup AfterQuery reached a $3.2 billion valuation, making it the fastest company in Y Combinator's history to achieve unicorn status
• ▪AfterQuery's annual recurring revenue grew to the "hundreds of millions" by July 2026, up from $100 million reported in April 2026

Expert reasoning data business

• ▪AfterQuery maintains working relationships with Chinese AI labs, positioning itself as a neutral data provider in a fragmented geopolitical landscape
• ▪AfterQuery runs an internal post-training pipeline to train models on its data and measure benchmark performance changes, proving data quality to AI labs before sale
• ▪AfterQuery's client list includes OpenAI, Google DeepMind, Microsoft AI, Nvidia's Nemotron models, Mira Murati's Thinking Machines Lab, and legal AI startup Legora
• ▪AfterQuery provides human-curated training datasets and reinforcement learning environments using a network of approximately 100,000 verified professionals across finance, medicine, law, and software engineering
• ▪AfterQuery uses custom software to validate human-generated training data, screening submissions to ensure they fall within a "Goldilocks" difficulty range that challenges but does not stump models

AI training data shortage

• ▪Frontier AI labs face a shortage of training data because they have already scraped most publicly available text on the open web, while synthetic data has limits
• ▪South Korean fintech firm Toss partnered with data infrastructure firm Poseidon to pay ordinary users to record real-world data for the AI data economy
• ▪Reddit signed a $60 million annual deal with Google to grant access to its data for AI training purposes

Founder pivot from finance

• ▪Spencer Mateega and Carlos Georgescu pivoted AfterQuery to expert data labeling after finding that leading AI models failed at nuanced white-collar workflows due to a lack of step-by-step reasoning training
• ▪Spencer Mateega and Carlos Georgescu founded AfterQuery in early 2025, initially planning to build AI agents for financial workflows

Data labeling billionaire founders

• ▪Mercor's founders became the youngest self-made billionaires at age 22 in October 2025, and the company is in talks with Nvidia for a funding round at a $20 billion valuation
• ▪Scale AI made Alexandr Wang the original data labeling billionaire in 2021 at age 24, and the company is currently in talks with Nvidia at a $20 billion valuation

3 sources