AfterQuery Becomes Y Combinator's Fastest-Ever Unicorn at $3.2 Billion

In AI, yesterday’s jaw-dropping valuation is today’s starting point. Eighteen months after two high school friends joined Y Combinator with no idea and no product, their AI training-data company AfterQuery has reached a $3.2 billion valuation, according to two people with direct knowledge of the matter who spoke to Forbes. The round, first reported on September 1, 2026, represents more than a tenfold increase from the $300 million valuation the San Francisco startup commanded just five months ago — and it makes AfterQuery the fastest company in Y Combinator’s history to go from inception to unicorn status, according to YC partner Gustaf Alströmer.

From No Idea to $3.2 Billion in 18 Months

The speed here is the story. Founders Spencer Mateega, 23, and Carlos Georgescu, 22, attended Y Combinator’s Winter 2025 cohort. In April 2026, AfterQuery announced a $30 million Series A at a $300 million valuation, led by Altos Ventures. At that time, the company said it had already reached an annualized revenue run rate of $100 million and was working with many of the biggest AI labs.

By July, Mateega was posting on X that recurring revenue had climbed into the “hundreds of millions.” And according to Forbes’ sources, AfterQuery is profitable and has lined up a lead investor for the new round — a detail that separates it from the capital-incinerating profile typical of AI startups scaling at this velocity.

What AfterQuery Actually Sells

AfterQuery is part of a fast-growing class of AI data providers that supply what frontier labs now need more than almost anything else: complex reasoning data created and vetted by humans. The company describes its mission as “encoding the patterns, decisions, and reasoning of the world’s best practitioners.”

Unlike first-generation data-labeling shops that annotated images or transcribed text, AfterQuery employs knowledge professionals — software engineers, lawyers, financial analysts, doctors — to generate step-by-step reasoning and judgment calls across finance, software engineering, law, and medicine. Rather than merely checking whether a model answers a question correctly, AfterQuery trains models and agents on how professionals actually work: how they sequence decisions, weigh trade-offs, and recover from mistakes.

The company’s differentiation is technical, not just thematic. Mateega has said AfterQuery’s edge over rivals like Mercor — which relies on large contractor pools selected by an AI interviewer — is custom software that validates human-generated training data. Expert contributions are screened to fall within a “Goldilocks” range: difficult enough to challenge state-of-the-art systems, but not so hard that the data is unlearnable.

Just as unusually, AfterQuery runs its own internal research pipeline. Instead of handing datasets to labs and letting them judge, the company trains models on its own data and measures how benchmark performance changes — proving quality before a lab even looks at a sample. Its work was used in Nvidia’s new series of Nemotron models (Nvidia publicly credited AfterQuery’s Off-The-Shelf Office Agent Training Dataset for improving Nemotron 3 Ultra on GDPval), and the company has also worked with former OpenAI CTO Mira Murati’s Thinking Machines Lab, legal AI startup Legora, and the Korean AI lab Motif Technologies. Unlike many of its peers, AfterQuery also works closely with Chinese AI labs.

Why the Market Is Paying These Prices

The valuation is explained by the state of the frontier-training market. Labs have already swallowed most of the high-quality text on the public web. Synthetic data helps, but it has well-documented limits — models trained on their own outputs tend to plateau. What remains scarce is domain-specific reasoning data produced by genuine experts: the kind of tacit professional judgment that never made it into any corpus.

That scarcity has turned formerly unglamorous data work into one of the quickest routes to behemoth valuations. Scale AI made then-24-year-old Alexandr Wang the original data-labeling billionaire in 2021; last October, Mercor’s founders surpassed him as the world’s youngest self-made billionaires at 22. Scale is now reportedly in talks with Nvidia at a $20 billion valuation. The unicorn pipeline is accelerating across the board — roughly 250 startups have crossed the $1 billion mark in 2026 so far, versus 193 in all of 2025 — and companies are going from launch to unicorn status in 18–24 months. AfterQuery did it at the extreme end of that range.

The Bigger Picture: Expertise Is the New Bottleneck

The AfterQuery story is the clearest signal yet of where the AI industry’s bottleneck has moved. For a decade, compute was the constraint. Then it was algorithms. As agents take on white-collar workflows — writing contracts, closing books, debugging production systems — the binding constraint is now access to high-quality demonstrations of expert human judgment. Frontier labs are willing to pay 10x premiums for companies that can manufacture it reliably.

There are caveats. AfterQuery declined to comment on the round, the $3.2 billion figure comes via unnamed sources, and “fastest unicorn in YC history” is a record built on an 18-month-old company whose revenue claims are self-reported. The premium-data market could also compress if labs succeed in generating expert-grade reasoning data synthetically, or if the agent economy’s growth disappoints.

But for now, the market has spoken: two founders who entered YC with nothing have built one of the fastest value-creation stories in Silicon Valley history — by selling the one thing the biggest AI companies cannot synthesize on their own: the way the best humans think.