AfterQuery Became YC’s Fastest Unicorn — and It Reveals What AI Companies Are Buying Next

A startup founded roughly 18 months ago is reportedly worth $3.2 billion.

Even by the standards of the AI boom, that is fast.

AfterQuery, an AI training-data startup founded by Spencer Mateega and Carlos Georgescu, has reportedly reached a $3.2 billion valuation — up more than tenfold from the $300 million valuation attached to its $30 million Series A in April. Y Combinator partner Gustaf Alströmer described it as the fastest company in YC history to go from launch to unicorn status.

But the valuation isn't the most interesting part of the story.

What AfterQuery sells tells us where the AI market may be heading next.

AI models have already consumed the obvious data

The first generation of large language models benefited from enormous quantities of publicly available text and online content.

The next generation needs something harder to find.

Expertise.

AfterQuery works with professionals including doctors, lawyers and other specialists to help train models and agents on complex professional workflows rather than simply providing basic data-labeling services.

The company describes its approach as teaching AI systems the patterns and decisions experts use while completing work.

That is becoming increasingly valuable as AI companies try to move beyond chatbots and build agents capable of performing actual tasks.

The data-labeling business is moving upmarket

Data labeling used to sound like one of the least glamorous businesses in technology.

AI has changed that.

Scale AI helped establish data infrastructure as a major venture category. Mercor pushed the market further toward recruiting skilled workers and professionals to help improve models.

Now startups such as AfterQuery are focusing on the reasoning behind professional work.

The distinction matters.

Teaching a model that an answer is incorrect is relatively straightforward.

Teaching it how an experienced analyst, lawyer or engineer approaches a complicated problem is considerably harder.

And scarcity tends to create valuable businesses.

The revenue claims are almost as notable as the valuation

In April, AfterQuery said it had reached an annualized revenue run rate of $100 million and named Nvidia, Legora and Motif Technologies among customers it had worked with.

Forbes subsequently reported that the company is profitable and that its recurring revenue had continued growing into the hundreds of millions.

Those numbers help explain investor enthusiasm.

But they also create enormous expectations for a company that barely existed two years ago.

The bigger startup lesson

AfterQuery's rise is another reminder that some of the biggest opportunities in AI may sit one layer away from the models themselves.

Not every breakout company needs to build the next frontier LLM.

Some will provide the data.

Others will provide infrastructure, security, evaluation, memory, orchestration or specialized workflows.

And as foundation models become more capable, the value may increasingly move toward whatever those models still lack.

Right now, expert reasoning appears to be one of those scarce inputs.

AfterQuery's $3.2 billion valuation may look extraordinary.

But investors appear to be betting that the data required to teach AI how professionals actually work could become one of the industry's most valuable commodities.