Undergrads Skip Class to Grade AI for $1,300 a Week. The Startup Behind Them Is Reportedly Worth $3.2 Billion

A $1,300 week grading the machine

Last spring, a Penn freshman skipped four straight days of classes to grade a client model's answers for eight hours a day, and cleared $1,300 in under a week ¹. The job came with no interview; a comment on a LinkedIn post was the application ¹. The work flows through AfterQuery Experts, a contractor Slack of more than 15,000 members, staffed heavily by students ¹.

A $3.2 billion mark nobody has confirmed

On 1 September, Forbes↗, citing two people with direct knowledge, put AfterQuery, founded in February 2025, at a $3.2 billion valuation; the company declined to comment, and TechCrunch↗ matched the figure ²³. The round would sit more than ten times above its $300 million April mark and, per YC partner Gustaf Alströmer, would be the fastest launch-to-unicorn in Y Combinator history ²³. The founder posted in July that recurring revenue had reached the hundreds of millions ².

The student scoop miscounted the round

The undergraduate workforce first surfaced in a 34th Street feature↗ that dates the April round at $50 million on a $500 million valuation ¹; the business press carries $30 million on $300 million ³⁴. The direction is the same either way. The gap matters because the record is thin: contractors cannot name their clients, only project codenames like Egg and Super Red for unnamed labs holding "huge contracts" ¹.

The students are the supply chain and the target market at once, an arrangement the launch post↗ barely disguises: "If AI is one day to replace jobs entirely, it can't just be good, it needs to be near perfect" ⁵.

Why frontier labs pay students

Demand is measurable: a $100 million annualized run rate by April, with Anthropic and OpenAI among buyers, and work used in Nvidia's Nemotron models ²⁴. Supply prices like a credentialed gig economy: listings run $30 to $150 an hour↗, median $50, paid weekly, in bursts ⁶. A third-party review↗ counts a network nearing 100,000 professionals ⁷.

The trade eats its trainers

The peer set is the warning label. Mercor closed a $350 million round at $10 billion in October 2025, reportedly lost Meta as a customer after a March 2026 breach, and is now in reported talks with Nvidia at $20 billion ²⁴⁸.

In this market a lab's favor is the whole balance sheet, and it can shift between rounds ⁴. One undergraduate finance trainer got there first: "we're basically becoming more replaceable" ¹.

The feature's darker labor-market numbers did not trace to a primary study in search. One claims a 19% employment shortfall for young software engineers; another, a poll in which a third of recruiters replace entry-level seats with AI ¹.

A named round settles the mark

Forbes's source says the company is profitable with a lead investor lined up ²; a named round or a filing settles the mark. The tell for everyone else is entry-level hiring in finance, software and law, measured against the benchmarks these students grade. The trainers' own hedge is fluency: "AI is not gonna replace us, but people who know how to use AI will" ¹.