As Amazon lets Mechanical Turk fade, Mercor hits a $2 billion gross run rate

Why it matters

Mercor's $2 billion gross run rate shows how quickly expert human labor has become a core AI input, but the contractor payout structure keeps the margin question front and center.

Brendan Foody's Mercor hit more than $2 billion in gross annualized revenue in June, double its pace earlier in 2026, The Information reported Monday, citing a person with direct knowledge of Mercor's financials.

The milestone lands against a sharp backdrop: Amazon has put Mechanical Turk, the old marketplace for internet-scale human microtasks, into maintenance mode. Mercor's expert-labor marketplace is the other side of that shift. AI labs and enterprises are no longer just buying cheap, interchangeable task completion. They are paying for credentialed human judgment that can train, evaluate and tune increasingly capable models.

That puts Mercor near the center of the current AI supply chain: models need expert feedback, enterprises need specialized data, and Mercor has built a marketplace that pays domain specialists by the hour to produce it. Foody, who co-founded Mercor in 2023 with Adarsh Hiremath and Surya Midha, has pushed the business from automated recruiting into a more lucrative market for expert-generated training data.

Mercor's reported revenue is gross revenue, not software-style ARR. The Information reported that Mercor pays 60% to 70% of topline revenue to contractors, which would put annualized net revenue at roughly $600 million to $800 million if that payout structure still holds at the June run rate. That distinction matters because Mercor's headline number includes large pass-through labor costs. It also explains why AI data companies can show unusually large revenue run rates while still being evaluated by investors on whether they can turn contractor supply into durable margin.

Mercor is profitable on a free cash flow basis, according to the person cited by The Information. Mercor has not publicly released audited financials, and the June run-rate figure is based on a single month annualized, a useful speedometer for a company growing this quickly and a weak substitute for full-year revenue.

The founder bet is expert labor as AI infrastructure

Mechanical Turk made human labor programmable for the web era. Mercor is trying to make expert labor programmable for the AI era. The difference is not just pricing, but who the workers are, what the work is used for and how much strategic value customers attach to the output.

Mercor's pitch has shifted as AI labs have moved from broad internet data to harder, more expensive data produced by people with credentials. The Information said Mercor's growth is coming from AI application developers and Fortune 500 customers building or fine-tuning models. Mercor pays contractors with domain expertise in areas such as physics and finance to answer questions and create specialized training data.

That is a different business than the one investors first backed. TechCrunch reported in February 2025 that Mercor, then run by three 21-year-old Thiel Fellows, had raised $100 million at a $2 billion valuation. At the time, Mercor was described as an AI recruiting platform that automated resume screening, matching, interviews and payroll management. Foody told TechCrunch that Mercor collected candidate performance data to improve predictions about who would perform best in future roles.

The growth since then came from a sharper use case: the same talent graph can be sold to AI labs and enterprises that need specialists to train and evaluate models. In February 2025, TechCrunch reported that Mercor said it worked with the world's top five AI labs, including OpenAI. By June 2026, according to The Information's latest report, demand had pushed Mercor's gross annualized revenue above $2 billion.

Mercor has been explicit about the strategic turn. In an October 27th, 2025 blog post, Foody wrote that Mercor connects human expertise with leading AI labs and enterprises, and that Mercor's talent network trains frontier AI models by sharing knowledge, experience and context. In the same post, Mercor announced a $350 million Series C led by Felicis, with participation from Benchmark, General Catalyst and Robinhood Ventures, valuing Mercor at $10 billion.

That valuation followed Mercor's September 2025 claim that it was "the fastest growing company of all time" after scaling from a $1 million run rate to a $500 million run rate in 17 months, according to The Information. Mercor's June 2026 run rate, if sustained, would be four times that September pace in less than a year.

Gross revenue is the scoreboard competitors are racing on

Mercor is competing in a market where the revenue numbers have become unusually large because the customer base is unusually concentrated and well-funded. AI labs are spending heavily to improve model behavior in domains where generic web data is insufficient. Enterprises are trying to adapt general models to internal workflows. Both groups need data that captures judgment, edge cases and specialized process knowledge.

The maintenance-mode fate of Amazon's Mechanical Turk underscores how much the market has moved. The old model treated human work as a general-purpose pool of small tasks. The new AI data market treats the right human output as infrastructure, with contractor supply, evaluation design and enterprise workflow context all becoming part of the product.

Handshake, a Mercor rival, reached $1 billion in gross annualized revenue earlier this year before contractor payouts, according to The Information. Sacra estimated in May that Handshake hit $1.1 billion in annualized gross revenue in April 2026 and roughly $450 million in net revenue after contractor payouts. Sacra also framed the key question for Handshake, Mercor and similar data-labeling businesses as whether they can move beyond marketplaces with minimally differentiated supply into a stickier data infrastructure layer.

Mercor is trying to answer that question with products around evaluations and enterprise AI deployment. On October 1st, 2025, Mercor introduced APEX, an AI Productivity Index designed to test whether models can perform economically valuable knowledge work across investment banking, law, consulting and medical practice. On March 26th, 2026, Hiremath introduced Mercor Enterprise AI, a platform Mercor says maps how employees actually work and turns that context into agent behavior specs, evaluations and quality guardrails.

Those moves show why the $2 billion gross run rate matters beyond scale. If Mercor remains a broker for expert labor, the 60% to 70% contractor payout will define the business. If Mercor can package expert work into benchmarks, workflow data and enterprise evaluation systems, Mercor has a path to more durable revenue than hourly markup.

Mercor's risks are also becoming more visible as Mercor grows. In late March 2026, Mercor was affected by a supply-chain cyberattack involving LiteLLM. Mercor said in a July 6th security update that it had nearly five million experts, that only a limited subset had sensitive information affected, and that all frontier labs had increased their work with Mercor over the prior few months. That last claim is Mercor's own account, but it lines up with The Information's report that revenue doubled by June.

For Foody, Hiremath and Midha, the current business is a bet that human expertise becomes more valuable as AI models get stronger because the scarce input shifts from generic labor to judgment-rich data. Amazon's Mechanical Turk shows what happens when human-task marketplaces become commodity infrastructure. Mercor's June run rate says customers are paying for a more specialized version of that labor at a pace few marketplaces have ever reached. The open question is how much of that revenue Mercor can keep once contractors, competitors and AI labs all understand where the money is.