AI recruiting platform Mercor hit $2B annualized run rate with 90% revenue from OpenAI, Anthropic, Google. The numbers validate the total addressable market for decentralized data-labeling protocols.
Alpha Score of 72 reflects strong overall profile with moderate momentum, moderate value, strong quality, strong sentiment.
A three-year-old AI recruiting and data platform, Mercor, generated $614 million in gross revenue in the first half of 2026, with more than 90% coming from artificial intelligence foundation model companies, according to internal documents cited by The Information.
The revenue implies an annualized run rate of roughly $2 billion, up from $1 billion four months earlier. The company connects domain experts and contractors with AI labs for data labeling and model evaluation, along with reinforcement learning from human feedback. Mercor compensates about 30,000 contractors at an average of $105 an hour, with daily payouts exceeding $2 million.
Its clients include OpenAI and Anthropic, with Google also among them, according to the company's website. The concentration of revenue among a few large AI labs creates vulnerability if any of those clients shift operations in-house or turn to alternative providers.
For the crypto industry, the numbers offer a concrete benchmark. Several blockchain projects are building decentralized alternatives to centralized data-labeling and model-training services. Mercor's $2 billion run rate provides a size estimate for the total addressable market those projects target, though the centralized model also demonstrates the operational scale required: a network of 30,000 contractors with daily payouts that would challenge many decentralized protocols.
CEO Brendan Foody, along with co-founders Adarsh Hiremath and Surya Midha, have raised $486 million to date at a $10 billion valuation. The H1 2026 haul represents about 70% growth over the company's full-year 2025 revenue, underscoring the pace of demand for human-in-the-loop AI training.
Mercor's growth reflects a market that decentralized AI projects are still trying to capture. The revenue concentration also highlights the risk of relying on a handful of foundation model developers for the bulk of revenue, a dynamic that decentralized networks aim to avoid through distributed incentives.
Drafted by a large language model from the source reporting linked above, then screened by automated publishing checks. It is not read by a journalist before publication. Some articles cite our Alpha Score. Verify prices and figures against the original source. Educational coverage, not personalized advice.