Mercor

Market position

Mercor is positioned as a leading AI data company and a principal competitor in the expert data-labeling and model-evaluation market alongside Scale AI, Surge AI, Turing, Invisible Technologies and Micro1. Its rise was accelerated by Meta's $14.3 billion purchase of a 49% stake in Scale AI in June 2025, after which some large labs reportedly cut ties with Scale; cofounder Adarsh Hiremath said the company's biggest competitor was 'torpedoed overnight.' Mercor was valued at $10 billion in October 2025 and was reported in mid-2026 to be in talks at a $20 billion valuation. It has appeared on the Forbes AI 50, Forbes Cloud 100, LinkedIn Top Startups, Bloomberg's 24 AI Startups to Watch and Newsweek's America's Greatest Startup Workplaces lists.

Mercor emphasizes depth of professional expertise rather than generic labeling: AI-driven, role-specific interviews to vet experts across 300+ fields, and its own APEX benchmark family to measure whether models can perform economically valuable professional work. It also combines the expert network with enterprise offerings such as custom AI agents and data monetization. Its perceived neutrality relative to Scale AI, after Meta acquired a large stake in that competitor, was cited as a commercial advantage.

Technology

Mercor's platform uses AI-driven, role-specific interviews to source, assess and match experts to projects; the original system evaluated candidates by analyzing interview transcripts, resumes and personal portfolio websites. The company also builds evaluation infrastructure in the form of the APEX benchmark family (APEX-Agents, APEX-Accounting, APEX-SWE), which measures AI model performance on real professional tasks, and deploys custom AI agents trained on client organizations' knowledge and workflows.

Go-to-market

Direct enterprise and AI-lab sales for data, evaluation and agent engagements, combined with self-serve recruitment of experts through public role listings on its website. The APEX benchmarks and published research, plus extensive press coverage, function as marketing and credibility channels; the company also maintains a newsroom, blog and expert 'stories' content.

Frontier AI labs and model developers such as OpenAI, Anthropic and Nvidia, plus Fortune 500 enterprises deploying AI in production. On the supply side, the platform targets skilled professionals including engineers, software developers, lawyers, doctors, bankers, quantitative traders, scientists, accountants, HR and sales professionals, and journalists.

Geography

Headquartered at 181 Fremont in San Francisco, California, with additional offices in New York City and London. The expert network is described as global; the company's earliest business connected freelance programmers in India with US companies.

History

The company was originally formed to connect freelance programmers in India with US companies, building an AI platform that interviewed programmers and matched them with hiring firms. It later shifted to data labeling and expert-sourced model training after inadvertently accumulating a network of specialized experts. In May 2025 it hired former Uber chief product officer Sundeep Jain as its first president. It raised a $100 million Series B led by Felicis, then in October 2025 announced a $350 million Series C at a $10 billion valuation, making its then-22-year-old founders the youngest self-made billionaires. In January 2026 it released its first APEX-Agents benchmark report. In late March 2026 it was affected by a supply-chain attack involving the LiteLLM package, followed by class-action lawsuits. In July 2026 it acquired AI agent-training startup Deeptune and was reported to be in talks to raise roughly $500 million at a $20 billion valuation; in August 2026 Nvidia was reported to be weighing participation in that round.

Risks & controversies

In late March 2026 Mercor was reportedly hit by a supply-chain attack tied to the LiteLLM open-source package, with hackers claiming up to 4 terabytes of internal data and contractor personally identifiable information; the incident triggered at least seven class-action lawsuits alleging exposure of interview recordings, facial biometric data and screenshots of workers' computers, and a temporary pause in work by clients including Meta. Mercor said it disputes the lawsuits' claims. Forbes also reported the company fired an employee for embezzling funds and that employees suspected North Korean operatives infiltrated the platform using stolen credentials. Contractors interviewed by The Verge, New York Magazine and Wired described a stressful work environment, poor management and declining pay. Business model risk also stems from concentration among a few large AI labs and intense competition in the data-labeling market.