Data Marketplaces & Intermediaries

Mercor's H1 2026 revenue hits $614M, but 91% comes from a handful of foundation model labs

Source: BigGo Finance · Aug 12, 2026

Mercor's growth numbers are, on their face, exceptional: $614 million in revenue for the first half of 2026 alone, up 70% year-over-year, with June's run rate annualizing to roughly $2 billion. Six months of 2026 revenue already exceeds Mercor's entire 2025 total. But the number that matters more for understanding the business's actual risk profile is a different one: approximately 91% of that H1 revenue came from AI foundation model companies — specifically named as OpenAI, Anthropic, Google DeepMind, Reflection AI, and Thinking Machines Lab.

That's an extreme concentration in a customer segment that is itself unusually volatile — foundation model labs' data-purchasing budgets move with their own funding cycles, model-training schedules, and competitive positioning, none of which Mercor controls. A company generating over 90% of revenue from roughly five customer accounts, however large those accounts are individually, carries a fundamentally different risk profile than one with a broad, diversified customer base — even at identical total revenue and growth rate.

Mercor appears to be aware of this exposure. Fundraising materials reportedly list financial institutions — Ramp, Blue Owl, and Citigroup among them — as evidence of expansion into enterprise clients, explicitly framed as a move to reduce dependence on foundation-model-company revenue. Whether that diversification effort is early-stage or already meaningfully underway isn't clear from available reporting, but the fact that it's being highlighted at all suggests the company itself sees the concentration as a fundraising talking point that needs addressing, not just a footnote.

For companies building licensable data businesses, this is a useful data point on customer-mix strategy. A narrow base of frontier-lab customers can produce spectacular growth numbers in the near term — the AI labs have deep pockets and urgent data needs — but it also means revenue quality is harder to defend to investors or partners than a diversified base would be. It's a trade-off worth naming explicitly when building a data-licensing go-to-market plan: chasing the handful of large lab budgets can move revenue fast, but building toward a broader enterprise customer base is what makes that revenue durable over a multi-year horizon.

Key Points

• H1 2026 revenue: $614M, up 70% year-over-year
• June's monthly run rate annualizes to roughly $2B
• ~91% of H1 revenue came from AI foundation model companies (OpenAI, Anthropic, Google DeepMind, Reflection AI, Thinking Machines Lab)
• Mercor is reportedly courting enterprise clients (Ramp, Blue Owl, Citigroup) to diversify away from lab concentration