Docs: AI data startup Mercor had $614M in gross revenue in H1 2026, up 70% from all of 2025, with ~91% from foundation model makers like OpenAI and Anthropic
Mercor, a three-year-old data startup whose army of contractors teaches AI to give better answers, is growing fast but relies heavily …
Context & Ripple Effects
Mercor's reported scale gives operating context to its rapid financing arc: it was raising $350M at a $10B valuation in late 2025 and was recently reported to be discussing funding at roughly a $20B valuation.
The company sits in an increasingly competitive market for AI-training contractors, where rival micro1 had also told investors it was generating substantial recurring revenue.
First-order effects
- Mercor gains a stronger revenue basis for fundraising and recruiting, but its business is immediately exposed to the purchasing decisions of a small set of foundation-model customers.
- OpenAI and Anthropic become especially consequential counterparties for Mercor because they account for roughly 91% of its reported revenue.
Second-order effects
- Rivals such as micro1 face greater pressure to demonstrate comparable customer access, contractor supply, and revenue durability as Mercor's scale becomes more visible.
- Mercor's concentration gives its largest model-maker customers leverage in commercial negotiations and makes retention of those accounts central to its growth profile.
Third-order effects
- If leading model developers continue to outsource large parts of training-data work, contractor marketplaces may become a more important layer of AI infrastructure rather than simply recruiting platforms.
- The market could consolidate around vendors that can reliably serve a few large model labs, though that structure also leaves suppliers vulnerable to customer concentration and shifts in labs' in-house capabilities.
The trend: AI model developers' demand for specialized human feedback is turning contractor marketplaces into high-revenue infrastructure suppliers, while concentrating their economics around a handful of labs.