Updated
Updated · The New York Times · Jul 10
Mercor Seeks $20 Billion Valuation as AI Labs Fuel Exploding Demand for Expert Data
Updated
Updated · The New York Times · Jul 10

Mercor Seeks $20 Billion Valuation as AI Labs Fuel Exploding Demand for Expert Data

3 articles · Updated · The New York Times · Jul 10

Summary

  • Mercor is discussing a funding deal at a $20 billion valuation, doubling from its $10 billion valuation in October as demand for high-end AI training data surges.
  • 30,000 contractors earn more than $4 million a day through Mercor to supply expert-vetted data, with postings seeking specialists from Hebrew voice actors to physicists and doctors.
  • OpenAI, Anthropic and other labs increasingly want lawyers, mathematicians and professors to refine advanced models rather than low-paid workers doing rote tagging or transcription.
  • That shift has turned data-training middlemen into some of Silicon Valley's fastest-growing start-ups: Handshake says its annualized revenue hit $1 billion in April, up from $550 million at the start of the year.
  • The boom carries a built-in tension: firms like Mercor need client models to keep improving to prove their value, but also to stay imperfect enough that labs keep buying more human data.

Insights

Is Mercor's $20B valuation a bubble, ignoring its toxic culture and major data breach?
By using elite experts to train AI, is Mercor building a new form of systemic bias?
Can Mercor fix its broken culture and security without killing the hyper-growth that made it famous?