TL;DR: The AI-powered Upwork, Mercor, is matching AI labs with pre-vetted domain experts from fields like law, medicine, and physics. Sacra estimates they hit $50M in annual recurring revenue (ARR) at the end of 2024, up 50x, with growing demand for high-skill experts for reinforcement learning with human feedback (RLHF). For more, check out our full report and dataset on Mercor.
Key points via Sacra AI:
- Early on, AI companies like OpenAI hired ~$1/hr overseas contractors from countries like Kenya and the Philippines to help fine-tune their AI models through a process known as reinforcement learning through human feedback (RLHF), outsourcing the work via companies like Scale AI ($950M ARR in 2024). Humans go through thousands of LLM outputs by hand, ranking the results and flagging any unsafe, explicit, or violent results, with the model then learning from that feedback.
- As labs started building LLMs focused on complex math, science, and other reasoning tasks, frontier labs began doing RLHF with $50-100/hr medical doctors, lawyers and PhDs to improve their models’ capabilities—creating the opportunity for Mercor to launch as an AI-powered Upwork for AI companies to hire large numbers of pre-vetted domain experts on hourly contracts. Mercor uses AI video interviews and tests to vet experts at scale (300,000 as of 2024), then matches them to companies like OpenAI and Meta, charging a monthly subscription for access to the platform and a 30% recruit fee for any candidates hired full-time.
- On the back of growing demand from AI labs for human experts to train the next generation of models, Sacra estimates that Mercor hit $50M in annual recurring revenue (ARR) at the end of 2024, growing 50x from $1M at the beginning of the year, with its September Series A led by Benchmark valuing it at $250M for a 5x forward revenue multiple. Compare to data labeling giant Scale AI, which exploded from $215M ARR in 2022 to $760M ARR in 2023 after the launch of ChatGPT, en route to $950M in in August 2024, up 64% YoY—and launched its own service for hiring high-skill contractors, Expert Match.
For more, check out this other research from our platform:
- Mercor (dataset)
- Scale AI (dataset)
- Scale at $760M ARR
- Scale: the $290M/year Mechanical Turk of machine learning
- Contractor Payroll: The $1.4T Market to Build the Cash App for the Global Labor Market
- Wingspan's 992x growth in contractor payroll
- Ved Sinha, Former VP of Product at Upwork, on gig marketplaces
- Samiur Rahman, CEO of Heyday, on building a production-grade AI stack
- Geoff Charles, VP of Product at Ramp, on Ramp's AI flywheel
- Mike Knoop, co-founder of Zapier, on Zapier's LLM-powered future
- OpenAI (dataset)
- Anthropic (dataset)
- Cursor (dataset)
Read more from
Mercor revenue, growth, and valuation
Unlocked Report
Continue Reading
Read more from
Read more from
Bolt.new revenue, growth, and valuation
Unlocked Report
Continue Reading
Crusoe revenue, growth, and valuation
Unlocked Report
Continue Reading