On October 19, Brendan Foody, Mercor’s CEO, posted a short correction to the market’s perception of his company: “many people don’t realize Mercor now has hundreds of customers.” The post suggests a business reaching beyond a handful of buyers.
For an expert selling work through an AI training platform, it raises a more personal economic question: how much of that expanding demand will reach next month’s calendar?
In a recent Reddit discussion about micro1, one contributor reported earning $73 an hour on a robotics project, with roughly 40 hours of work before a pause. In a follow-up, they distinguished about 30 hours of production from other activity. Others in the same discussion described reliable payments, interesting assignments, and substantial ongoing work. These are unverified individual accounts, not a representative earnings survey.
Together, they describe a market worth taking seriously: professional expertise can command attractive prices while the quantity of work available remains uncertain.
AI training work includes creating examples, evaluating model outputs, designing tests, and applying professional judgment. Its impact on the labor market extends beyond adding a new job title. It introduces another buyer for specialized skills and another way to divide the risks of employing them.
The hourly rate is visible. The duration of demand, the costs of finding work, and the allocation of downtime are harder to see.
That is where the economics becomes interesting.
Here for practical AI training advice? Start with World Model.
If you joined to prepare for an assessment, understand what a project expects, or improve your evaluation work, World Model is the newsletter I write for you. It brings the practical side into focus: preparation steps, scoring rubrics, and worked examples you can use.
Start here: How to prepare for an AI training assessment. Semi Diffusion complements that practical work by examining the businesses, incentives, and labor market behind it.
Continue with the paid analysis: compare a $100 rate with a $60 rate, examine when uncertainty deserves a premium, and use four operating measures to judge whether a platform can turn demand into repeat work.