According to Forbes, Mercor’s three 22-year-old cofounders became the world’s youngest self-made billionaires last month after raising $350 million at a $10 billion valuation. Just over a week later, the AI training startup abruptly canceled Project Musen, which involved over 5,000 contractors reviewing Meta’s Facebook and Instagram Reels content. Contractors found themselves locked out of Slack on Tuesday without warning, despite being told in October the project would run until at least December. Hours later, Mercor emailed contractors offering new roles on Project Nova but at a reduced rate of $16 per hour, down from their previous $21 pay. The company had hit a $500 million annualized revenue run rate in September and saw fivefold growth since March, making the sudden wage cuts particularly jarring for workers.
The human cost of the AI boom
Here’s the thing about these rapid AI valuations – they’re built on the backs of real people doing often tedious work. These contractors weren’t just numbers on a spreadsheet; they were people who’d been told they had stable work through December, only to get locked out of systems with zero notice. One contractor put it perfectly: “It’s contract work but we are real people who deserve some notice, or warning or some consideration.”
And let’s talk about that pay cut. Dropping from $21 to $16 hourly isn’t just a minor adjustment – in states like California, Washington, and Connecticut, that’s actually below minimum wage. Mercor’s justification? They’re offering “steadier task volumes” and “greater earning stability.” But come on – does anyone really believe that’s anything but corporate spin for “we want to cut costs while maintaining output”?
This is part of a bigger pattern
This isn’t some isolated incident. The entire AI data labeling industry has become a gold rush, with companies like Scale, Surge, and Mercor scrambling for contracts. Remember when OpenAI was paying Kenyan workers $2 per hour to make ChatGPT less toxic? We’re seeing the same dynamic play out here, just with slightly higher dollar amounts.
The irony is palpable. While Mercor is cutting pay for basic data work to $16 hourly, they’re simultaneously running job ads offering lawyers, journalists, and doctors up to $200 per hour for specialized AI training. It creates this bizarre two-tier system where some knowledge workers get premium rates while the people doing the fundamental grunt work get treated as disposable.
Where does this go from here?
I’ve got to wonder – how sustainable is this model? When you treat your workforce as completely expendable, what happens to quality? These contractors were reviewing content to train AIs to identify people and products in videos – that’s not exactly trivial work. If you constantly churn through workers and slash their pay, you’re going to get what you pay for.
Meanwhile, the legal battles are already heating up. Scale is suing Mercor for allegedly stealing trade secrets, which Mercor’s CEO casually dismissed as “not something we spend a lot of time thinking about.” That kind of attitude might work when you’re riding a valuation tsunami, but it rarely ages well.
Basically, we’re watching the same Silicon Valley playbook get applied to AI: move fast, break things, and worry about the human consequences later. The difference this time? The scale is massive, and the workers being affected aren’t just in distant countries – they’re right here in the US, trying to make ends meet while billionaires get minted overnight.
