Earnings down
Scale AI was once known as a fast-growing company that helped major tech firms train and check AI systems. But the mood among workers has changed since Meta invested $14 billion and hired founder Alexandr Wang. Contractors say their earnings have gone down, and the amount of work available has also decreased. Many taskers on Outlier, the company’s gig-work platform, report long unpaid onboarding sessions with no guarantee of tasks afterward. Some say that pay for certain projects has dropped from about $50 an hour to around $20. A few workers shared cases in which $20-an-hour projects allowed only a few minutes of work every two days, resulting in payouts of less than a dollar. Scale AI spokesperson Joe Osborne says the company is performing well and that pay rates are clear before workers accept tasks. He also says more users have joined Outlier since the Meta deal.
Contrasting claims and reality
Even as the company says business is growing, its valuation in private markets has fallen sharply. Before the Meta deal, Scale AI was valued at $29 billion. Recently, shares on secondary platforms have traded at implied values between $9 billion and $15 billion. One marketplace listed an even lower figure, though Scale AI denies its accuracy. Some investors believe Meta mainly wanted to hire Alexandr Wang rather than invest in Scale AI’s broader operations. Others say the company still has enough money and time to recover, with more than $1 billion in cash reserves.
Aftermath of Meta deal
Soon after the Meta deal, Scale AI laid off about 14% of its full-time staff. The company later cut red-team contractors and shut down a Dallas-based group. Former workers say these cuts followed months of slow project flow. Rival companies have moved quickly to take advantage. Surge AI has grown rapidly, and Mercor, run by three young founders, has raised large sums and even won work from Meta. Scale AI has responded by filing a lawsuit against Mercor, accusing it of trying to take major clients. The former has also faced criticism over security lapses. Some internal training documents were found stored in public Google Docs, exposing personal data and confidential project details. The company says it has fixed the issue.
