Meta’s US$14.3 billion investment in data-labeling company Scale AI in June is facing challenges, as tensions emerge between the two firms and key executives depart.

Scale AI CEO Alexandr Wang and several executives joined Meta Superintelligence Labs (MSL) after the deal, but at least one senior hire, Ruben Mayer, has left Meta after only two months.

Sources said Mayer oversaw AI data operations but was not part of Meta’s core AI superintelligence team, though Mayer disputes this, saying he helped set up the lab and was involved from day one.

Meta’s AI unit is now working with other data-labeling vendors, including Surge and Mercor, despite its large investment in Scale AI.

Some Meta researchers have reportedly cited quality concerns with Scale AI’s data, preferring to use Surge and Mercor.

After losing OpenAI and Google as clients, Scale AI laid off 200 employees in July, with its new CEO citing the cuts as due to changing market demand.

Several AI researchers and executives, including recent hires and long-time staff, have left Meta in recent months.

🔗 Source: TechCrunch

🧠 Food for thought

1️⃣ The data labeling industry faces a fundamental business model shift

Scale AI’s struggles highlight a broader transformation in the data annotation industry, where traditional crowdsourcing approaches are becoming less effective.

The company originally built its business on using large numbers of low-cost workers for simple data labeling tasks, but modern AI models now require highly-skilled domain experts like doctors, lawyers, and scientists to generate quality training data1.

This shift explains why Meta’s researchers reportedly prefer competitors like Surge and Mercor, which built their business models around high-paid expert talent from the beginning1.

The industry transformation is significant, as Amplify Partners observed that “annotation for AI doesn’t scale” using traditional methods, since consensus-based labeling by non-experts fails to meet the quality standards required by advanced AI systems2.

Scale AI’s response included launching its Outlier platform to attract subject matter experts, but this pivot appears insufficient to address the mismatch between its original infrastructure and current market demands.

2️⃣ Big tech acquisitions often struggle with cultural integration despite massive investments

Recent Meta developments