Meta’s $14.3 billion investment in Scale AI was expected to cement one of the most significant alliances in artificial intelligence, with Scale CEO Alexandr Wang and several senior leaders joining the newly formed Meta Superintelligence Labs (MSL).
Just months later, however, the collaboration is showing early signs of strain.
Executive Turnover Raises Questions
One of Meta’s notable hires, Ruben Mayer—formerly Scale AI’s Senior VP of GenAI Product and Operations—left the company after only two months.
Reports suggested Mayer managed data operations but was not involved in MSL’s core research unit, TBD Labs, where ex-OpenAI talent now works.
Mayer disputed this characterization, stating he was part of TBD Labs “from day one” and that his role was to “help set up the lab, with whatever was needed.” He also clarified that he did not directly report to Wang and described his time at Meta as a positive experience.
Meta Broadens Data Labeling Beyond Scale
Despite the investment, Meta is diversifying its data-labeling partnerships. Insiders revealed that TBD Labs is working with Scale’s competitors, including Surge and Mercor, to train its next generation of AI models.
While distributing labeling tasks across vendors is common, the move is notable given Meta’s financial commitment to Scale. Some researchers reportedly prefer Surge and Mercor, citing stronger domain expertise and higher-quality data compared to Scale’s workforce.
Shifting Data Labeling Economics
Scale AI built its early success on crowdsourced, low-cost labeling. As AI models advance, however, demand has shifted toward expert-generated datasets requiring doctors, lawyers, and scientists.
Competitors such as Surge and Mercor positioned themselves early for this shift, while Scale has been adapting through its Outlier platform to attract specialists.
Scale AI’s Broader Challenges
The Meta partnership comes amid a turbulent period for Scale AI,
- Customer losses: OpenAI and Google ended contracts with Scale following Meta’s investment.
- Layoffs: Roughly 200 data-labeling staff were cut in July, attributed to changes in market demand.
- Leadership shift: New CEO Jason Droege has redirected focus toward government contracts, securing a $99 million deal with the U.S. Army.
Meta’s Strategic Gamble on Wang
Some analysts argue Meta’s investment was aimed less at Scale’s business and more at bringing Alexandr Wang into its ecosystem.
Wang has already helped attract researchers from OpenAI and Anthropic to MSL. Still, several former Scale executives at Meta are not working within TBD Labs, raising questions about their integration.
Organizational Strain Inside Meta Superintelligence Labs
MSL itself has faced turbulence. Employees describe internal friction since Wang’s arrival, with concerns over bureaucracy, diminished roles for existing teams, and a wave of resignations.
Departures include researcher Rishabh Agarwal, product director Chaya Nayak, and engineer Rohan Varma.
CEO Mark Zuckerberg’s urgency to catch up with rivals has fueled aggressive hiring, acquisitions, and infrastructure investment, including the $50 billion Hyperion data center in Louisiana.
Yet despite lavish compensation packages, talent turnover remains high, with some new hires reportedly considering returning to their previous companies.
Pressure of the AI Talent Arms Race
Across the industry, AI specialists are commanding unprecedented pay, with reports of Meta offering contracts worth up to $100 million.
But high compensation has not always ensured retention. Cultural clashes, constant restructuring, and limited autonomy are cited as reasons for exits.
Disruptions to Scale’s Business Model
Meta’s deal has also impacted Scale AI’s broader business relationships. Google—once a major client, reportedly spending around $200 million annually—reduced its work with Scale after the Meta partnership.
In response, Scale has cut staff, consolidated operations, and shifted its strategy toward enterprise and government clients.
