Meta's Partnership with Scale AI Shows Signs of Fraying

Meta's partnership with data vendor Scale AI is showing signs of fraying. One of the executives brought over to run Meta Superintelligence Labs has departed after just two months. Additionally, researchers in TBD Labs, the core unit tasked with building AI superintelligence, are working with third-party data vendors other than Scale AI, citing the low quality of Scale AI's data. This is notable as Meta invested $14.3 billion in Scale AI in June.

Meta's $14.3 billion investment in data vendor Scale AI, announced in June, has already begun to face challenges. The company's partnership is showing signs of fraying, with one executive leaving after just two months and researchers in Meta's Superintelligence Labs (MSL) expressing dissatisfaction with the quality of Scale AI's data.

Ruben Mayer, the former Senior Vice President of GenAI Product and Operations at Scale AI, departed Meta after two months. Mayer oversaw AI data operations teams at Meta but was not part of the core TBD Labs team, which is tasked with building AI superintelligence [2].

Additionally, researchers at TBD Labs have begun working with third-party data vendors such as Mercor and Surge, citing the low quality of Scale AI's data. This is a notable development given Meta's substantial investment in Scale AI. While AI labs commonly use multiple data vendors, the reliance on competitors suggests that Meta is not fully confident in Scale AI's product [2].

The departure of Mayer and the preference for third-party data vendors raise questions about the value of Meta's investment in Scale AI. One current MSL employee stated that several Scale executives brought over to Meta are not working on the core TBD Labs team, and Meta is not exclusively relying on Scale AI for data labeling work [2].

This situation is particularly notable as Meta's AI unit has become increasingly chaotic since the arrival of Alexandr Wang and a wave of top researchers. New talent from OpenAI and Scale AI have expressed frustration with navigating the bureaucracy of a big company, while Meta's previous GenAI team has seen its scope limited [2].

The challenges faced by Meta's partnership with Scale AI come at a time when the company's AI development has been under scrutiny. The lackluster launch of Llama 4 in April led to frustration from Meta CEO Mark Zuckerberg with the company's AI team [2].

References:
[1] https://www.businessinsider.com/meta-superintelligence-team-hires-deepmind-scale-ai-2025-8
[2] https://techcrunch.com/2025/08/29/cracks-are-forming-in-metas-partnership-with-scale-ai/