Leadership exits, shifting vendor allegiances, and quality concerns signal trouble in Meta’s bold AI partnership with Scale AI.
A High-Stakes Partnership Under Strain
In June 2025, Meta invested $14.3 billion in Scale AI, integrating CEO Alexandr Wang and several top Scale executives into its newly-formed Meta Superintelligence Labs (MSL). The goal: build next-generation AI superintelligence. Just months later, cracks are beginning to show.
- Ruben Mayer, a key hire and former SVP at Scale AI, exited Meta after only two months.
- Conflicting narratives emerged around Mayer’s role in TBD Labs, the core AI research team.
- Mayer claims he was part of TBD Labs “from day one” but left for personal reasons, not internal conflict.
While early executive exits alone aren’t unusual, the discrepancy in public messaging raises questions about Meta’s internal alignment.
Meta Turns to Competitors Amid Quality Concerns
Despite its multi-billion dollar stake in Scale AI, Meta is actively sourcing data from other vendors — a surprising move in the AI ecosystem.
- Meta has long used data vendors Surge and Mercor, but TBD Labs’ preference for them over Scale is telling.
- Several sources cited “low-quality data” from Scale as a major concern, especially as AI systems demand domain-specific, expert-labeled datasets.
Scale AI initially built its business using a low-cost crowdsourcing model, but this model has struggled to scale with the evolving demands of advanced generative AI.
Competitors Thrive on High-Skill Foundations
Scale AI is attempting to shift strategies via its Outlier platform, which recruits experts like doctors and lawyers to provide high-fidelity labels. But it may be too late.
- Surge and Mercor, founded on expert-first data labeling models, are gaining ground.
- Meta’s decision to continue and possibly expand partnerships with those competitors underlines a lack of confidence in Scale AI’s current offerings.
This pivot suggests that Meta is hedging its bets — maintaining ties with Scale while diversifying to avoid dependence.
Strategic Motivations Behind the Investment
Some insiders believe Meta’s investment was less about Scale AI’s core product and more about recruiting Alexandr Wang.
- Wang, though not an AI researcher himself, brought credibility and helped attract top talent from rivals like OpenAI and Anthropic.
- However, multiple Scale hires at Meta reportedly don’t have direct roles in TBD Labs, raising concerns about their influence.
Wang’s appointment may have been a strategic move to fast-track recruitment, but it’s unclear whether this approach will deliver long-term value.
Internal Friction Threatens Meta’s AI Ambitions
Since forming MSL, Meta’s AI division has experienced turmoil.
- New hires from OpenAI and Scale AI are reportedly frustrated by Meta’s internal bureaucracy.
- The previous GenAI team has seen its mandate curtailed, leading to multiple high-profile departures, including:
- Rishabh Agarwal, AI researcher
- Chaya Nayak, product management director
- Rohan Varma, research engineer
These exits point to growing instability and a potential culture clash between existing Meta teams and new arrivals.
A Race Against Time
After the underwhelming launch of Llama 4, Meta CEO Mark Zuckerberg took swift action: restructuring teams, poaching rival talent, and investing heavily in infrastructure.
- Meta is building massive data centers, including the $50B Hyperion in Louisiana.
- The company aims to launch its next-gen AI model by year-end, but lingering instability may delay progress.
Wang’s leadership, combined with Zuckerberg’s urgency, created early momentum. Yet, whether that can sustain execution at scale remains in doubt.
What’s Next for Meta and Scale AI?
As Meta ramps up its AI infrastructure and talent acquisition, the long-term value of its Scale AI investment remains murky.
- Meta is clearly diversifying data sources, despite public claims of partnership strength.
- Scale AI, meanwhile, has suffered setbacks: losing OpenAI and Google as clients and laying off 200 employees in July.
- The startup is now pivoting toward government contracts, recently securing a $99M deal with the U.S. Army.
The coming months will test whether Meta can stabilize MSL, retain talent, and deliver on its superintelligence vision—or whether its largest AI bet to date unravels further.
