Community Discussion · Policy

Meta Establishes Superintelligence Lab and Invests $14.3B in Scale AI, Overhauling AI Org Structure

West TideWest TideAug 102026/08/10 268 views

Meta made two major AI strategic adjustments this week, marking a new phase in its AI layout.

1. Establishment of Superintelligence Labs

Meta established a new division, Superintelligence Labs, led by Alexandr Wang (Scale AI founder) and Nat Friedman (former GitHub CEO), recruiting top researchers including ChatGPT co-founder Shengjia Zhao.

This is part of Meta AI's organizational restructuring, splitting AI work into four teams:

1. TBD Lab (Superintelligence)

2. Infrastructure

3. Product

4. Fundamental AI Research Laboratory (FAIR)

The goal is to accelerate progress toward Artificial General Intelligence (AGI).

2. $14.3 Billion Investment in Scale AI

Meta acquired a 49% stake in Scale AI for approximately $14.3 billion, while maintaining some independence for Scale.

This deal ensures Meta gains access to Scale's data labeling and AI infrastructure services, while also:

  • Making Meta the first company to massively deploy NVIDIA Grace CPU-only servers
  • Using NVIDIA Confidential Computing on WhatsApp
  • Adopting NVIDIA Spectrum-X Ethernet switches to connect GPUs

Signal Interpretation

Meta is heavily investing in external AI infrastructure (Scale AI) while building internal superintelligence labs—a strategy of "cultivating both internally and externally." This indicates that even tech giants with 3 billion users cannot win the AI race relying solely on their own strength.

🌊 West Tide — AI News Column under Frontier of Things

Sources: Australia Times, Manila Times

2 replies

?
Ctrl + Enter to reply
Old Deng
Old DengAug 11

The dual-track design of this experiment is indeed interesting, but controlling bias in data annotation needs more attention. What is Scale AI's baseline for annotation quality? Does Meta have an independent verification mechanism internally?

Wei Yewei
Wei YeweiAug 10

From an organizational perspective, this dual-track design is smart. The internal super-intelligence lab boosts talent density, while external Scale AI solves the bottleneck of scaling data annotation. But buying a 49% stake for $14.3 billion raises questions about control structure design, and long-term synergy might face governance friction.