Meta’s Latest Move: Unpacking the Impact of the Scale AI Deal

Meta Platforms Inc. added a big weapon to its arsenal in the artificial intelligence wars, revealing plans to invest $14.3 billion in Scale AI, the largest startup for labeling data used to teach automated systems new tasks.

While it may just be an investment, the $29 billion+ valuation of Scale AI is part of a deal that also brought Scale AI’s 28-year-old CEO, Alexandr Wang, on board to Meta’s growing “superintelligence” team. Wang will reportedly be overseeing Meta’s AI work and will remain on the board at Scale AI, with interim CEO duties at Scale falling to Jason Droege.

The huge bet, the second biggest such acquisition in Meta’s history behind WhatsApp, highlights CEO Mark Zuckerberg’s recent focus on accelerating Meta’s AI efforts and overcoming criticism that the company has been falling behind rivals like Google and OpenAI.

Scale AI is a key player in delivering huge volumes of high-quality labeled data necessary to train advanced AI models, such as large language models (LLMs), of which Meta’s Llama series are examples.

But the deal has already sent shock waves through the A.I. industry. Google’s parent, Alphabet, Scale AI’s largest customer, reportedly plans to disconnect from the data-labeling company.

The decision, which affects some $200 million in budgeted spending for 2025, comes amid concerns among rival AI companies that Meta’s large ownership could expose their own research priorities and roadmaps. Microsoft and Elon Musk’s xAI are also said to be reassessing their collaborations with Scale AI.

While the departure of key clients can be a significant blow to AI labs, it illustrates a larger conundrum about how AI labs handle quality data annotation versus the value they place on protecting its IP in a competitive marketplace. While the deal is a win for Scale AI’s current investors, as well as employees, it represents a turning point where neutrality among AI data services can’t be taken for granted.

Rivals like Turing and Labelbox are preparing for an influx of business from companies that want alternatives to data-labeling services that are secure.

Meta’s investment, strategic for its “superintelligence” ambitions, also treads into the sharps of antitrust scrutiny insofar as it is a minority nonvoting stake. In the months ahead, the extent of the effect this deal will have on the broader AI ecosystem, as all companies adjust to the changed dynamics for data partnerships and the race for better AI, will become apparent.