Meta to invest $15B in Scale AI, deepening push into artificial intelligence

Meta Platforms Inc (NASDAQ:META, ETR:FB2A, SWX:FB) has agreed to invest $15 billion for a 49% stake in data-labeling startup Scale AI, according to multiple reports Tuesday.

The move marks one of the tech giant’s largest strategic bets to accelerate its artificial intelligence ambitions.

The deal, which values San Francisco-based Scale AI at around $30 billion, will give Meta significant access to the company’s data-labeling infrastructure without triggering antitrust scrutiny, according to the report.

The investment is designed to help Meta process and train large AI models using high-quality labeled data at scale.

As part of the agreement, Scale AI founder and CEO Alexandr Wang is expected to take a top role within Meta’s newly established AI research lab focused on "superintelligence," according to reporting from The New York Times. Meta has reportedly been in discussions to bring Scale AI employees over as part of the arrangement.

The investment signals a major escalation in Meta’s AI efforts as CEO Mark Zuckerberg seeks to catch up to rivals OpenAI and Google in developing advanced artificial general intelligence (AGI) systems. Bloomberg reported that Zuckerberg has been personally recruiting AI researchers from top labs, hosting candidates at his homes in Palo Alto and Lake Tahoe to build a new elite team.

Meta recently announced that its AI platform has reached 1 billion monthly active users. Still, Zuckerberg is said to be frustrated with the pace of progress, prompting the creation of the new lab and the push to hire a new head of AI research.

The company is already expected to spend $72 billion on capital expenditures this year, with AI infrastructure representing a large share of that budget. The Scale AI investment further underscores Meta’s commitment to expanding its foundational AI capabilities through data and talent acquisition.

Founded in 2016, Scale AI provides data labeling services critical for training machine learning models and counts major government agencies and private sector companies among its clients.