Meta Allocates $14 Billion to Scale AI to Enhance Model Training

Meta, the owner of Facebook, has completed the acquisition of a 49% investment in data-labeling startup Scale AI for US$14.3 billion, as reported by two informed sources.

Scale AI CEO Alexandr Wang testifies during a 2023 House Armed Services Subcommittee hearing in Washington, D.C. Photo: Drew Angerer/Getty Images

Scale announced late Thursday that the agreement values the company at $29 billion, and its CEO, Alexandr Wang, will join Meta to assume a significant position in the Facebook owner's AI strategy.

Jason Droege, Scale's chief strategy officer, will assume the role of interim CEO, according to Scale AI. The corporation failed to respond to a request for further information regarding the financial specifics of the transaction.

The agreement would be classified as Meta's second-largest acquisition, following its $19 billion purchase of WhatsApp. The sources opted for anonymity due to the confidential nature of the material.

By recruiting Wang, who lacks a research background yet successfully established one of the largest enterprises in providing critical data for AI model training, Meta CEO Mark Zuckerberg is wagering that an accomplished business leader, akin to OpenAI's Sam Altman, can revitalize Meta's AI initiatives more effectively than the research scientists leading most rival laboratories.

As numerous laboratories have engaged Scale for data services, Zuckerberg might gain an advantageous insight into his competitors' data priorities, a crucial element in the development of contemporary AI models.

In a funding round in May 2024, Scale was appraised at approximately $14 billion, securing $1 billion from investors such as technology leaders Nvidia, Amazon, and Meta.

Established in 2016, Scale AI delivers huge quantities of precisely labeled data, essential for training advanced technologies such as OpenAI's ChatGPT.

To achieve this, Scale established subsidiary platforms like Remotasks and Outlier to recruit and oversee gig workers responsible for manual data labeling.

Wang, the 28-year-old co-founder, has guided the company to deliver data labeling services across prominent technological areas, ranging from autonomous vehicles in previous years to generative AI currently.