Unlock the Editor’s Digest for free
Roula Khalaf, Editor of the FT, selects her favourite stories in this weekly newsletter.
Meta plans to invest about $15bn in data-labelling start-up Scale AI and hire the group’s co-founder and top researchers, in one of the biggest deals of its kind as the Big Tech company seeks to catch up with rivals.
The deal, which could be announced as soon as Wednesday, would give Meta a 49 per cent stake in Scale AI and value the start-up at roughly $28bn, according to people with knowledge of the matter. It would mark the second consecutive year that Scale AI has doubled its valuation.
The investment in Scale AI and attempt to poach its top talent was part of Meta’s plan to build a “superintelligence” lab that would outperform OpenAI, Anthropic and Google, which are also developing models they claim will exceed human intelligence, according to one of the people.
Scale AI declined to comment, and Meta did not respond to a request for comment.
The launch of Meta’s latest large language model, Llama 4, underwhelmed critics after it underperformed on independent reasoning and coding benchmarks.
Meanwhile, competitors such as Google, OpenAI and Anthropic have each unveiled a new generation of powerful “reasoning” models, which solve problems by breaking them down step by step. Meta is also facing pressure from open source competitors such as China’s DeepSeek that have built powerful models for a fraction of the cost.
Meta, with a market capitalisation of nearly $2tn, has invested heavily in generative artificial intelligence. But progress has been halting and chief executive Mark Zuckerberg has reorganised the efforts multiple times. Meta announced in April the departure of Joelle Pineau, vice-president of AI research.
Alexandr Wang, a 28-year-old paper billionaire who co-founded Scale AI in 2016, is set to join Meta’s “superintelligence” lab, the details of which were first reported by The New York Times. Details of Meta’s investment were first reported by Bloomberg and The Information.
Scale AI’s core business involves manually labelling the data that is used to train advanced AI models to ensure it is accurate.
Wang has forged relationships with Silicon Valley’s biggest investors and technologists, including OpenAI’s Sam Altman, and has positioned Scale AI to serve companies developing autonomous vehicles and more recently those building generative AI models.
But his talents lie in promoting the company rather than managing its staff or furthering AI research, according to multiple people who have worked with him.
Jason Droege, who joined Scale AI from Uber Eats less than a year ago, was expected to step up from chief strategy officer to chief executive, according to a person with knowledge of the matter.
The fate of Scale AI’s remaining employees is less clear. Wang recently spoke about his desire to take the start-up public, but the potential deal with Meta casts uncertainty over that goal.
Scale AI had been attempting to broaden its revenue sources following investor concerns about its concentrated services, according to one person with knowledge of the matter. The group has increasingly focused on building custom applications for enterprises and bidding for government contracts.
Last year, Microsoft paid $650mn to hire Inflection boss Mustafa Suleyman and his top lieutenants, and to license the start-up’s technology. Google also paid $2.7bn for a similar arrangement with Character AI.
The bespoke structures used by the Big Tech groups were partly designed to avoid probes from regulators, according to people with knowledge of the deals. But Google and Microsoft have nonetheless faced scrutiny from antitrust enforcers.
Additional reporting by Hannah Murphy
