Meta plans to invest $15bn in Scale AI in bid to catch up to rivals

Digest opened free editor

Rola Khaleda, FT editor, chooses her favorite stories in this weekly newsletter.

Meta plans to invest about $ 15 billion in the allowance for the permissible data starting and employ the co -founder of the group and senior researchers, in one of the largest deals of its kind as the large technology company seeks to catch up with competitors.

The deal, which can be announced after Wednesday, will give META 49 percent stake in Scale AI and the startup evaluation of about 28 billion dollars, according to people who have knowledge of the issue. In the second year in a row, this will coincide with Amnesty International’s evaluation.

Investing in Scale AI and trying to fish in his upper talent was part of the Meta plan to build a “Supertelligence” laboratory that surpasses Openai, Noteropic and Google, which also develops models that they will go beyond human intelligence, according to a person.

Scale AI refused to comment, and Meta did not respond to a comment.

The launch of the latest major linguistic model in Meta, Llama 4, which was enjoyed after it was a weak performance in independent standards of thinking and coding.

Meanwhile, both competitors such as Google, Openai and Anthropic have revealed a new generation of strong “thinking” models, which solve problems by dividing them step by step. Meta also faces pressure from the competitors of open sources like Chinese Deepseek, which designed strong models of a small part of the cost.

Meta, with a market value of about 2 Trin, has invested significantly in obstetric artificial intelligence. But progress has stopped, and CEO Mark Zuckerberg has re -efforts several times. Mita announced in April the departure of Joel Pino, Vice president of Artificial Intelligence Research.

Alexander Wang, a 28 -year -old billionaire who participated in the establishment of Scale AI in 2016, is scheduled to join the META “Super Wanvelign” laboratory, which the New York Times informed the first time. Meta investment details were first reported by Bloomberg and Information.

Scale Ai Core Business manually includes the data of the data used to train advanced artificial intelligence models to ensure their accuracy.

Wang has forged relationships with the largest investor and technician in the Silicon Valley, including Sam Altman from Openai, and Amnesty International has put the measure to serve companies that develop independent vehicles and recently those who build models of artificial intelligence.

However, his talents lie in promoting the company instead of managing its employees or strengthening artificial intelligence research, according to several people who worked with him.

Jason Drewig, who joined the Amnesty International scale from Uber eating less than a year ago, is expected to ascend from the chief strategy officials to the CEO, according to a person aware of the issue.

The fate of the remaining employees of artificial intelligence is less clear. Wang recently talked about his desire to take the emerging audience, but the potential deal with Meta raises uncertainty about this goal.

Scale AI was trying to expand its revenue sources after the investor’s concerns about its concentrated services, according to one person to know the matter. The group has increasingly focused on building requests for institutions and providing bids to government contracts.

Last year, Microsoft paid $ 650 million to hire the coach of the turn of Mustafa Suleiman and his largest lieutenant, and to license the startup technology. Google also paid $ 2.7 billion to a similar arrangement with AI for the letter.

The detailed structures used by large technology groups are partially designed to avoid investigations by organizers, according to people with deals. But Google and Microsoft, nevertheless, were scrutinized by the anti -monopoly perpetrators.

Additional reports from Hana Murphy