Zuckerberg makes the biggest bet of meta on AI, with $14 billion AI deal

Mark Zuckerberg is very frustrated Meta Those familiar with the issue remain in artificial intelligence that is willing to spend billions of dollars to convince AI CEO Alexandr Wang of scale to join his company.

Meta is finalizing a deal that would invest $14 billion in scale AI, according to anyone familiar with the issue they asked not to name because the terms are confidential. Bloomberg reported earlier this week that the investment could exceed $10 billion, and stories from information Tuesday said Meta would pay nearly $15 billion.

As the founder of one of AI's most well-known startups, Wang has built a reputation as an ambitious leader who understands the technical complexity of AI and how to build a business that is not simply focused on research. Zuckerberg is relying on Wang to better carry out Meta's AI ambitions after the launch of the company's latest Llama AI model, lukewarm water.

By not directly obtaining scale AI, META appears to be taking a similar strategy to companies like Google and Microsoftbrought prominent AI leaders from startup characters. Meta is currently in trial against the Federal Trade Commission over an antitrust claim, and the company doesn't want to further disrupt regulators by obtaining AI of size, according to people familiar with the issue.

As part of the transaction, Meta will acquire 49% stake in the data label and annotation startup, and Wang will lead the social networking company's new AI research lab, joining some of his colleagues. The New York Times first reported on the new AI lab.

Founded in 2016, Scale AI splashed in the age of generated AI by helping major tech companies such as Openai, Google and Microsoft prepare the data they use to train cutting edge AI models. According to two people familiar with the issue, Meta is one of AI's biggest customers.

The startup, which was valued in a funding round for $14 billion about a year ago, ranked 28th on CNBC's Disruptor 50 list. In mid-2024, the company recently signed one of its most recent commercial leases in San Francisco, gaining about 180,000 square feet of space in its downtown building. Airbnb.

Scale AI has been increasingly reciprocating towards the defense industry, and in March announced a multi-million dollar contract with the Department of Defense. In November, he said, “We worked with Defense Lalama, a custom version of Meta's open source Llama Foundation model, specifically designed to support the US national security mission.

Meta and Scale AI declined to comment.

Heading into 2025, AI has been one of the top priorities of the meta. But Zuckerberg has upset that rivals like Openai appear to be ahead in both the underlying AI models and consumer apps, a current and former meta employee said.

Zuckerberg is stripping Meta of its basic artificial intelligence research unit or justice in favor of more product-oriented Genai teams to advance with AI and improve the Llama family of AI models.

Meta's release of the Llama 4 AI model in April has not received good reviews with developers, people said, making Zuckerberg even more frustrating. At the time, Meta said it would only release two smaller versions of the Llama 4, and eventually release a bigger, more powerful “Behemoth” model.

The model has not yet been available due to Zuckerberg's concerns about its capabilities for competing models, people said. There are concerns, particularly about how Behemoth likes the wider developer community for the latest in companies like Openai and China's Deepseek.

Following the inactive debut of the Llama 4, Meta has reorganized the Genai unit and split it into two. Connor Hayes, a longtime Meta employee, will be in charge of AI products, and the AGI foundation was given to Amir Frenkel, previously vice president of engineering and products for Meta's Reality Lab Hardware Unit, and Ahmad Aldar, former head of Jenai.

The new position as Al-Dahle's co-leader was seen as a sign that Zuckerberg had lost confidence in him, people said.

Zuckerberg praised Wang, people said he believes he can play a major role in Meta as an AI leader. A dropout from the Massachusetts Institute of Technology, Wang has built quite a bit of business and is well versed in the technical complexities of AI. People describe the king as a “wartime CEO” in line with Zuckerberg's position where the US faces increasing competition with China, and therefore need the help of the tech industry.

Wang said in January that CNBC had believed there was an “AI war” between the US and China, and that the US would need more computing power to compete.

“The US will need huge amounts of computing power and huge amounts of infrastructure,” Wang said at the time. “We need to unleash our energy to make this AI boom possible.”

This is a rare move for Zuckerberg, who has traditionally placed loyalists in high positions. But it shows the magnitude of the moment, indicating Zuckerberg's belief that prominent outsiders like King could be superior to current META employees to strengthen the company's position in AI.

Wang also brings a lot of external knowledge about how competitors like Openai are building consumer chatbots and AI models. The increasing capabilities of AI models have made data labeling and training more complicated in recent years, said Vahan Petrosyan, CEO of Superannotate, one of the AI ​​scale competitors.

“I think the scale covers 70% of all models that are built,” Petrosian said. With King and others in Scale AI, Meta has “gained collective intelligence on how to build a better ChatGPT.

“When Meta is buying them, they are buying their intelligence,” Petrosian said.

clock: Zuckerberg Lobby Tramp's Mark to avoid meta-antitrust tests