Meta to Invest Hundreds of Billions of Dollars More in Artificial Intelligence Infrastructure
Zuckerberg’s Meta to Invest Billions in AI Infrastructure, Aiming for “Superintelligence”
Meta’s CEO, Mark Zuckerberg, plans to invest “hundreds of billions of dollars” in artificial intelligence (AI) infrastructure, the latest major investment announcement aimed at achieving his stated goal: building a “superintelligence.”
“We are building computer networks with several gigawatts of power,” he announced on Monday, July 14, in a message posted on his social networks Facebook and Threads. These data centers, far larger than those already existing to house cloud servers, are specifically designed to train and host generative AI models. These models require enormous computing power, and therefore cutting-edge computer chips and a lot of energy.
Meta has fallen behind in generative AI, dominated by OpenAI since the launch of ChatGPT, and Google. All industry leaders have their sights set on a so-called “general” AI, or a “superintelligence,” with cognitive abilities that would be superior to those of humans, capable of making scientific discoveries and inventing technologies itself.
“The first [computer network], which we call Prometheus, will be commissioned in 2026. We are also building Hyperion, which will be able to reach a capacity of 5 gigawatts over the years,” said Mark Zuckerberg. A power of 5 gigawatts represents the equivalent of the annual electricity consumption of 1 to 4 million American households.
“The Most Talent-Dense Team”
In mid-June, the social media giant spent more than $14 billion to acquire 49% of the capital of Scale AI, which specializes in preparing data used to develop AI models.
According to OpenAI’s CEO, Sam Altman, Meta also offered an individual signing bonus of more than $100 million to “many” of his company’s employees and about the same amount annually in salary.
Mark Zuckerberg then confirmed in an internal memo that no fewer than seven of them had changed companies, as had the CEO of Scale AI, Alexandr Wang, and several members of rivals Anthropic and Google.
The billionaire repeated on Monday that he intended to build “the most elite and talent-dense team in the entire industry,” and that it would have “a level of computing power unparalleled in the sector.” “We have the capital from our activities to do it,” he assured.
Meta’s latest major AI model, Llama 4, launched in early April, was disappointing. It lags behind all the heavyweights, American, Chinese and French, in the ranking established by the independent evaluation platform LMArena on code writing, and even behind its predecessor Llama 3 for the text interface.
Enjoyed this post by Thibault Helle? Subscribe for more insights and updates straight from the source.
