Meta Platforms plans to invest heavily in building out its network of data centers, with Meta CEO Mark Zuckerberg posting on Threads that the company will construct “industry-leading levels of compute and by far the greatest compute per researcher.”
Daniel Newman, CEO of The Futurum Group, noting the depth of Meta’s investment, commented that “The Capex commits continue to come fast and furious,” on a Twitter post. Companies standing to benefit include NVIDIA, AMD and Broadcom, he added. “AI Bubble bears will continue to be wrong on this trend. The race to super intelligence will fuel a staging multi-year infrastructure investment. The enterprise number is barely baked into expectations at this point.”
Meta’s massive data center build out is part of its announced $64 to $72 billion on capital expenditures for 2025—a boost from its earlier announcement of $60 to $65 billion, which was a jump from the $35-40 billion annually the year prior. Much of this investment is earmarked for developing AI and the data infrastructure needed to support AI training and inference. Zuckerberg said in January that Meta would eventually commit hundreds of billions of dollars on staying ahead in the AI race.
Today’s announcement demonstrated the depth of the Meta CEO’s commitment. “We’re actually building several multi-GW clusters,” he wrote on Threads. “We’re calling the first one Prometheus and it’s coming online in ’26. We’re also building Hyperion, which will be able to scale up to 5GW over several years. We’re building multiple more titan clusters as well.”
Zuckerberg said the planned data centers will be multi-gigawatt clusters, which would mean these combined facilities will be at the scale of those operated by the largest hyperscalers, in some cases larger.
Describing how gargantuan the planned data centers will be, he added “Just one of these covers a significant part of the footprint of Manhattan.” With the post he added an image of the upcoming facility superimposed over a map of Manhattan. The data center is set to be built in Richland Parish, Louisiana and isn’t expected to be online until 2030. When completed it will be part of something very big: “Meta is on track to be the first lab to bring a 1GW+ supercluster online,” Zuckerberg wrote.
Meta is also investing heavily in the expertise required to make these data centers viable—and the numbers are even more staggering. “For our superintelligence effort, I’m focused on building the most elite and talent-dense team in the industry,” he noted. To hire top experts for what Zuckerberg calls a Superintelligence team, focused on dominating the AI sector, the typical offer is in the neighborhood of $200 million over 4 years. If nothing else, Zuckerberg—who’s personally involved in recruiting efforts—is exponentially rising the price of top AI talent for competitors.
Among his costly hires, he’s snatched up Daniel Gross, the CEO and co-founder of Safe Superintelligence, and Nat Friedman, former GitHub CEO. Additionally, Zuckerberg bought a 49 percent stake in Scale AI, and brought on its former CEO Alexandr Wang for Meta’s AI team.
