Meta Doubles Down on AI Superintelligence With Massive Data Center Investment

Meta Platforms is dramatically scaling up its artificial intelligence ambitions, with CEO Mark Zuckerberg revealing that the company will invest “hundreds of billions of dollars” to build several large-scale AI data centers dedicated to developing what he calls superintelligence.

In a post shared on his Threads platform, Zuckerberg said Meta’s first multi-gigawatt facility, named Prometheus, is expected to go online in 2026, with another center, Hyperion, designed to scale up to 5 gigawatts in the coming years. “We’re building multiple more titan clusters as well,” he wrote, adding that just one of these clusters will cover “a significant part of the footprint of Manhattan.”

Meta (NASDAQ: META), which generated nearly $165 billion in revenue in 2023, has faced persistent investor skepticism over whether its AI investments will deliver meaningful returns. According to SemiAnalysis, Meta is on track to be the first AI lab to deploy a supercluster exceeding one gigawatt of compute power. The company reorganized its AI operations last month under a new unit, Superintelligence Labs, after facing setbacks with its open-source Llama 4 model and key departures on the research team.

Zuckerberg highlighted the strength of Meta’s core advertising business to justify the massive expenditure, stating that improvements in AI have already boosted ad sales volume and pricing power. Shares of Meta were up about 1% following the announcement and have gained more than 20% year-to-date, according to LSEG data.

D.A. Davidson analyst Gil Luria noted that while Meta’s near-term AI gains are evident in its ad business, its aggressive capital spending is more about securing a leading position in the race to build the world’s most advanced AI systems — a goal that could take years to materialize. Meta raised its capital expenditure forecast for 2025 to between $64 billion and $72 billion in April, up sharply as it ramps up competition with rivals OpenAI, Google, and Microsoft (LSEG).

To support this ambition, Meta has also been on a talent acquisition spree. In recent weeks, Zuckerberg has personally led recruitment efforts for the Superintelligence Labs, bringing in Scale AI co-founder Alexandr Wang and former GitHub CEO Nat Friedman, following Meta’s $14.3 billion investment in Scale AI earlier this year.

The company is betting that the new AI division will generate fresh revenue streams beyond ads, with potential products ranging from the Meta AI assistant app to image-to-video ad creation tools and smart wearables like augmented reality glasses. Meanwhile, a report from the New York Times suggests the unit may abandon its open-source “Behemoth” model in favor of a closed, proprietary alternative — a sign that Meta is willing to pivot strategies to stay ahead in the AI arms race.

While Meta’s plans promise huge computational power — with some clusters covering footprints the size of Manhattan — they also mark one of the boldest bets yet on the future of AI superintelligence. Whether these massive investments ultimately deliver a competitive edge or become a costly misfire remains to be seen.