Zuckerberg’s Mega AI Plan Targets AGI Race

Meta CEO Mark Zuckerberg has announced an ambitious initiative to invest hundreds of billions of dollars into constructing a vast network of advanced AI data centers. The move is part of the tech giant’s bid to dominate the global pursuit of artificial general intelligence (AGI), or what Zuckerberg calls “superintelligence.”

The plan, shared in a post on Threads, highlights Meta’s increasing efforts in artificial intelligence as it competes with OpenAI and Google for leadership in infrastructure and talent.

Among the new developments are two mega-scale data centers: Prometheus, expected to go live in 2026, and Hyperion, which may reach a capacity of 5 gigawatts. These facilities are designed to power Meta’s so-called “titan clusters.”

“We’re building multiple more titan clusters as well. Just one of these covers a significant part of the footprint of Manhattan,” Zuckerberg stated.

Referencing an analysis by SemiAnalysis, he emphasized that Meta could soon become the first company to operate a gigawatt-plus AI supercluster.

Zuckerberg defended the vast spending by pointing to Meta’s robust advertising revenues: “We have the capital from our business to do this.”

The company recently consolidated its AI initiatives under a new division, Superintelligence Labs, following internal disruptions tied to Llama 4 and several key researcher departures. Alexandr Wang, former Scale AI CEO, and Nat Friedman, ex-GitHub chief, are now steering the division.

Meta has revised its 2025 capital expenditure estimate to between $64 billion and $72 billion. The company anticipates AI will significantly boost revenue through products like Meta AI, advertising technology, and smart devices such as AR glasses.

Analysts remain divided—some lauding Meta’s recent advertising performance powered by AI, while others warn that such large-scale investments are geared toward long-term supremacy and may take years to yield measurable returns.