Meta to spend $70B on AI supercenters

Mark Zuckerberg has announced a significant investment strategy for Meta Platforms, committing "hundreds of billions of dollars" to develop massive AI data centers, referred to as "titan clusters." The first of these, named Prometheus, is expected to be operational next year, with plans for another, Hyperion, that can scale up to 5 gigawatts. This initiative aims to position Meta at the forefront of the race for "superintelligence," an AI that surpasses human capabilities.

Zuckerberg emphasized the strength of Meta's core advertising business as a foundation for this ambitious spending. The company has recently reorganized its AI efforts under a new division called Superintelligence Labs, following challenges with its Llama 4 model and the departure of key personnel. The new unit is expected to generate revenue through products like smart glasses and innovative advertising tools, and Meta is actively recruiting top AI talent, offering salaries exceeding $100 million.

For 2025, Meta has increased its capital expenditure to between $64 billion and $72 billion, with approximately $70 billion allocated specifically for AI infrastructure—almost double the previous year's budget. To expedite the construction process, Meta is even utilizing temporary "tent" data centers.

In addition to its internal investments, Meta has already put $14.3 billion into Scale AI and is seeking an additional $29 billion in private capital to accelerate its data center projects. The company is also pursuing clean energy agreements, including 791 megawatts of solar and wind power, to meet the growing energy demands of its infrastructure while aligning with sustainability goals.

While the announcement led to a slight increase in Meta's share price, analysts caution that achieving dominance in superintelligence through these investments is a long-term endeavor, and significant returns may not materialize in the near future.