Meta’s Chief Executive, Mark Zuckerberg, has announced an ambitious strategy to develop several large-scale datacenters that collectively will require multiple gigawatts of power. Set to commence their operations in 2026, these datacenters underscore Meta’s aim to be a leader in AI infrastructure.

The revelation follows an analysis report from SemiAnalysis suggesting that Meta is quietly building one of the biggest AI clusters in Ohio, dubbed Prometheus. This will be followed by another project named Hyperion, projected to scale up to 5GW over several years.

Zuckerberg used his social media platform to clarify the company’s plans, stating, “We’re establishing several massively scaled, multi-gigawatt operations. Each site is designed to accommodate the infrastructure needed for powerful superintelligence computing.”

Illustrating the project’s magnitude, a video was released showing that just one of the planned facilities could span almost the entirety of Manhattan. Such massive developments signify Meta’s push towards dominating AI capabilities.

Despite the grand scale and potential of these initiatives, Meta faces considerable challenges. Previously, the company invested approximately $60 billion into Reality Labs, targeting VR and metaverse products with limited financial returns. As Zuckerberg shifts focus towards AI, he assures that Meta’s resources will drive profound advancements in AI development and computing power.

While Meta enjoys vast financial capital, translating these resources into profitable ventures remains to be seen, especially given meta’s prior attempts in other high-tech domains.