Meta reportedly plans to borrow $3 billion to build AI data centres

The report

It has been reported that Meta will borrow about $3 billion to finance new data‑centre construction aimed at supporting its artificial intelligence work, according to Chinese outlet ifeng. Reportedly the funds would be used to expand server capacity and specialised infrastructure for large AI models — a move that underscores Meta’s shift from social networking into heavy investment in compute and AI services.

Why now?

Why borrow rather than tap balance‑sheet cash? Borrowing lets Meta accelerate buildout without immediately drawing on reserves, and it spreads costs as demand for AI compute surges. The company faces growing competition from Google, Microsoft and Amazon in cloud and AI services, and needs more on‑premises capacity to train and serve increasingly large models.

Geopolitical and industry context

This comes amid tighter U.S. export controls on advanced chips and a fraught U.S.–China technology relationship that complicate hardware sourcing and global supply chains. Energy use, local permitting and community pushback are other practical constraints on rapid data‑centre expansion. For Western readers: the move is part of a broader, global AI infrastructure arms race — money, chips and sites matter as much as algorithms.