Zuckerberg unveils Meta’s multi-gigawatt AI data clusters
Zuckerberg is scaling Meta’s AI capabilities by constructing huge data centres in Louisiana and beyond, prioritising internal infrastructure investment instead of outsourcing storage needs.
Meta Platforms is building several of the world’s largest data centres to power its AI ambitions, with the first facility expected to go online in 2026.
Chief Executive Mark Zuckerberg revealed on Threads that the site, called Prometheus, will be the first of multiple ‘titan clusters’ designed to support AI development instead of relying on existing infrastructure.
Frustrated by earlier AI efforts, Meta is investing heavily in talent and technology. The company has committed up to $72 billion towards AI and data centre expansion, while Zuckerberg has personally recruited high-profile figures from OpenAI, DeepMind, and Apple.
That includes appointing Scale AI’s Alexandr Wang as chief AI officer through a $14.3 billion stake deal and securing Ruoming Pang with a compensation package worth over $200 million.
The facilities under construction will have multi-gigawatt capacity, placing Meta ahead of rivals such as OpenAI and Oracle in the race for large-scale AI infrastructure.
One supercluster in Richland Parish, Louisiana, is said to cover an area nearly the size of Manhattan instead of smaller conventional data centre sites.
Zuckerberg confirmed that Meta is prepared to invest ‘hundreds of billions of dollars’ into building superintelligence capabilities, using revenue from its core advertising business on platforms like Facebook and Instagram to fund these projects instead of seeking external financing.
Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot!
