Meta is investing billions in superintelligence development with new data center projects
Meta's CEO Mark Zuckerberg has announced the construction of two massive data centers to power the company's superintelligence ambitions. These facilities, named Prometheus and Hyperion, will generate gigawatt-scale energy for AI training and operations.
According to BBC reports, the Prometheus facility will be located in New Albany, Ohio with 2026 deployment expected. Hyperion's Louisiana site is projected to launch in 2030. Zuckerberg also mentioned additional large-scale computing clusters in a Threads post, including one covering a Manhattan-sized area (59 km²).
The Hyperion project aims to expand to 5 gigawatts (GW), which would make it one of the world's largest data centers. Currently, the largest single facility is China Telecom's Inner Mongolia campus at 0.15 GW capacity. Competitors like Oracle, Google, OpenAI, and Amazon are also developing multi-gigawatt AI infrastructure. Shark Tank investor Kevin O'Leary is building what could become the world's largest AI data center in Canada.
Environmental and community risks from mega data centers
A single gigawatt of power can supply approximately 750,000 homes, but large data centers strain electricity grids and cause blackouts. They also consume significant water resources and contribute to air pollution, leading to local opposition. The New York Times reported that Meta's Newton County Georgia data center accounts for about 10% of the region's daily water usage. Residents now face water pressure issues from sediment buildup and must pay for mitigation costs while struggling to sell their homes.
A company representative managing Meta's data center complex told the New York Times that the water issues might coincidentally coincide with facility construction. New applications request up to 6 million gallons daily - exceeding the county's total water use. However, rejecting these requests proves challenging due to the potential millions in tax revenue they bring. Regulatory oversight is difficult as operators often remain opaque about water and energy consumption.
Zuckerberg's AI ambitions for Meta
Though Meta has pursued AI since 2013, the ChatGPT era revealed competitive shortcomings against OpenAI and Google. Strategic missteps included open-sourcing models that enabled DeepSeek to create more advanced alternatives. The latest Llama model underperformed, AI chatbots gained negative reputations, and top engineers departed.
Recently, Zuckerberg has heavily invested in reversing this trend. He established the Meta Superintelligence Lab, reportedly offering $100 million sign-on bonuses to lure engineers from competitors like OpenAI and Google DeepMind. The project is led by Scale AI founder Alexandr Wang and former GitHub CEO Nat Friedman. Zuckerberg also allegedly secured Apple's top AI executive Peng Ruming with a $200 million compensation package. The team's first major decision involved abandoning Meta's powerful open-source Behemoth model for closed-source alternatives that cannot be stolen.
"The Meta Superintelligence Lab will have industry-leading compute capabilities, with more computational power per researcher than any other organization," Zuckerberg wrote on Threads.
Can Meta's billions win the superintelligence race?
OpenAI CEO Sam Altman suggested Meta lacks genuine commitment to its superintelligence goals, claiming the company will eventually pivot to "their next pet project." After starting as Facebook, Meta maintains dominant social media control while having lost $60 billion in metaverse investments.
However, Zuckerberg's memo outlining lab recruits stated Meta has "unique advantages" for superintelligence development, including financial resources for massive compute needs, a proven track record reaching billions of users, AI hardware experience through smart glasses, and corporate structures allowing major experiments. In his Threads posts, Zuckerberg reaffirmed, "We have capital from our business to do this." Meta generated over $160 billion in 2024 revenue primarily from social media advertising.
