The Birth of the Manhattan-Class Data Center: The Full Scope of Zuckerberg's "Prometheus Project"

The Birth of the Manhattan-Class Data Center: The Full Scope of Zuckerberg's "Prometheus Project"

On July 14, 2025, Mark Zuckerberg, CEO of Meta Platforms, announced on Threads that the company would invest hundreds of billions of dollars to achieve "superintelligence." As the first phase, they plan to construct a 1 GW-class cluster named "Prometheus," set to be operational in 2026, and expand it to 5 GW with "Hyperion" after 2030. Over a million NVIDIA H100/H200 units will be deployed, with the newly formed in-house organization "Superintelligence Labs" leading the research. Despite the massive expenditure, the stock price saw a slight increase. Bank of America raised its target stock price, while environmental groups criticized the power consumption and carbon footprint. On social media, there was a mix of welcoming the initiative as the "Avengers of AI" and condemning it as "greenwashing." IDC estimates that superintelligence will impact various fields, including advertising, healthcare, education, and the metaverse, with a potential $5.7 trillion impact on the global economy. Balancing technical feasibility and social acceptance will be the focus moving forward.