Meta 將斥資數千億美元興建 AI 資料中心

Meta, the parent company of Facebook and Instagram, has announced plans to invest hundreds of billions of dollars in building massive artificial intelligence data centers across the United States. Mark Zuckerberg, Meta's founder and CEO, revealed that the company is working on several large-scale projects, including two major data centers named Prometheus and Hyperion. These facilities are designed to support Meta's ambitious goal of developing 'superintelligence,' a form of AI that could surpass human intelligence. The first of these, Prometheus, will be located in New Albany, Ohio, and is expected to be operational by 2026. Hyperion, which will be built in Louisiana, is planned to be fully online by 2030.

The scale of these new data centers is enormous. Zuckerberg mentioned that one of the sites would cover an area nearly the size of Manhattan, which is about 59 square kilometers. The centers will be made up of multi-gigawatt clusters, with Hyperion alone potentially reaching up to five gigawatts over several years. These clusters are essentially groups of powerful computers that work together to process huge amounts of data, making them essential for advanced AI research and development. Meta has already invested heavily in AI, and these new centers represent a significant step forward in their efforts to lead the industry.

Meta's financial strength comes from its dominance in online advertising, generating over $160 billion in revenue in 2024. The company is now channeling this wealth into AI, aiming to build the most advanced technology in the world. Industry experts, like Karl Freund from Cambrian AI Research, believe that Zuckerberg is determined to make Meta the top player in AI by investing in the best talent and hardware. The announcement has already had a positive impact on Meta's stock, which rose by 1 percent and has increased by more than 20 percent this year. However, there are concerns about the environmental impact of these massive data centers. AI-driven facilities consume vast amounts of energy and water, with one study estimating that they could use 1. 7 trillion gallons of water globally by 2027. Even a single AI query, such as asking a question to ChatGPT, requires significant resources. Despite these challenges, Meta's new data centers are set to become some of the largest and most advanced in the world, positioning the company at the forefront of AI innovation.