Recently, OpenAI, Oracle, and Japanese SoftBank jointly announced that they will build five artificial intelligence (AI) data centers in the United States. This ambitious project, named "Stargate," is expected to invest more than $400 billion over the next three years, with computing power potentially reaching 7 gigawatts and eventually expanding to 10 gigawatts, making it one of the largest AI data center projects in history.
Sam Altman, CEO of OpenAI, stated that the rapid development of artificial intelligence relies on powerful computing capabilities. He emphasized that only by building sufficient computing power can AI truly realize its potential.
Image Source Note: The image was generated by AI, and the licensing service provider is Midjourney.
According to the public plan, OpenAI and Oracle will build three new data centers in Travis County, Texas; Dona Ana County, New Mexico; and an undisclosed location in the central United States. At the same time, OpenAI, SoftBank, and its affiliated companies will establish two data centers in Lordstown, Ohio, and Miller County, Texas. This plan not only includes new facilities but also integrates the collaboration project with CoreWeave. The total computing power is expected to approach 7 gigawatts within three years, with an investment scale as high as $400 billion.
This project is expected to create about 25,000 direct jobs locally. It is worth noting that former U.S. President Trump publicly supported this project in January this year, emphasizing the importance of AI in the national economic and security strategy.
Meanwhile, NVIDIA also announced that it will invest up to $10 billion in OpenAI to supply millions of AI chips to support the construction of its next-generation data centers. According to the memorandum of understanding signed by both parties, the energy consumption of this cooperation is expected to exceed 10 gigawatts.
However, this massive investment has also raised some market doubts. Stacy Rasgon, an analyst at Bernstein, pointed out that such investments may increase the risk of "circular transactions," where AI companies rely on financing to expand data centers, thereby continuously purchasing NVIDIA products. He believes that the scale of this deal exceeds all previous cases, and the rationale behind it needs further review.
According to PitchBook, NVIDIA participated in over 50 AI-related venture capital transactions in 2024, and it is expected to break records this year. However, analysts are concerned that the huge bet on OpenAI may exacerbate the valuation bubble in the AI industry chain.
