On Tuesday, September 23, OpenAI, Oracle, and SoftBank announced their plan to establish five large AI data centers in the United States as part of the ambitious “Stargate” project.
OpenAI stated that they would collaborate closely with Oracle and SoftBank. The three new data centers in partnership with Oracle will be located in Shackelford County, Texas, Doña Ana County, New Mexico, and an undisclosed location in the Midwest.
On the other hand, SoftBank will develop two additional sites in Lordstown, Ohio, and Milam County, Texas. Construction has already begun in Ohio, with the new data center expected to start operations next year.
The expansion of these new data centers is expected to increase the total capacity of the “Stargate” data centers to nearly 7 gigawatts in the next three years, driving investments of over $400 billion.
With the announcement of these projects, it is anticipated that the goal of “10GW, $500 billion” set earlier in the year will be achieved ahead of schedule by the end of 2025.
In January, President Trump convened top tech executives to launch the “Stargate” project. This privately-led initiative aims to invest up to $500 billion in building AI infrastructure to ensure America’s competitiveness in future AI battles.
The role of AI in sensitive areas such as national defense, coupled with Beijing’s efforts to catch up, has made this emerging technology a top priority for the Trump administration.
OpenAI emphasized that the nationwide site selection process was rigorous, reviewing over 300 proposals from more than 30 states. These new sites are expected to create over 25,000 on-site job opportunities and generate tens of thousands of additional jobs nationwide, bringing significant economic benefits to local communities.
Currently, OpenAI is continuing to evaluate more potential building locations. The company stated that as they progressively fulfill and surpass the commitment to invest $500 billion, more sites will be announced in the future.
Nvidia also announced on Monday that they would invest $100 billion in supplying data center chips to OpenAI.
In recent times, there have been growing concerns among critics that the surge in AI capital expenditure could fuel market bubbles, drawing parallels to the dot-com bubble of 25 years ago. However, the rapid progress and substantial investments in the “Stargate” project not only demonstrate the industry’s strong drive but also reflect the transition of AI technology from conceptual to large-scale physical infrastructure development.
OpenAI CEO Sam Altman acknowledged signs of a bubble but also expressed long-term belief in the technology.
“We will push forward on infrastructure because it will enhance our ability to develop advanced technologies, as well as basic products and services,” Altman said at a press conference.
“Only by establishing the computational power that drives it can AI fulfill its promise. This computational power is essential to ensure everyone benefits from AI and unlock key breakthroughs for the future. Through ‘Stargate,’ we have made historic progress towards this goal and are rapidly taking action not only to fulfill our initial commitments but also to lay the groundwork for future developments,” Altman added.
Oracle CEO Clay Magouyrk echoed these sentiments.
“To meet this substantial demand, we will continue to expand the landscape of cloud infrastructure at an unparalleled pace to provide the highest performing and cost-effective AI training and inference,” Magouyrk said. “This work is not something a single company can accomplish alone.”