OpenAI Stargate accelerates with five new US data centers

Last update: 24/09/2025

  • Five new data centers in the US under OpenAI's Stargate initiative with Oracle and SoftBank.
  • Planned capacity of nearly 7 GW and more than 400.000 billion euros in committed investment.
  • Oracle will develop three locations (TX, NM, and the Midwest) and an expansion in Abilene; SoftBank will develop two (Ohio and Texas).
  • Cash and debt financing, with support from NVIDIA chips and a possible deal of up to $100.000 billion.
OpenAI Stargate

OpenAI, together with strategic partners such as Oracle and SoftBank, has launched a ambitious deployment to raise five new data centers artificial intelligence in the United States under the Stargate brandThe move reinforces the commitment to creating the computing foundation needed for the next wave of large-scale AI services.

With these locations, The Stargate roadmap reaches a planned capacity of nearly 7 gigawatts and a committed investment of 400.000 over million, with a stated goal of reaching the 10 GW and 500.000 billion as the program progresses.

Locations and initial scope

OpenAI Stargate

Involvement Oracle infrastructures It includes three projects and a significant expansion, forming the core of the first phase of this expansion in the US.

  • Shackelford County, Texas, where the joint deployment of Oracle and OpenAI is already progressing.
  • Doña Ana County, New Mexico, which will strengthen the computing footprint in the Southwest.
  • An undisclosed location in the Midwest, selected to diversify latency and resilience.

To these is added a 600 MW expansion in Abilene (Texas). Together, Oracle-led developments will exceed 5,5 GW of capacity and are expected to boost the creation of around 25.000 direct jobs.

Exclusive content - Click Here  How to uninstall directx

For its part, Softbank will co-develop two other sites: Lordstown, Ohio y Milam County, Texas, with an initial deployment of 1,5 GW in 18 monthsThe Ohio site is already under construction and is targeted to be operational as of 2026, while the one in Texas will take advantage of the rapid construction model of SB Energy, the energy subsidiary of the Japanese group.

Capacity, schedule and objective

The joint plan pushes the Stargate's planned capacity up to almost 7 GW, with the goal of getting closer to the 10 GW as new locations and phases are confirmed. This scale equates, in terms of power, to the demand for large metropolitan areas.

The works already started in Ohio and the increase in activity in Texas and New Mexico They have established a schedule that prioritizes the rapid operational start-up of the first computing blocks, while also reserving room for modular expansions as demand and the supply chain mature.

Technology and energy

NVIDIA GB200 racks

Oracle highlights its infrastructure LEISURE equipped with racks NVIDIA GB200, already deployed in Abilene, as a pillar for scaling efficiently. This technical profile is key to accelerating the training and inference of advanced models, with improvements in density, efficiency and cost per computation, and in data management through the best compression format for copies and sending.

Involvement SB Energy The Texas site focuses on energy availability and rapid construction schemes, critical elements for supplying high-power data centers and stabilizing supply in the face of demand peaks.

Exclusive content - Click Here  How to Change Language in Word

Financing and alliances

The new centers will be funded by a combination of cash and debt. OpenAI is considering structures that include the GPU leasing and other asset financing formulas, while a possible investment agreement of up to $100.000 billion with NVIDIA, according to reports, would facilitate obtaining debt and securing chip capacity.

The Stargate Initiative was conceived as a LLC, but the brand has expanded to encompass data center agreements outside that entity, including collaborations with Oracle and other vendors, and the integration of secure connectivity solutions such as WireGuard. Also ongoing projects with CoreWeave They are added to the overall calculation that pushes capacity towards 7 GW.

Selection process and next steps

The program began a competitive process in January in which the following were reviewed: more than 300 proposals in 30 states. OpenAI is keeping new locations open for evaluation, so the number of 500.000 million associated with Stargate may not be a closed ceiling as computing demand consolidates.

What the protagonists say

In a joint message, the companies stressed that these sites will allow for a faster deployment, with greater Scalability and better cost efficiencies in the provision of AI capacity.

Sam Altman, CEO of OpenAI, insisted that the promise of AI depends on build the infrastructure of adequate computation and recalled that, although there will always be limitations, the goal is to reduce bottlenecks to continue delivering progress.

Exclusive content - Click Here  What to Do When Taskbar Disappears in Windows

From Oracle, responsible for LEISURE They emphasize that the effort is necessarily collaborative and that the combination of cloud capacity and cutting-edge hardware is enabling OpenAI to scale its business rapidly.

Masayoshi Son, president and CEO of SoftBank, highlighted that the project takes advantage of the energy experience of the group to drive a new era in which AI contributes tangibly to economic and social progress.

Impact on the sector

OpenAI Stargate Infrastructure

The thrust of Stargate is framed in a broader investment wave: Giants like are expected to Meta, Alphabet, Amazon and Microsoft allocate hundreds of billions to infrastructure this year, much of it directed to data centers already run AI models. This capability also supports massive services such as ChatGPT, which according to industry estimates already has hundreds of millions of weekly users.

At the same time, the voices that warn of a possible overheating of AI spendingThe companies involved maintain that mature business demand, financial discipline, and technological efficiency must act as anchors for sustainable deployment.

With five sites on the way, With a plan of nearly 7 GW and an investment exceeding $400.000 billion, Stargate is shaping up to be one of the largest AI infrastructure deployments in the US.., with distributed locations to build resilience and with technological and financial partnerships designed to accelerate implementation time without losing sight of long-term efficiency.

Related article:
China vetoes Nvidia's purchase of AI chips from its tech companies