OpenAI, Oracle, and SoftBank power Stargate with five US data centers.

  • Five new AI data centers in the US under the Stargate project, with three led by Oracle and two with SoftBank.
  • Planned capacity close to 7 GW and more than $400.000 billion committed in the short term.
  • Locations: Shackelford, Texas, Doña Ana, New Mexico, a Midwest site, Lordstown, Ohio, and Milam County, Texas.
  • Mixed financing (cash and debt) and collaboration with Nvidia for chips; 25.000 jobs are expected to be created.

Stargate Project AI Data Centers

OpenAI, Oracle and SoftBank have taken a step forward with five new artificial intelligence data centers in the United States as part of its joint Stargate initiative. With this wave of projects, the consortium is accelerating its roadmap to deploy large computing capacity and support the next generation of AI services.

According to the companies, the expansion will raise the planned infrastructure to almost 7 gigawatts (GW) of power and an investment volume of more than 400.000 billion dollars in the short term, with the aim of moving towards the announced goal of 500.000 billion and 10 GW under the Stargate umbrella. The planned infrastructure seeks to accelerate operational deployment in different regions.

Where the new facilities will be located

AI infrastructure from OpenAI, Oracle, and SoftBank

Three sites will be developed with Oracle, within the OpenAI and Oracle cloud agreement: Shackelford County (Texas), County of Doña Ana (New Mexico) and an as yet undisclosed location in the MidwestThese plants, together with a possible expansion of 600 MW near the Abilene, Texas, campus, account for the bulk of the planned power increase.

The other two centers will be developed by SoftBank: lordstown (Ohio) and Milam County (Texas). The Ohio project has already begun construction and is expected to begin operations next year, while the Texas location will rely on the energy expertise of SB Energy to speed up construction.

Together, the facilities promoted with Oracle will provide more than 5,5 GW of new capacity, and those developed with SoftBank will scale up to 1,5 GW in approximately 18 months, figures that place the project on the path of its ambitions.

The consortium notes that this selection of sites is the result of a national process initiated in January, during which they were evaluated more than 300 proposals from more than 30 states, and which are still being studied additional locations for future phases.

Capacity, schedule and scope

Stargate Project: Data Centers in the United States

With the new announcements, Stargate aims for a total capacity close to 7 GW In the medium term, combining ongoing projects and planned expansions, to approach the 10 GW target. This scale, comparable to the energy demand of several cities, is designed to support the deployment of AI models and services increasingly demanding.

OpenAI has indicated that the committed investment already exceeds the 400.000 million dollars over the next few years, aligning with the overall goal of 500.000 million communicated by the companies involved. The initiative was presented after a meeting held in January with leaders of the technology sector, a movement aimed at accelerate critical infrastructure of AI in the US

In addition to the new quintet of centers, Stargate includes the expansion of the Oracle-OpenAI hub in Abilene, Texas and ongoing collaborations with CoreWeave, reinforcing the diversification of partners to scale high-performance computing.

The companies have stressed that the Stargate brand is articulated both through a structure LLC as well as through data center agreements outside of that company, allowing flexibility for expand in phases and in different regions when opportunities arise.

Financing and chips: the muscle behind Stargate

To support this deployment, OpenAI and its partners will use mixed financing which combines cash and debt, and they contemplate resorting to agreements to rent chip capacity dedicated to infrastructure. This approach allows hardware availability to be synchronized with the construction and commissioning stages.

In parallel, OpenAI and Nvidia have signed a letter of intent with a view to a strategic alliance that includes the implementation of at least 10 GW in chips for AI infrastructure, along with a possible investment of up to $100.000 billion by Nvidia at OpenAI. The first phase of hardware deployment is being considered for the second half of 2026, subject to the final formalization of the agreements.

Oracle strengthens its role with its cloud LEISURE and the arrival in Abilene of the first racks with NVIDIA GB200 GPU last June, while SoftBank provides center design and energy expertise to accelerate construction. OpenAI insists that the key to AI delivering real value lies in increasing available computing power without compromising security and cost efficiency.

Employment and impact on the local economy

The new facilities foresee the creation of around 25.000 jobs on site, as well as indirect jobs in construction, energy, and supply chain. Stargate's push will translate into increased demand for electrical infrastructure, hiring technical talent and agreements with local suppliers, especially in Texas, Ohio, and New Mexico.

Executives from the three companies have stressed that the challenge is of such a scale that it requires a industrial collaboration broad, with multiple technological and financial partners, to maintain the pace of adoption of AI and make it easier for companies and administrations to access this computing power.

The announcement places Stargate as one of the major AI infrastructure programs underway in the US, combining new centers, expansions and chip supply agreements to move towards a computer network more powerful, scalable and efficient.

OpenAI's Stargate
Related article:
OpenAI and Oracle's Stargate Megaproject: Massive AI Infrastructure in the United States

Follow us on Google News