OpenAI’s Stargate expansion fuels a new era in US AI infrastructure
Subscribe to our free newsletter today to keep up to date with the latest construction news.
OpenAI, Oracle, and SoftBank have announced five new US sites under their Stargate AI infrastructure initiative. The move pushes the consortium toward its $500 billion commitment and advances its goal of delivering 10 gigawatts of compute capacity by the end of 2025.
The sites include locations in Texas, New Mexico, Ohio and an undisclosed Midwest state. These are in addition to the operational flagship site in Abilene, Texas, and ongoing projects with CoreWeave. Together, they bring Stargate’s planned capacity to nearly 7 gigawatts and underscore a broader transformation in how AI infrastructure is scaled across the country.
Backed by a July agreement between OpenAI and Oracle to build up to 4.5 gigawatts in new capacity, the Stargate initiative now spans more than $400 billion in planned investment. SoftBank is leading development on two sites, including a modular project in Milam County, Texas, and a new data center design in Lordstown, Ohio, through its energy subsidiary SB Energy.
The developments are expected to create over 25,000 onsite jobs, with wider impact across national supply chains. The five new sites were selected from more than 300 proposals submitted by 30 states, highlighting growing regional interest in the AI economy.
New demands on construction
For the US construction sector, Stargate signals a shift in scope and urgency. AI data centers require power infrastructure, high-density cooling, and fiber networks at scales that differ from traditional builds. Spending on data center construction is expected to exceed $40 billion this year, with hyperscale projects leading that growth.
Many selected regions benefit from lower land and labor costs, but that introduces new challenges around speed, logistics and skilled workforce availability. Contractors are turning to modular construction and prefabrication methods to meet tight timelines. Efficient delivery is becoming just as critical as technical execution.
The materials demand alone is reshaping parts of the supply chain. Steel, concrete, thermal systems and transformers must be sourced and deployed at an accelerated pace. AI’s infrastructure footprint is growing, and developers are now optimizing designs for repeatability across multiple sites.
Powering the next generation of AI
Stargate’s approach reflects a shift in how AI companies manage compute needs. By building dedicated facilities on Oracle Cloud Infrastructure, OpenAI gains more control over cost, scale and system performance. It also reduces dependence on third-party cloud providers.
The initiative launched in January during an event at the White House. Policy alignment has helped fast-track permits and approvals, with local governments offering incentives to attract AI investment. As the buildout continues, energy access and regulatory efficiency are becoming key differentiators.
The scale of consumption is hard to ignore. US data centers account for over four percent of national electricity use. The environmental impact of AI workloads is already being scrutinized, especially as training models demand more power. SoftBank’s sites aim to integrate renewable sources, though energy reporting remains limited across the sector.
Infrastructure for the intelligence era
Stargate is more than a series of buildings. It represents a new model for infrastructure investment that connects digital growth with physical construction. The demands of large language models are now driving real-world spending, employment and regional development.
What makes this different from past technology cycles is the pace. AI training requirements are doubling every few months, and the supporting infrastructure must match that velocity. The companies behind Stargate are not just scaling servers. They are designing physical systems to power the next phase of global AI deployment.
In doing so, they are creating a new template for infrastructure, one where compute power is as vital as roads or electricity. For now, the buildout continues, bringing new jobs, shifting industry norms and redefining what construction means in the AI age.
Sources