Key Takeaway:
OpenAI, Oracle, and SoftBank are expanding the $500 billion Stargate initiative with five new U.S. Stargate AI data centers, boosting capacity to nearly 7 GW (comparable to seven nuclear reactors) and accelerating the timeline to achieve 10 GW by the end of 2025, driven by current unmet demand (ChatGPT serves ~700 million weekly users) and next-gen model training needs
Stargate AI Data Centers Ahead of Schedule – Key Points
Scale of Expansion:
The new Stargate AI data centers bring the platform to nearly 7 GW planned capacity and more than $400 billion investment over the next three years. This expansion places the program ahead of schedule to meet its 10 GW / $500 billion goal by 2025. HSBC estimates global AI infrastructure buildouts already exceed $2 trillion, positioning Stargate as the largest single initiative.
Oracle Partnership:
OpenAI and Oracle signed a $300 billion+, 4.5 GW agreement in July 2025. Oracle’s projects include Stargate AI data centers in Shackelford County (TX), Doña Ana County (NM), and an upcoming Midwest site, plus a 600 MW expansion near Abilene. Oracle reports ~6,000 daily construction workers and ~1,700 permanent jobs across its Stargate projects. Overall, Oracle-led sites are projected to create 25,000 onsite jobs and tens of thousands more indirectly.
SoftBank Partnership:
Two new facilities will add up to 1.5 GW of scalable capacity over the next 18 months:
Lordstown, Ohio – advanced data center design, operational in 2026.
Milam County, Texas – rapid-build site powered by SB Energy (SoftBank Group).
These aim to deliver fast deployment, energy integration, and cost efficiency, forming a template for future large-scale rollouts.
Demand and Capacity Rationale:
ChatGPT serves ~700 million weekly active users, creating ongoing strain on existing compute. OpenAI cites a “massive compute crunch”, with shortages affecting both inference and training of next-gen models. The multi-gigawatt expansion of Stargate AI data centers aims to close this gap and sustain future growth.
Financing Structure:
OpenAI CFO Sarah Friar disclosed $13 billion expected revenue in 2025, with expansion financed through operating cash flow and debt. Nvidia is providing equity tied to systems deployment and will be paid for delivered GPUs. Analysts highlight circular funding risks (e.g., infrastructure providers invest in OpenAI, which then contracts them at scale). Reporting also notes Nvidia exploring GPU leasing models via a separate entity (another layer of financial engineering) while Oracle’s reported ~$30B/year build-operate arrangement similarly centralizes spend with a key supplier.
Abilene Flagship Campus:
The Abilene, Texas campus (~180 miles west of Dallas) is the first Stargate site online. One building is live and another near completion, with the campus able to scale past 1 GW (enough power for ~750,000 U.S. homes). The site runs on Oracle Cloud Infrastructure (OCI); NVIDIA GB200 racks started arriving June 2025. OpenAI has begun training and inference there and plans to incorporate NVIDIA’s next-gen Vera Rubin chips as capacity comes online in 2026.
Site Selection Process:
Launched January 2025, the nationwide RFP reviewed 300+ proposals from 30+ states. The five new selections are the first set, with additional U.S. sites planned as Stargate meets and surpasses the initial commitment.
Industry and Political Weight:
Stargate was first announced at the White House in January 2025 with President Trump and has been framed as an economic and national-security priority. OpenAI’s infrastructure paper argues the buildout could reshape the U.S. power grid via new energy technologies and bolster U.S. influence over global AI infrastructure.
Why This Matters:
Stargate is the largest AI infrastructure project in history, with $500 billion committed to 10 GW of compute power. By anchoring U.S. leadership in advanced compute, these Stargate AI data centers secure GPU supply chains, create tens of thousands of jobs, and enable training of the next generation of AI models. The project also highlights the rise of supplier-linked financing, with Nvidia and Oracle both investors and beneficiaries, signaling a new financial model for building “AI factories” at planetary scale.
This article was drafted with the assistance of generative AI. All facts and details were reviewed and confirmed by an editor prior to publication
Explore the U.S. government’s strategic push in semiconductor manufacturing, featuring TSMC’s and Samsung’s major investments and technological leaps.
All you need to know about the critical components of AI infrastructure, hardware, software, and networking, that are essential for supporting AI workloads.
Oracle and OpenAI ink a record $300B cloud contract, reshaping enterprise infrastructure and raising new questions about AI compute and energy.
Alibaba plans to invest $53B in AI and cloud infrastructure over three years, aiming for AGI leadership and global cloud dominance, surpassing past decade’s spending.
Read a comprehensive monthly roundup of the latest AI news!
