OpenAI and Oracle have announced a collaboration to build an additional 4.5 GW of Stargate AI data‑centre capacity across the US, pushing the total under development to more than 5 GW.
The expanded network will house over two million AI training and inference chips, marking a significant advance toward OpenAI’s commitment, made at the White House in January, to invest $500 billion in 10 GW of US AI infrastructure over the next four years.
Construction of the initial Stargate I facility in Abilene, Texas, is already underway, with Oracle delivering the first Nvidia GB200 GPU racks last month and early workloads running at the site.
According to OpenAI CEO Sam Altman, the partnership with Oracle will accelerate America’s reindustrialisation, enhance US leadership in AI, and generate more than 100 000 jobs across construction and operations. Those roles span from specialised electricians and equipment operators to full‑time data centre technicians and ancillary positions in manufacturing and local services.
“Stargate is our long‑term vision to democratise AI benefits,” Altman said. “With Oracle and SoftBank as critical partners, we’re rapidly scaling the compute backbone necessary to power frontier research and bring advanced AI tools to everyone.” OpenAI’s collaboration with SoftBank is also progressing, focusing on site assessments and innovative data‑centre designs tailored for energy efficiency and reliability.
The joint effort represents one pillar of OpenAI’s broader Stargate platform, which includes partnerships with CoreWeave and continues to leverage Microsoft’s cloud services. It also complements OpenAI’s “OpenAI for Countries” initiative, which invests in international AI infrastructure built to US standards.
The White House has championed the Stargate rollout as essential to maintaining America’s economic competitiveness. Administration officials have noted that robust onshore AI compute capacity will help safeguard sensitive workloads and reduce reliance on foreign hardware supply chains.
Oracle’s contribution, ranging from GPU sourcing to data‑centre design and management, underscores the company’s expanding role beyond traditional enterprise systems into generative‑AI infrastructure. For Oracle, the deal follows its 2024 launch of a dedicated AI cloud region, positioning it as a strategic enabler for major AI developers.
As OpenAI and Oracle break ground on the expanded Stargate network, analysts predict the scheme will reshape the domestic data‑centre landscape, drive local hiring booms and set new benchmarks for sustainable, large‑scale AI operations. With the first phase already operational in Texas and the next 4.5 GW slated for development nationwide, Stargate is poised to underpin the next generation of AI applications across industry and government.
