Building the compute infrastructure for the Intelligence Age
Key Points
- Surpassed 10GW capacity goal with 3GW added in 90 days
- Ecosystem-wide partnerships across utilities, chipmakers, and cloud providers
- GPT-5.5 trained at Abilene flagship site with responsible resource management
Summary
OpenAI's Stargate initiative aims to build the compute foundation required to deliver AGI benefits globally. The project has already surpassed its initial 10GW capacity goal for the United States by 2029, with over 3GW added in the last 90 days alone.
Key Points
- Accelerating capacity: Stargate has exceeded the original 10GW milestone ahead of schedule, demonstrating rapid infrastructure expansion to meet surging AI demand
- Ecosystem partnership: Success requires coordination across utilities, chipmakers, cloud providers, construction firms, and public sector partners—no single company can build at this scale alone
- Community integration: Projects include local investments in education, workforce development, and responsible resource stewardship (e.g., closed-loop cooling systems)
- Abilene flagship site: Demonstrates responsible infrastructure development, operating on Oracle Cloud Infrastructure with NVIDIA GB200 systems and trained GPT-5.5
- Strategic expansion: OpenAI and partners are evaluating additional data center locations nationwide to meet growing demand while preserving technological flexibility
- Workforce development: Partnerships with North America's Building Trades Unions create job pathways for skilled workers and apprentices in infrastructure construction
- Efficiency focus: Improved compute efficiency enables lower costs, faster model serving, and broader access to AI capabilities across businesses and consumers