OpenAI Accelerates Compute Expansion with Stargate
Note OpenAI's Stargate expansion and compute capacity growth.
Monitor OpenAI's compute expansion for potential impact on AI model availability.
Summary
OpenAI’s Stargate initiative has surpassed its 10GW compute goal, adding more than 3GW of capacity in the last 90 days alone. The company is partnering with cloud providers, utilities, chipmakers, and local communities to build data centers that deliver the power needed for next‑generation AI models. Stargate’s Abilene, Texas site demonstrates rapid, responsible construction, using closed‑loop cooling that reduces water use to the equivalent of four average households once fully built.
The expanded infrastructure powered the training of GPT‑5.5 on NVIDIA GB200 systems running on Oracle Cloud Infrastructure, bringing the latest model closer to global users. OpenAI also launched a community engagement program, donating to the Port Washington‑Saukville Education Foundation to support workforce readiness. The company’s compute strategy is designed to keep pace with accelerating demand while preserving flexibility for future technology shifts.
Key changes
- Surpassed 10GW compute goal with 3GW added in last 90 days.
- Partnerships with cloud providers, utilities, chipmakers, and local communities.
- Abilene site uses closed‑loop cooling, reducing water use to four households.
- GPT‑5.5 trained on NVIDIA GB200 systems at Abilene.
- Community engagement program donated to Port Washington‑Saukville Education Foundation.
- Compute strategy designed to scale with demand while preserving flexibility.