OpenAI announced a long-term partnership with Amazon Web Services (AWS), a seven-year deal worth $38 billion for cloud and AI infrastructure. The agreement gives OpenAI access to hundreds of thousands of Nvidia GPUs within AWS regions and availability zones, with deployments targeted to reach full scale by late 2026. Official post: AWS and OpenAI partnership.

The new capacity will power everything from ChatGPT’s live interactions and enterprise workloads to training and evaluation for upcoming foundation and reasoning models. The contract includes flexible scaling, letting OpenAI burst capacity during peak periods while aligning spend with utilization and performance milestones.

• Diversification: The move follows OpenAI’s recently updated arrangement with Microsoft that removed exclusivity constraints, clearing the way to purchase compute from multiple providers.
• Scope: AWS will provision large, multi-region GPU clusters tuned for AI training and inference, alongside storage, networking, security, and MLOps services.
• Ecosystem fit: The AWS pact joins OpenAI’s broader infrastructure buildout that includes partnerships and procurement plans spanning Oracle, Google, Nvidia, and Broadcom.
• Operational focus: Terms emphasize reliability, latency, and sustainability targets to keep ChatGPT responsive while supporting frontier model R&D.

Why it matters: With demand for AI compute exploding, OpenAI is locking in multi-vendor capacity to reduce risk and improve bargaining power. The $38B, seven-year AWS commitment signals a bet that diversified, hyperscale infrastructure will be essential to ship safer, faster, and more capable models—without single-cloud dependence. It also underlines how cloud providers are competing not only on GPUs but on end-to-end AI platforms that can keep pace with OpenAI’s aggressive product roadmap.