OpenAI signed a $38 billion, seven-year cloud deal with AWS, a move that immediately expands the ChatGPT maker’s access to computing power and lifted Amazon shares more than 4% on November 3, 2025. The pact underscores how the AI race is increasingly about who can secure the most reliable, scalable infrastructure.
Announced on November 3, 2025, the multi‑year agreement allows OpenAI to run and scale its core AI workloads on Amazon Web Services right away. The companies said OpenAI will draw on AWS data centers equipped with hundreds of thousands of Nvidia GPUs, with all targeted capacity slated to come online by the end of 2026 and room to expand into 2027. The arrangement is framed as a strategic partnership rather than an exclusive commitment, adding a major new cloud provider alongside Microsoft.
The deal reflects the industry’s escalating demand for high‑end compute to train and serve increasingly capable models. It also follows governance changes at OpenAI that loosened prior constraints tied to its Microsoft relationship, opening the door to multi‑cloud diversification. For Amazon, landing OpenAI as a marquee customer helps counter perceptions that AWS was ceding AI momentum to rivals. More broadly, it signals that access to energy‑hungry infrastructure—chips, data centers, and specialized networking—has become a defining competitive edge in generative AI.
OpenAI has layered substantial long‑term infrastructure commitments across providers in recent months. While that has prompted questions about sustainability, the company says steep revenue growth supports the investment cadence. Amazon, meanwhile, continues to back other AI startups—including Anthropic—highlighting a strategy of building the underlying compute rails for the broader ecosystem.
Investors welcomed the announcement. Amazon shares rose more than 4% on November 3, adding to gains after its latest earnings. Analysts framed the pact as validation of AWS’s capacity to deliver at frontier scale and as a potential incremental contributor to high‑margin cloud revenue over time.
OpenAI begins using AWS immediately, with full intended capacity targeted by late 2026 and optional expansion in 2027. All capacity targeted by end of 2026 sets a clear execution clock for both companies. Watch for how workloads are split across clouds, the pace of new data‑center buildouts, and whether the partnership spurs further chip and power‑supply investments across the AI supply chain.