OpenAI signs seven-year AWS cloud deal to power AI
OpenAI will spend $38 billion on AWS cloud capacity, gaining Nvidia chips to train and run its models, with workloads starting now.
The seven-year agreement allows OpenAI to purchase Amazon Web Services capacity, the Amazon.com (NASDAQ: AMZN) cloud unit, gaining access to hundreds of thousands of Nvidia (NASDAQ: NVDA) accelerators to train and run its AI models.
Announced on Nov 3, 2025, the pact has OpenAI starting on AWS immediately. Planned capacity is slated to come online by the end of 2026, with options to add more starting in 2027. The arrangement supplies large-scale compute to build and operate services such as ChatGPT.
Amazon intends to deploy Nvidia GB200 and GB300 accelerators in data clusters configured to power ChatGPT responses and train OpenAI’s next generation of models. The commitment arrives amid investor scrutiny of AWS’s AI competitiveness versus Microsoft and Google. Amazon shares hit an all-time high on Nov 3, 2025.
OpenAI has set long-term targets of 30 gigawatts of computing resources, and CEO Sam Altman has outlined ambitions to eventually add about 1 gigawatt of compute per week.
Altman described the need for "massive, reliable compute," adding that partnering with AWS "strengthens the broad compute ecosystem" and can "bring advanced AI to everyone." PP Foresight analyst Paolo Pescatore called it "a hugely significant deal" and "a strong endorsement of AWS compute capabilities to deliver the scale needed to support OpenAI."
As GNCrypto wrote previously, Amazon Web Services revenue grew 20% in Q3 2025, its strongest pace in three years, and Amazon signaled heavier data‑center investment to meet AI demand. The company is restructuring to streamline operations, including previously announced 14,000 corporate layoffs. Its market capitalization jumped by roughly $330 billion on Oct 31, 2025, as investors rewarded record cloud growth.
