OpenAI and AWS Ink $38 Billion Compute Pact to Accelerate AI Deployment

Post by : Bianca Suleiman

OpenAI has finalised a $38 billion, multi-year agreement with Amazon Web Services (AWS) that gives the company broad access to AWS’s high-performance cloud infrastructure, intended to support substantially larger AI workloads.

The announcement follows OpenAI’s recent organisational restructuring that ended its non-profit status and opened new commercial pathways. The arrangement is also a strategic move to reduce dependence on any single cloud partner, while preserving access to top-tier computing resources.

Under the seven-year contract, OpenAI will deploy AWS EC2 UltraServers outfitted with thousands of Nvidia GB200 and GB300 GPUs. The environment is engineered to handle both inference for current ChatGPT models and the training demands of future systems, with the capacity to scale across millions of CPU cores. AWS has highlighted the configuration’s performance, efficiency and security, with full roll-out expected by the end of 2026 and potential extension into 2027.

OpenAI has also broadened its cloud footprint elsewhere: its foundational models were added to Amazon Bedrock earlier this year, and the organisation maintains separate computing arrangements with Microsoft Azure, Google Cloud and Oracle to meet surging demand.

The deal marks a clear operational shift, enabling OpenAI to speed research and commercial deployments without being tied to a single provider. It also underlines the intensifying contest among AI developers to secure vast, high-performance compute capacity essential for next-generation innovation.