OpenAI has signed a $38 billion deal with Amazon Web Services (NYSE: AMZN) to run its expanding artificial intelligence workloads, marking a notable shift away from its previous exclusive cloud partnership with Microsoft Corporation (NASDAQ: MSFT). This new multi-year agreement not only secures hundreds of thousands of Nvidia Corporation (NASDAQ: NVDA) graphics processing units for OpenAI’s use but also gives the company the flexibility to scale its infrastructure well into the coming years.
Since its founding, OpenAI developed a close cloud relationship with Microsoft, which provided its exclusive cloud services until earlier this year. However, OpenAI faced constraints with computing power from Microsoft, leading to a recent renegotiation that ended Microsoft’s preferential status and allowed OpenAI to pursue additional cloud partnerships. The Amazon deal follows agreements OpenAI has also made with Google, Oracle, and others, reflecting its aim to diversify its infrastructure providers.
The AWS contract guarantees immediate access to Amazon’s state-of-the-art computing resources, including EC2 UltraServers powered by Nvidia’s advanced GPUs. These clusters of GPUs are designed to deliver the performance and low latency necessary to support both training cutting-edge AI models and handling real-time inference tasks for applications like ChatGPT. This architectural design clusters Nvidia GPUs to optimize efficiency, enabling OpenAI to run workloads with high throughput across interconnected systems.
Amazon plans to have the full AWS infrastructure capacity slated for OpenAI’s use deployed before the end of 2026, with the flexibility to continue scaling into 2027 and beyond. This arrangement offers OpenAI significant operational freedom to grow its AI compute power rapidly and reliably on a large scale without being tied exclusively to any single provider.
AWS Vice President Dave Brown explained the agreement by emphasizing the distinct and dedicated computing capacity AWS is providing. Amazon CEO Matt Garman stated that AWS’s comprehensive and immediately available optimized infrastructure supports OpenAI’s ambitious AI workloads and furthers the company’s efforts to push AI boundaries.
While OpenAI continues to invest heavily in Microsoft’s Azure cloud services, with plans to purchase an additional $250 billion in Azure capacity, the new agreement signals a clear strategic broadening of its cloud footprint. For Amazon, this deal is a major win, especially as it competes with other cloud giants like Microsoft and Google for AI infrastructure supremacy. It also highlights AWS’s prowess in handling demanding long-term AI workloads and reinforces its standing in the fast-evolving AI landscape.
Financially, the $38 billion commitment over seven years is one of the largest cloud infrastructure agreements in recent memory, underscoring the scale of investment necessary to support frontier AI development. OpenAI’s use of Nvidia GPUs through AWS complements its direct agreements with Nvidia and other chip makers like AMD and Broadcom. Together, these partnerships enable OpenAI to expand both its AI model training and deployment capabilities.
This broad network of agreements reflects OpenAI’s focus on operational agility and preparing for future growth, including a likely public offering. By diversifying its compute sources across multiple top-tier cloud services and chip providers, OpenAI is positioning itself for long-term scalability and resilience in a highly competitive and capital-intensive space.
Supporting millions of users worldwide, OpenAI’s models are now accessible through Amazon Bedrock, AWS’s managed AI foundation service, serving a variety of clients in fields including coding, scientific research, and business automation. This collaboration with AWS aims to enhance accessibility to advanced AI capabilities through robust, scalable cloud infrastructure.
The signing of this landmark deal marks a new chapter for OpenAI’s infrastructure strategy, reflecting the growing importance of flexible, reliable computing power as artificial intelligence continues to evolve and expand across industries worldwide.
