In a landmark move reshaping the artificial intelligence landscape, OpenAI has entered a $38 billion, multi-year partnership with Amazon Web Services (AWS) to access massive computing power for its next generation of AI models. The agreement, announced today, grants OpenAI unprecedented scale and flexibility to run its core AI workloads on AWS’s global cloud infrastructure.
Expanding Beyond Microsoft
The partnership marks OpenAI’s first collaboration with a leading cloud provider outside of Microsoft, signaling a major shift in the company’s compute strategy. Under the deal, OpenAI will operate its ChatGPT, research models, and future agentic systems using AWS’s high-performance infrastructure powered by hundreds of thousands of Nvidia GPUs, with expansion potential into tens of millions of CPUs by 2027.
“This partnership strengthens the broad compute ecosystem that will power the next era of AI,” said Sam Altman, OpenAI’s co-founder and CEO. “Scaling frontier AI requires massive, reliable compute, and AWS offers the infrastructure to make that possible.”
AWS Delivers Scalable Frontier Infrastructure
The AWS deployment is among the largest AI infrastructure projects ever built, utilizing Amazon EC2 UltraServers that interconnect Nvidia GB200 and GB300 GPUs with ultra-low latency. These clusters are optimized for both training large-scale frontier models and serving inference for billions of ChatGPT users worldwide.
“AWS’s best-in-class infrastructure will serve as the backbone of OpenAI’s ambitions,” said Matt Garman, CEO of AWS. “The immediate availability of optimized compute at scale demonstrates why AWS is uniquely positioned to support AI leaders like OpenAI.”
The project’s capacity rollout begins immediately, with all initial infrastructure targeted for deployment by the end of 2026, followed by further scaling through 2027.
Massive Demand for Compute Power
The agreement underscores the surging global demand for AI computing resources, as companies race to build and deploy increasingly complex foundation models. AWS currently operates clusters exceeding 500,000 GPUs, giving OpenAI secure and reliable access to unmatched processing power while maintaining U.S. data security standards.
Broader AI Ecosystem Integration
The partnership extends an already deep collaboration between the two firms. Earlier this year, OpenAI’s open-weight foundation models became available through Amazon Bedrock, enabling thousands of enterprises, including Peloton, Comscore, Triomics, and Thomson Reuters, to integrate OpenAI’s models for agentic workflows, scientific analysis, and AI coding applications.
With this $38 billion deal, OpenAI cements AWS as a key partner in its next phase of growth, ensuring it can train larger, more capable AI systems while continuing to make ChatGPT and other tools accessible to millions of users worldwide.
Industry analysts describe the agreement as a “tectonic shift in AI infrastructure partnerships,” positioning AWS and OpenAI at the center of global AI innovation heading into 2026.