US$38B cloud computing deal explained

• The AWS OpenAI partnership commits $38 billion over seven years, providing immediate access to hundreds of thousands of Nvidia GPUs—one of the largest cloud computing deals in AI history
• OpenAI is already using the infrastructure, with full deployment targeted by the end of 2026, raising questions about AI infrastructure consolidation among major cloud providers

Every AI company faces a fundamental constraint: even the most brilliant algorithms are useless without massive computing power to run them. The partnership announced yesterday between AWS and OpenAI tackles this bottleneck head-on, with Amazon Web Services committing US$38 billion over seven years to provide the latter with immediate access to hundreds of thousands of Nvidia GPUs.

This AWS OpenAI partnership represents one of the largest cloud computing commitments in history, giving the maker of ChatGPT the infrastructure needed to develop increasingly sophisticated AI models.

For OpenAI, this infrastructure translates into tangible capabilities: the ability to serve millions of ChatGPT users simultaneously while training next-generation AI models in parallel. The computational resources include not just GPUs for intensive AI processing, but the capacity to scale to tens of millions of CPUs for what the industry calls “agentic workloads”—AI systems that can plan, reason, and execute complex tasks autonomously.

Unlike many partnership announcements that promise future collaboration, OpenAI began utilising AWS compute immediately upon announcement. This means the infrastructure is already processing real workloads, from answering user queries to potentially training more advanced versions of GPT models.

The technical infrastructure

The deployment features Amazon EC2 UltraServers equipped with state-of-the-art Nvidia GB200 and GB300 GPUs. These chips are clustered on the same network, enabling low-latency performance across interconnected systems—essential for training large AI models where thousands of processors must communicate constantly and efficiently.

According to the announcement, AWS will deploy infrastructure comprising hundreds of thousands of GPUs, with all capacity targeted for deployment before the end of 2026, and expansion potential extending into 2027 and beyond. The clusters are designed with flexibility to adapt to OpenAI’s evolving needs, supporting various workloads as AI technology advances.

“As OpenAI continues to push the boundaries of what’s possible, AWS’s best-in-class infrastructure will serve as a backbone for their AI ambitions,” said Matt Garman, CEO of AWS. “The breadth and immediate availability of optimised compute demonstrates why AWS is uniquely positioned to support OpenAI’s vast AI workloads.”

Garman’s confidence stems from AWS’s track record running large-scale AI infrastructure, with clusters exceeding 500,000 chips—an experience few cloud providers can match at this scale.

Why did this partnership happen now?

The timing reflects a critical inflexion point in AI development. As models grow more sophisticated, their computational requirements have increased exponentially. Training cutting-edge AI models now requires coordinating hundreds of thousands of processors working in concert for weeks or months—a logistical and technical challenge that demands both massive capital investment and operational expertise.

“Scaling frontier AI requires massive, reliable compute,” said OpenAI co-founder and CEO Sam Altman. “Our partnership with AWS strengthens the broad compute ecosystem that will power this next era and bring advanced AI to everyone.”

This isn’t the company’s first collaboration. Earlier in 2025, OpenAI’s open-weight foundation models became available on Amazon Bedrock, AWS’s managed AI service. OpenAI quickly became one of the most popular model providers on the platform, with thousands of customers—including Peloton, Thomson Reuters, and Verana Health—using their models for tasks ranging from coding assistance to scientific analysis.

Industry implications and critical questions

The AWS OpenAI partnership signals a broader trend: access to computing infrastructure has become as strategically important as the algorithms themselves. The US$38 billion commitment underscores the enormous capital requirements of modern AI development—a reality that concentrates AI capabilities among well-funded organisations with access to extensive computational resources.

For enterprise customers already using OpenAI models through Amazon Bedrock, this expanded partnership suggests more reliable service and potentially faster development of new capabilities. Companies using AI for agentic workflows, mathematical problem-solving, and scientific analysis stand to benefit from the enhanced infrastructure.

However, the deal also raises questions about consolidation in AI infrastructure. OpenAI’s substantial reliance on a single cloud provider creates dependencies, though the company likely maintains relationships with other infrastructure providers. The seven-year timeframe highlights the long-term planning required in AI development, but also locks both parties into a relationship that spans multiple technology generations.

Whether this concentration of AI development on major cloud platforms accelerates innovation or limits competition remains an open question. For businesses and researchers watching the AI space, this partnership makes clear that access to cutting-edge AI capabilities will increasingly depend on relationships with major cloud infrastructure providers—a dynamic that could reshape the competitive landscape of artificial intelligence for years to come.

Want to experience the full spectrum of enterprise technology innovation? Join TechEx in Amsterdam, California, and London. Covering AI, Big Data, Cyber Security, IoT, Digital Transformation, Intelligent Automation, Edge Computing, and Data Centres, TechEx brings together global leaders to share real-world use cases and in-depth insights. Click here for more information.