• OpenAI to spend $38 billion on AWS compute capacity to reduce reliance on one cloud provider.
• Deal adds Nvidia GPUs and supports inference and training.

OpenAI has agreed to buy $38 billion-worth of capacity from AWS, marking its first large cloud deal with Amazon and showing that the startup is no longer tied to a single provider.

Under the agreement, OpenAI will start running workloads on AWS right away. The company will access hundreds of thousands of Nvidia GPUs located in the US, with more capacity expected over time.

Amazon’s stock closed up 4% in value on Monday, hitting a record high. Shares have climbed 14% in two days, the strongest run since late 2022. At first, AWS will lean on existing data centres to support OpenAI, and Amazon plans to add new infrastructure later.

“It’s completely separate capacity that we’re putting down,” said Dave Brown, vice president of compute and machine learning services at AWS. “Some of that capacity is already available, and OpenAI is making use of that.”

OpenAI has been lining up huge infrastructure deals in recent months, totalling about $1.4 trillion with companies like Nvidia, Broadcom, Oracle, and Google. The pace has fueled concerns about an AI bubble and whether the US has enough power, resources, and hardware to deliver on these plans.

More cloud partners in the mix

Until recently, OpenAI relied on Microsoft under an exclusive cloud contract. Microsoft first backed OpenAI in 2019 and has invested around $13 billion in the company. In January, Microsoft said it would shift to a model that gives it first refusal on new requests, rather than full exclusivity.

Last week, Microsoft’s special status ended under new commercial terms. That change has opened the door to more deals with other major cloud providers. OpenAI had already reached agreements with Oracle and Google, but AWS remains the biggest player in cloud services.

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

OpenAI still plans to spend heavily on Microsoft’s cloud. The company said last week that it will buy an additional $250 billion in Azure services.

The announcement also matters for Amazon, which has close ties to Anthropic, one of OpenAI’s key rivals. Amazon has invested billions in Anthropic and is building an $11 billion data centre campus in Indiana for Anthropic workloads.

“The breadth and immediate availability of optimised compute demonstrates why AWS is uniquely positioned to support OpenAI’s vast AI workloads,” AWS CEO Matt Garman said in the release.

Amazon reported over 20% annual revenue growth in AWS last week, beating analyst expectations. Microsoft and Google grew even faster, posting cloud growth of 40% and 34%.

Hardware choices shape future growth

The deal centres on Nvidia chips, including two popular Blackwell models. There may be room to add other hardware later, and Anthropic is already using Amazon’s custom Trainium chip in the new facility.

“We like Trainium because we’re able to give customers something that gives them better price performance and honestly gives them choice,” Brown said. He declined to share any details about Trainium’s work with OpenAI.

The infrastructure will support both inference – like ChatGPT responding to users – and training new models. OpenAI can continue to scale with AWS over the next seven years, but plans past 2026 are still uncertain.

OpenAI’s foundation models, including open-weight versions, are already available on Bedrock, AWS’s managed AI service. Companies like Peloton, Thomson Reuters, Comscore, and Triomics use OpenAI models on AWS today for tasks like coding, math, research, and agent-based workflows.

Monday’s announcement strengthens the business relationship. “As part of this deal, OpenAI is a customer of AWS,” Brown said. “They’ve committed to buying compute capacity from us, and we’re charging OpenAI for that capacity. It’s very, very straightforward.”

For OpenAI, this agreement is another step toward preparing for a future public offering. By working with multiple cloud vendors and securing long-term access to hardware, the company is trying to show that it can support growth at scale. Altman recently said in a livestream that an IPO is “the most likely path” as the company raises more capital. CFO Sarah Friar also suggested that recent changes inside the company are meant to help prepare for going public.

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.

Tech Wire Asia is powered by TechForge Media. Explore other upcoming enterprise technology events and webinars here.