ChatGPT creators OpenAI have struck a $38B deal with Amazon Web Services (AWS), comitting to infrastructure expansion and unlocking access to “hundreds of thousands” of AI GPUs already powering AWS behind the scenes. The colloboration will allow OpenAI to scale even further, and the AI firm is to be given access to Amazon’s EC2 UltraServers for AI training and inference.

These AI GPUs are, unsurprisngly, courtesy of Nvidia, which recently partnered with OpenAI in $100 billion deal for the production of at least 10 gigawatts’ worth of data centers – that’s around the peak power required in New York City with its population of ~8.5 million people.

OpenAI’s seven-year deal with Amazon Web Services

“Under this new $38 billion agreement, which will have continued growth over the next seven years, OpenAI is accessing AWS compute comprising hundreds of thousands of state-of-the-art NVIDIA GPUs, with the ability to expand to tens of millions of CPUs to rapidly scale agentic workloads. AWS has unusual experience running large-scale AI infrastructure securely, reliably, and at scale–with clusters topping 500K chips. AWS’s leadership in cloud infrastructure combined with OpenAI’s pioneering advancements in generative AI will help millions of users continue to get value from ChatGPT.”

Source: Amazon

This is the same AWS which had a major outage recently, disrupting ‘half’ of the internet, given just how many services rely on Amazon’s servers. In any case, the “world-class infrastructure” of AWS is well equipped for advanced AI workloads, and OpenAI will be able to benefit from the collaboration, all the while giving the company space to scale even further. An ambitious figure of “tens of millions of CPUs” has already been thrown out there for generative AI workloads.

The targeted capacity is due to be deployed before the end of 2026, while the opportunity for expansion will open up in 2027 onwards. Amazon’s EC2 UltraServers will provide the infrastructure required for AI inference (the process of training AI models like GPT with unseen data), built on clusters of Nvidia’s GB200 and GB300 GPUs.

About the Author

Jack Goodall

At PC Guide, Jack is mostly responsible for reporting on hardware deals. He also specializes in monitors, TVs, and headsets and can be found putting his findings together in a review or best-of guide.