Following last week's confirmation of cooperation with OpenAI$380 billion cloud deal reachedSubsequently, Amazon Web Services (AWS) and OpenAI further announced details of this strategic partnership, under which AWS will provide OpenAI with its world-class infrastructure to "start running its advanced AI workloads immediately".
This $380 billion agreement will continue to grow over the next seven years. This means that after ending its exclusive cloud agreement with Microsoft, OpenAI has officially made AWS one of its core AI infrastructure providers.
Importing EC2 UltraServers, equipped with hundreds of thousands of GB200/GB300 GPUs
According to details released by AWS, the core technical aspects of this collaboration include:
Hardware specifications:
• AWS will provide OpenAI with Amazon EC2 UltraServers compute clusters, featuring hundreds of thousands of state-of-the-art NVIDIA GPUs (including GB200s and GB300s).
• AWS highlighted its experience in operating large-scale AI infrastructure (comprising over 50 chips) and stated that this deployment could be scaled up to tens of millions of CPUs in the future to meet the massive scale requirements of agentic AI workloads.
Deployment schedule:
• OpenAI will begin using AWS computing resources immediately.
• All computing capacity in the agreement is targeted to be deployed by the end of 2026, with the capacity to be further expanded after 2027.
Architecture Design:
AWS points out that the infrastructure it built for OpenAI uses a sophisticated architecture design that interconnects NVIDIA GPU computing clusters on the same network through EC2 UltraServers to achieve low-latency performance, enabling it to efficiently run various workloads, including ChatGPT inference and next-generation model training.
Strengthening the computing ecosystem, AWS holds a unique position.
Sam Altman, co-founder and CEO of OpenAI, stated, "Scaling up cutting-edge AI requires massive and reliable computing. Our partnership with AWS strengthens the broader computing ecosystem, which will power the next era and bring advanced AI to everyone."
AWS CEO Matt Garman responded, "As OpenAI continues to drive possibilities, AWS's world-class infrastructure will be a pillar of its AI ambitions. The breadth and immediacy of optimized computing demonstrate why AWS is uniquely positioned to support OpenAI's massive AI workloads."
AWS consolidates its strategic position in AI.
This collaboration is clearly a major victory for AWS in the AI infrastructure race. Previously, there had been doubts about AWS lagging behind competitors Microsoft (a key partner of OpenAI) and Google (one of the key partners of Anthropic) in the AI wave.
However, with OpenAI seeking to diversify its computing power and Anthropic recently announcing the adoption of millions of Google TPUs, AWS's successful acquisition of this $380 billion order from OpenAI not only alleviates market concerns about its over-concentration of AI customers (such as Anthropic), but also confirms that it remains a key player in the market for providing large-scale NVIDIA GPU computing power.
