Amazon Web Services [AWS] and OpenAI have entered a multi-year, $38 billion strategic partnership. It will see OpenAI’s growing artificial intelligence workloads migrate to AWS’s cloud infrastructure. The deal, running over seven years, gives OpenAI access to hundreds of thousands of NVIDIA GPUs hosted on AWS, with capacity to scale to tens of millions of CPUs as demand for large-scale “agentic” AI systems intensifies.
The partnership marks a deepening convergence between infrastructure giants and AI model developers. This is a reflection of how computing power, not just algorithms, is defining the next phase of the AI race. For OpenAI, which operates ChatGPT and develops frontier models. The arrangement ensures sustained access to advanced, secure compute clusters amid a global shortage of high-end chips. For AWS, it signals confidence in its ability to deliver hyperscale AI environments that can support increasingly autonomous workloads.
Compute Power that will Fuels Next AI Race Phase
AWS’s infrastructure has long been built for scale and reliability, running clusters that exceed half a million chips. The new deployment designed for OpenAI will integrate NVIDIA’s latest GB200 and GB300 GPUs through Amazon EC2 UltraServers. It enables low-latency performance across interconnected systems critical for both real-time inference on ChatGPT and the training of future models. The first phase of capacity will be online by 2026, with expansion projected into 2027.
“Scaling frontier AI requires massive, reliable compute,” said Sam Altman, OpenAI’s CEO. “Our partnership with AWS strengthens the broad compute ecosystem that will power this next era and bring advanced AI to everyone.” AWS chief executive Matt Garman added that the collaboration demonstrates why AWS is “uniquely positioned to support OpenAI’s vast AI workloads.”
Beyond immediate infrastructure gains, the deal carries strategic subtext. OpenAI’s decision to diversify its compute providers hints at a pragmatic shift in an industry previously dominated by single-vendor dependencies. For AWS, it reinforces its standing amid intensifying competition from Microsoft Azure and Google Cloud both already embedded in the AI arms race.
The companies have previously worked together to distribute AI models through Amazon Bedrock. It offers OpenAI’s open-weight foundation models to millions of enterprise customers. That existing relationship, now expanded, suggests a deepening of mutual interests. AWS securing a flagship AI customer to showcase its next-generation infrastructure, and OpenAI ensuring resilience and agility as model complexity and compute demands continue to soar.
