AWS & OpenAI Strike $38 Billion Partnership to Power Next-Gen AI Infrastructure

The AI industry is booming, and so are the partnerships around the top AI giants in the industry. After a recent news about Microsoft investing $15.2 billion in the UAE for expanding AI and cloud solution, we have another development around AI.

Today, OpenAI and Amazon Web Services (AWS) have announced a multi-year, $38 billion strategic partnership to power OpenAI’s next-generation AI workloads. The agreement allows OpenAI immediate access to AWS’s world-class infrastructure, including clusters featuring hundreds of thousands of NVIDIA GPUs.

This partnership will also help to scale up to tens of millions of CPUs in the coming years. No wonder, that’s a staggering amount of compute, and a confirmation that the generative AI arms race has intensified like never before.

Under the agreement, AWS will provide Amazon EC2 UltraServers equipped with NVIDIA’s GB200 and GB300 chips. Worth noting that these will be optimized for high-efficiency AI workloads. Not to mention, they will support everything from ChatGPT inference tasks to training frontier models, giving OpenAI the flexibility to handle rapidly growing user demand.

All capacity is expected to be deployed by the end of 2026, with expansion plans stretching into 2027 and beyond. OpenAI CEO Sam Altman called the partnership essential for scaling frontier AI, noting that “massive, reliable compute” is what enables the future in intelligence.

AWS CEO Matt Garman comments, similarly, adding that the company’s best-in-class infrastructure is “uniquely positioned” to support OpenAI’s goals. The announcement also builds on previous collaborations between the two companies.

Earlier this year, OpenAI’s foundation models became available through Amazon Bedrock, quickly becoming some of the platform’s most-used models by enterprises.