OpenAI has secured a significant multi-year partnership with Amazon Web Services (AWS), valued at $38 billion. This deal provides OpenAI with immediate access to AWS’s advanced infrastructure, enabling the AI company to enhance and scale its AI workloads efficiently. The collaboration marks a pivotal shift in OpenAI’s cloud computing strategy, allowing it to leverage AWS’s powerful computing resources.
AWS will supply OpenAI with its EC2 UltraServers, which are equipped with hundreds of thousands of Nvidia GPUs and the capability to scale up to tens of millions of CPUs. This state-of-the-art infrastructure is designed for low-latency, high-performance AI processing, crucial for both training new models and serving applications like ChatGPT. The sophisticated clustering of Nvidia GB200s and GB300s GPUs on these servers will accommodate a range of workloads, adapting to OpenAI’s evolving requirements.
Sam Altman, co-founder and CEO of OpenAI, stated that scaling frontier AI necessitates substantial and reliable computing power. He emphasized that partnering with AWS enhances the overall compute ecosystem, paving the way for advanced AI solutions accessible to a wider audience.
In a statement on X, Andy Jassy, CEO of Amazon, announced that OpenAI would begin utilizing AWS’s infrastructure immediately. He projected that the full capacity of the infrastructure would be deployed by the end of next year, with provisions for further expansion in 2027 and beyond.
This strategic move comes after a previous relationship with Microsoft, which was OpenAI’s exclusive cloud-computing partner until January 2023. Microsoft initially invested $1 billion in OpenAI in 2019 and subsequently increased its stake to around $10 billion in 2023. Reports indicate that last month, OpenAI contracted Microsoft for additional Azure services valued at $250 billion, granting the tech giant a 27 percent stake worth approximately $135 billion.
As part of its recapitalization strategy, OpenAI will now have the opportunity to develop some products jointly with third parties. However, Microsoft’s intellectual property rights will not extend to OpenAI’s consumer hardware, allowing the AI company to diversify its offerings.
Prior to finalizing its current partnership, OpenAI also secured a deal with Oracle to rent substantial computing power from the tech firm for use in U.S. data centers. This arrangement is part of OpenAI’s Stargate joint venture project, showcasing the company’s commitment to maintaining robust computing capabilities across multiple platforms.
This landmark agreement with AWS positions OpenAI at the forefront of AI development, ensuring it has the necessary resources to drive innovation and growth in the rapidly evolving tech landscape. The collaboration not only enhances OpenAI’s operational flexibility but also signifies a new chapter in the competitive cloud computing market.
