Oracle Cloud Infrastructure announced plans to deploy 50,000 AMD Instinct MI450 graphics processors starting in the second half of 2026, marking one of the largest single expansions of AI computing power by a major cloud provider.
The move reflects how cloud companies are increasingly turning to AMD as an alternative to Nvidia’s dominant GPU lineup amid surging global demand for artificial intelligence infrastructure.
AMD shares rose about 3% in early trading following the announcement, while Oracle stock slipped slightly. The deployment will use AMD’s MI450 accelerators, introduced earlier this year, which represent the company’s first rack-scale AI system.
The design allows up to 72 chips to work in tandem as a single unit, providing the kind of high-bandwidth performance required for large-scale training and inferencing workloads.
Karan Batta, senior vice president of Oracle Cloud Infrastructure, said AMD’s chips will play a growing role in the company’s cloud portfolio. “We feel like customers are going to take up AMD very, very well – especially in the inferencing space,” he said during an interview. Batta added that AMD’s software stack is “critical” to enabling developers to train and deploy models efficiently on Oracle’s platform.
The partnership further strengthens the ties between Oracle, AMD, and OpenAI, which have deepened over the past several months. Earlier this year, OpenAI and AMD signed a multi-gigawatt compute deal to power future generations of large language models, potentially involving up to 160 million AMD shares if performance milestones are met.
The agreement complements OpenAI’s five-year, $300 billion cloud partnership with Oracle announced in September, positioning Oracle as a primary infrastructure provider for OpenAI’s rapidly growing workloads.
Oracle’s founder and chairman Larry Ellison is expected to outline further details of the company’s AI strategy at Oracle AI World 2025, emphasizing how Oracle plans to challenge Microsoft, Amazon, and Google in the race for AI cloud leadership. Analysts see the AMD partnership as part of a broader strategy to reduce reliance on Nvidia while leveraging Oracle’s deep enterprise data assets and software capabilities.
“The company must now prove that beyond capacity, it can capitalize on its massive underlying data and enterprise capabilities to add meaningful value to the AI wave,” said Daniel Newman, CEO of The Futurum Group.
OpenAI CEO Sam Altman has praised AMD’s progress, appearing alongside AMD CEO Lisa Su earlier this year to unveil the MI450 series. The collaboration highlights OpenAI’s multi-supplier approach, which also includes partnerships with Broadcom for custom AI chips and Nvidia for high-performance GPU clusters.
By 2026, Oracle’s deployment of AMD systems could become one of the largest dedicated AI infrastructures in the world – a cornerstone of the company’s effort to establish itself as a critical backbone for enterprise-scale AI computing.