Oracle Cloud to Deploy 50,000 AMD AI Chips, Challenging Nvidia’s Dominance

Oracle Cloud Infrastructure (ORCL) announced on Tuesday a significant expansion of its AI capabilities, slated to deploy 50,000 Advanced Micro Devices (AMD) graphics processors, commencing in the latter half of 2026. This move underscores the intensifying competition within the cloud AI infrastructure landscape.

Following the announcement, AMD shares saw a modest rise of approximately 2%. Conversely, Oracle’s stock experienced a decline of 4%, while Nvidia (NVDA), the current market leader in AI GPUs, dipped by over 3%. These market reactions reflect the nuanced investor sentiment regarding the evolving dynamics of AI hardware providers.

Oracle’s decision highlights a broader trend: cloud providers increasingly diversifying their GPU offerings with AMD’s solutions as a viable alternative to Nvidia’s dominance, particularly in AI inferencing workloads.

“We anticipate strong customer adoption of AMD, especially within the inferencing domain,” stated Karan Batta, Senior Vice President of Oracle Cloud Infrastructure, emphasizing the performance and cost-effectiveness that AMD’s GPUs can deliver for specific AI tasks.

The partnership will leverage AMD’s Instinct MI450 chips, unveiled earlier this year. These chips represent AMD’s foray into creating large-scale, interconnected AI compute systems. The MI450’s architecture allows for the assembly of rack-sized systems comprising 72 individual chips operating as a unified processing unit. This scalability is crucial for handling the computational demands of training and deploying sophisticated AI algorithms.

The announcement builds on a pre-existing relationship and public endorsement of AMD by OpenAI. OpenAI CEO Sam Altman appeared alongside AMD CEO Lisa Su in June at a unveiling event.

Earlier this month, OpenAI further solidified its commitment to AMD through an expanded infrastructure deal including multi-year support, with a 1-gigawatt rollout starting in 2026. This deal hints at a deeper collaboration, where, if the deployment proves fruitful, OpenAI could potentially acquire up to 160 million AMD shares, approximately 10% of the company’s equity.

In September, OpenAI inked a substantial five-year cloud services agreement with Oracle, estimated to be worth up to $300 billion. This deal signifies OpenAI’s significant investment in expanding its computing resources.

While OpenAI’s previous work, particularly the development of ChatGPT, has been closely associated with Nvidia’s hardware, the company’s current actions show diversifying their compute infrastructure with AMD provides further resilience.

OpenAI management has repeatedly emphasized the need to secure diverse sources of computing power. This strategy not only mitigates supply chain risks and vendor lock-in but also allows OpenAI to leverage custom fit AI chipset to match performance and power needs. The company also has plans to design its own AI chips with Broadcom.

“Both AMD and Nvidia have demonstrated exceptional capabilities, and we believe they each have unique strengths to offer,” emphasized Batta.

Oracle founder and chairman Larry Ellison is slated to present at Oracle AI World, where he will share insight on the latest OpenAI deal and how Oracle continues to strategically position itself against key cloud-computing competitors: Microsoft, Amazon, and Google.

“Oracle is making significant commitments in the AI space. The organization now needs to show customers that beyond increased compute capacity, it can take advantage of customer data and enterprise resources to add meaningful value in the AI market”, said Daniel Newman, CEO of The Futurum Group.