Challenging NVIDIA’s Dominance? AMD and Oracle Partner to Build the World’s Largest Non-NVIDIA AI Cluster

TradingKey - On Tuesday (local time), chip giant AMD (AMD.US) and cloud provider Oracle (ORCL.US) announced a major strategic partnership. Year-to-date, shares of AMD and Oracle have surged approximately 79% and 86%, respectively.

Under the agreement, AMD will begin deploying around 50,000 of its most advanced MI450 AI accelerator chips into Oracle’s data centers starting in Q3 2026, forming what will be the largest AI computing cluster not based on NVIDIA technology to date.

This marks the first large-scale deployment of the AMD MI450 in a public cloud environment. The chips will be integrated into AMD’s proprietary Helios server rack system and paired with AMD CPUs to deliver a full-stack AI computing solution — directly competing with NVIDIA’s next-generation “Vera Rubin” series.

While financial terms were not disclosed, the deal is widely seen as a bold challenge to NVIDIA’s dominance in the AI hardware space. As demand for generative AI surges globally, AI compute capacity has become critically scarce. Leading AI firms like OpenAI have repeatedly delayed model iterations due to insufficient processing power.

“Demand for AI is so strong that I’m turning away customers far more often than I’m saying yes,” admitted Mahesh Thiagarajan, Head of Cloud Infrastructure at Oracle.

The partnership builds on both companies’ recent momentum. AMD has signed a five-year agreement with OpenAI, expected to generate tens of billions of dollars in revenue. Oracle, meanwhile, has secured a long-term cloud services deal with OpenAI valued at approximately $300 billion.

Daniel Newman, CEO of Futurum Group, said:“Oracle has demonstrated its determination to bet on the AI era. Going forward, the company needs to prove whether, beyond computing power, it can leverage its vast data resources and enterprise software capabilities to create greater value amid the AI wave.”