Oracle Bets Big on AMD: AI Training Revolution with MI355X GPUs

In a significant move to bolster its AI capabilities, Oracle has announced a collaboration with AMD to integrate the latest AMD Instinct MI355X GPUs into its Oracle Cloud Infrastructure (OCI). This strategic partnership involves deploying up to 131,072 of these GPUs, starting with a confirmed initial deployment of 30,000 units, to create a zettascale AI cluster designed for large-scale AI training and inference workloads.

The AMD Instinct MI355X GPUs, built on AMD’s CDNA 4 architecture and utilizing TSMC’s 3nm process, promise substantial improvements over previous generations. They offer nearly triple the compute power and a 50% increase in high-bandwidth memory, making them ideal for demanding AI applications.

Oracle’s decision to leverage AMD’s technology is expected to significantly enhance the price-performance ratio for AI workloads, potentially reducing costs and accelerating the development of large-scale AI models. This move also positions Oracle as a major player in the AI infrastructure market, offering customers more flexible and cost-effective alternatives to Nvidia-based solutions.

Industry analysts and commentators view this partnership as a strategic challenge to Nvidia’s dominance, highlighting the potential for broader multi-cloud adoption and lower AI compute prices. The collaboration is seen as particularly beneficial for enterprises seeking alternatives to Nvidia and for startups looking to access high-performance AI resources without the hefty price tag.

Sources

• Oracle News: Oracle and AMD Collaboration
• Data Center Dynamics: Oracle’s AMD Deployment