Oracle to deploy cluster of more than 130,000 AMD MI355X GPUs

Oracle is significantly expanding its AI infrastructure offerings, highlighted by the deployment of an AMD MI355X GPU cluster scalable to 131,072 GPUs, promising over 2x better price-performance for AI workloads compared to previous AMD generations. The company has also made its Nvidia GB200 NVL72 supercluster generally available and is joining Nvidia's DGX Cloud Lepton platform, alongside a new customer partnership with Seekr, who will leverage OCI and AMD's MI300X GPUs for AI training; these moves underscore Oracle's commitment to providing a broad AI infrastructure and solidify its position as a key player in the AI cloud computing landscape.

Analysis

Oracle is substantially augmenting its AI cloud capabilities through significant infrastructure investments and strategic partnerships, positioning itself as a key provider for demanding AI workloads. The company announced plans to deploy a massive AMD Instinct MI355X GPU cluster, scalable up to 131,072 units, a considerable increase from the previously stated 30,000. This deployment on Oracle Cloud Infrastructure (OCI) is projected to offer over twice the price-performance for AI training and inference compared to AMD's prior GPU generation, with Oracle aiming to be among the first hyperscalers to offer these new 3nm-based chips featuring 288GB HBM3E memory and 79 TFLOPS of FP64 performance. Concurrently, Oracle has achieved general availability for its Nvidia GB200 NVL72 OCI Supercluster, also comprising 131,072 Blackwell GPUs, and plans to deploy 64,000 Nvidia GB200s in its Texas data center by late 2026. Further strengthening its Nvidia ties, Oracle is joining the Nvidia DGX Cloud Lepton platform. Underscoring its multi-vendor AI strategy and market traction, Oracle has secured a multi-year agreement with AI company Seekr, which will utilize OCI and AMD's MI300X GPUs for training next-generation large language and vision-language foundation models. These developments collectively signal Oracle's aggressive pursuit of the AI cloud market by offering diverse, high-performance GPU options from both leading providers, aiming to enhance its appeal for AI training, inference, and emerging agentic applications.

Key Decisions for Investors

• Oracle's aggressive dual-sourcing strategy for high-end GPUs from both AMD and Nvidia significantly de-risks its AI infrastructure expansion and enhances its competitive positioning in the cloud AI market; this warrants a positive outlook on ORCL's growth potential in this segment.
• The large-scale adoption of AMD's new MI355X GPUs by Oracle, coupled with AMD's claims of superior performance metrics over competitors, signals potential market share gains for AMD in the data center AI accelerator space, which could positively impact AMD's valuation.
• Investors should closely monitor the customer adoption rates and realized price-performance of Oracle's new AMD and Nvidia-based AI clusters, as these will be critical indicators of OCI's ability to effectively compete and capitalize on the escalating demand for AI compute.
• The Seekr partnership, utilizing AMD MI300X GPUs for specialized vision-language model training, exemplifies a tangible use-case and successful go-to-market strategy that could foreshadow further enterprise AI workload acquisitions by Oracle.