Oracle has deployed and optimized its initial wave of liquid-cooled NVIDIA GB200 NVL72 GPU racks within its data centers, positioning thousands of NVIDIA Blackwell GPUs for immediate availability. The advanced hardware is integrated across NVIDIA DGX Cloud and Oracle Cloud Infrastructure (OCI), enabling enterprises to develop and operate the latest reasoning models and Artificial Intelligence agents. Core to this installation are high-speed NVIDIA Quantum-2 InfiniBand and Spectrum-X Ethernet networking solutions, which deliver scalable, low-latency interconnects essential for intensive computational workloads. In addition, a comprehensive suite of software and database integrations from both NVIDIA and OCI ensures streamlined deployment and performance optimization for customers using these new GPUs.
OCI has established itself as one of the world´s leading and rapidly expanding cloud service providers. By being among the first to implement NVIDIA GB200 NVL72 systems, Oracle demonstrates a significant commitment to accelerated computing. The company plans to construct one of the largest Blackwell-based clusters globally, with its OCI Superclusters designed to scale beyond 100,000 NVIDIA Blackwell GPUs. This colossal scale aims to satisfy the rapidly increasing demand for inference tokens and high-performance computing resources, driven by the global surge in Artificial Intelligence research and application deployment.
The move comes as the pace of innovation in Artificial Intelligence continues to accelerate, with firms like OpenAI recently launching more advanced reasoning models. Oracle´s collaboration with NVIDIA and the swift build-out of GPU infrastructure mark a notable step in the industry-wide race to support complex Artificial Intelligence workflows, large language models, and emerging agent-based technologies. The company´s integrated approach—including high-speed networking and tailored software stacks—positions OCI as a highly competitive platform for organizations aiming to stay at the forefront of Artificial Intelligence development and deployment at scale.
