Meta and Oracle are adopting NVIDIA’s Spectrum-X Ethernet switches to enhance the performance of their AI data centre networks, aiming to accelerate deployment at scale, improve AI training efficiency, and shorten the time needed to generate insights.
Both companies are leveraging Spectrum-X as an open and accelerated networking architecture that can support the growing scale of AI workloads. As trillion-parameter models transform data centres into massive AI computing facilities, Spectrum-X enables hyperscalers to connect millions of GPUs into unified systems capable of training the largest AI models.
“Trillion-parameter models are transforming data centres into giga-scale AI factories, and industry leaders like Meta and Oracle are standardising on Spectrum-X Ethernet to drive this industrial revolution,” said Jensen Huang, founder and CEO of NVIDIA. “Spectrum-X is not just faster Ethernet — it’s the nervous system of the AI factory, enabling hyperscalers to connect millions of GPUs into a single giant computer to train the largest models ever built.”
Oracle builds giga-scale AI factories
Oracle plans to use Spectrum-X Ethernet switches to build giga-scale AI supercomputers powered by NVIDIA’s Vera Rubin architecture. These facilities will interconnect millions of GPUs with high efficiency, enabling customers to train, deploy, and benefit from the next wave of generative and reasoning AI technologies more quickly.
“Oracle Cloud Infrastructure is designed from the ground up for AI workloads, and our partnership with NVIDIA extends that AI leadership,” said Mahesh Thiagarajan, executive vice president of Oracle Cloud Infrastructure. “By adopting Spectrum-X Ethernet, we can interconnect millions of GPUs with breakthrough efficiency so our customers can more quickly train, deploy and benefit from the next wave of generative and reasoning AI.”
This move reflects Oracle’s commitment to building infrastructure optimised for the scale and complexity of future AI applications, which require seamless GPU connectivity and rapid data movement.
Meta enhances AI infrastructure with Spectrum-X
Meta will integrate Spectrum Ethernet switches into its Facebook Open Switching System (FBOSS), a software platform developed to manage and control network switches at massive scale. This integration, including within Meta’s Minipack3N switch, will enable the company to expand its open networking approach while unlocking new levels of efficiency and predictability in AI training.
“Meta’s next-generation AI infrastructure requires open and efficient networking at a scale the industry has never seen before,” said Gaya Nagarajan, vice president of networking engineering at Meta. “By integrating NVIDIA Spectrum Ethernet into the Minipack3N switch and FBOSS, we can extend our open networking approach while unlocking the efficiency and predictability needed to train ever-larger models and bring generative AI applications to billions of people.”
Meta’s adoption of Spectrum-X is part of its broader strategy to support increasingly complex AI workloads and deliver new applications to a global user base.
Spectrum-X sets new benchmarks for AI networking
The NVIDIA Spectrum-X Ethernet platform, which includes Spectrum-X Ethernet switches and Spectrum-X Ethernet SuperNICs, is the first Ethernet platform purpose-built for AI. It delivers the scalability and performance required to interconnect millions of GPUs, enabling the construction of the world’s most advanced AI infrastructure.
Trillion-parameter models and generative AI are pushing the limits of data centre capabilities. Spectrum-X addresses these challenges by delivering unprecedented levels of efficiency. It has already enabled the world’s largest AI supercomputer to achieve 95% data throughput through advanced congestion-control technology. In contrast, standard off-the-shelf Ethernet struggles with throughput capped at around 60% due to thousands of flow collisions.
This performance leap significantly improves the economics and scalability of AI networking. The Spectrum-XGS Ethernet technology, part of the Spectrum-X platform, supports scale-across capabilities that link data centres across cities, countries, and continents, enabling the creation of vast, interconnected AI “super-factories”.
Spectrum-X builds on NVIDIA’s broader full-stack ecosystem — including GPUs, CPUs, NVLink, and software — to deliver seamless performance from compute to network. Its adaptive routing, advanced congestion control, and AI-driven telemetry ensure predictable and efficient performance for large-scale AI training and inference workloads.
