NVIDIA Spectrum-X Ethernet Switches Speed Up Networks for Meta and Oracle

NVIDIA has announced that Meta and Oracle will enhance their AI data center networks using NVIDIA Spectrum-X™ Ethernet networking switches.

Both companies are deploying Spectrum-X Ethernet switches as part of an open, accelerated networking architecture designed to scale efficiently, boost AI training performance, and shorten time to insights.

“Trillion-parameter models are transforming data centers into giga-scale AI factories, and industry leaders like Meta and Oracle are standardizing 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, connecting millions of GPUs into a single, unified computer capable of training the largest models ever built.”

Oracle plans to build giga-scale AI factories powered by the NVIDIA Vera Rubin architecture and interconnected through Spectrum-X Ethernet.

“Oracle Cloud Infrastructure is purpose-built for AI workloads, and our collaboration with NVIDIA strengthens that leadership,” said Mahesh Thiagarajan, executive vice president of Oracle Cloud Infrastructure. “By adopting Spectrum-X Ethernet, we can connect millions of GPUs with breakthrough efficiency, allowing customers to train, deploy, and benefit from the next generation of generative and reasoning AI more quickly.”

Meta will integrate Spectrum Ethernet switches into its networking infrastructure for the Facebook Open Switching System (FBOSS), the company’s large-scale network management software platform. This integration will accelerate deployment at scale, improving AI training efficiency and reducing time to insights.

“Meta’s next-generation AI infrastructure demands open and efficient networking at an unprecedented scale,” 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 achieving the efficiency and consistency needed to train larger models and bring generative AI applications to billions of people.”

The NVIDIA Spectrum-X Ethernet platform — comprising Spectrum-X Ethernet switches and Spectrum-X Ethernet SuperNICs — is purpose-built for the trillion-parameter model era. It enables hyperscalers to interconnect millions of GPUs with unmatched efficiency and scalability, meeting the growing demands of AI at scale.

Spectrum-X Ethernet has already demonstrated record-setting performance, achieving 95% data throughput using NVIDIA’s advanced congestion-control technology. In comparison, traditional Ethernet solutions typically achieve around 60% due to flow collisions. This leap in efficiency represents a breakthrough in both the performance and economics of AI-scale networking.

NVIDIA’s Spectrum-XGS Ethernet technology, part of the Spectrum-X networking platform, allows data centers to interconnect across cities, countries, and continents — creating vast, giga-scale AI super-factories.

Built on NVIDIA’s full-stack platform — including GPUs, CPUs, NVIDIA NVLink™, and software — Spectrum-X delivers seamless performance from compute to network, with adaptive routing, congestion control, and AI-driven telemetry ensuring efficiency and reliability across massive AI training and inference clusters.