Meta and Oracle are to adopt Nvidia Spectrum-X Ethernet networking switches for their AI data center efforts.

The deployments will see cloud giant Oracle build its giga-scale AI factories using Nvidia Vera Rubin architecture, which will be interconnected by Nvidia’s Ethernet networking platform, Spectrum-X.

Meta, meanwhile, plans to integrate Nvidia Spectrum Ethernet into its networking framework for the Facebook Open Switching System (FBOSS), a software stack run on Meta’s proprietary Minipack switch. It handles the massive scaling and performance needs of the firm’s data center network infrastructure. The social media giant's new integration is hoped to accelerate large-scale deployment, leading to greater efficiency in AI training and faster insights.

Nvidia CEO Jensen Huang cited the customer deals as industry validation for its Ethernet offering and a further sign of the times regarding AI data center evolution.

“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 Huang. “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 Cloud Infrastructure is designed from the ground up for AI workloads, and our partnership with Nvidia extends that AI leadership,” added Mahesh Thiagarajan, EVP for software development at Oracle Cloud Infrastructure (OCI). “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.”

“Meta’s next-generation AI infrastructure requires open and efficient networking at a scale the industry has never seen before. 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,” said Gaya Nagarajan, Meta's VP for networking engineering.

Explaining Spectrum-X Ethernet

As recently explored by SDxCentral, Spectrum-X is Nvidia's Ethernet-based networking fabric specifically designed for AI.

First revealed last year, Spectrum-X merges Nvidia’s networking hardware and software into an AI-tuned platform, pairing Spectrum SN5600 switches – built on the Spectrum-4 architecture – with BlueField-3 DPUs to boost generative-AI network performance to 1.6 times that of traditional Ethernet fabrics.

Nvidia recently unveiled a new scale-out technology, Spectrum-XGS Ethernet, designed to leverage Spectrum-X to connect multiple data centers to create a unified 'AI super-factory.' The networking algorithm solution integrates Nvidia Spectrum-X switches with ConnectX-8 SuperNICs, the company's AI-centric 800 Gb/s networking adapters, to offer adaptive distance congestion control, precision latency management, and end-to-end telemetry.

Oracle and Meta’s adoption of Spectrum-X furthers the chip giant’s relationship with both brands. Formalized in 2022, Nvidia’s partnership with Oracle has seen it launch AI services on OCI, specifically Nvidia AI Enterprise and Nvidia DGX Cloud.

The pair also partnered last year to deliver sovereign AI solutions to customers worldwide, while collaborating this year on agentic AI solutions via OCI and Nvidia AI Enterprise.

Meta’s infrastructure, meanwhile, includes 350,000 NVIDIA H100 GPUs as of 2025, with Nvidia’s chips powering its AI Research SuperCluster (RSC).

According to reports from last month, Meta is in talks to sign a multi-year cloud deal with Oracle worth about $20 billion in a bid to augment its growing AI training and inference needs.