Why Will Meta and Oracle Adopt Nvidia Spectrum-X Ethernet?

Nvidia has announced that Meta and Oracle are deploying its Spectrum-X Ethernet networking switches to accelerate giga-scale AI data centre performance. 

The adoption of Spectrum-X marks a major expansion of Nvidia’s networking ecosystem among hyperscale operators, enabling higher efficiency, faster model training and improved scalability across massive compute environments.

Enabling giga-scale AI performance

Meta and Oracle are integrating Nvidia Spectrum-X Ethernet switches into their respective infrastructures as part of a push to develop next-generation AI facilities capable of connecting millions of graphics processing units (GPUs). 

These systems are designed to support trillion-parameter models – large-scale ML systems that demand unprecedented data throughput and stability.

“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,” says 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.”

The announcement highlights a growing focus among hyperscalers on purpose-built networking architectures optimised for AI workloads.

Spectrum-X delivers an open and accelerated Ethernet solution that reduces deployment time, increases efficiency and shortens time to insights for AI developers and enterprises.

Oracle builds giga-scale AI supercomputers

Oracle will integrate Spectrum-X Ethernet switches into its Oracle Cloud Infrastructure (OCI), forming part of new AI supercomputers powered by the Nvidia Vera Rubin architecture. 

The collaboration expands Oracle’s cloud capabilities to meet rising global demand for generative and reasoning AI.

“Oracle Cloud Infrastructure is designed from the ground up for AI workloads, and our partnership with Nvidia extends that AI leadership,” says 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.”

The use of Nvidia Spectrum-X in OCI data centres represents a strategic step towards creating AI-ready network environments at scale. It also underscores Oracle’s commitment to offering the bandwidth, reliability and speed required for complex model training.

Meta integrates open networking at scale

Meta will deploy Spectrum Ethernet switches within its Facebook Open Switching System (FBOSS) – the company’s proprietary software platform for managing network switches across large-scale data centres.

The integration is designed to enhance network efficiency, enable faster AI model training and maintain the flexibility of Meta’s open networking approach.

“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 decision to deploy Nvidia Spectrum technology highlights a trend toward customisable, high-performance networking platforms that can evolve alongside AI innovation.

Spectrum-X – built for the AI era

The Nvidia Spectrum-X Ethernet platform, which includes Spectrum-X Ethernet switches and Spectrum-X Ethernet SuperNICs, is the first Ethernet solution purpose-built for AI data centres.

It enables hyperscale operators to interconnect millions of GPUs with high data throughput, providing the scalability required for trillion-parameter models and other intensive AI workloads.

Spectrum-X has already demonstrated record-setting efficiency, achieving 95% data throughput on large-scale AI supercomputers through its congestion-control technology.

Standard Ethernet, by contrast, typically operates at around 60% throughput under comparable conditions due to packet collisions and flow limitations.

This performance leap represents a major advance in both the economics and design of AI-scale networking.

The Spectrum-XGS Ethernet technology, part of the same platform, also introduces “scale-across” capabilities, allowing data centres to be linked across cities and continents – creating what Nvidia describes as giga-scale AI super-factories.

Spectrum-X builds on Nvidia’s broader full-stack architecture, which combines GPUs, CPUs, Nvidia NVLink and software to deliver consistent performance from compute to network.

The platform’s features include adaptive routing, congestion management and AI-driven telemetry – all designed to maintain efficiency and predictability across complex AI training and inference clusters.

Powering the future of AI data centres

As AI workloads expand, hyperscalers are prioritising networking innovation to maintain performance and control costs.

The adoption of Spectrum-X by Meta and Oracle underscores the increasing importance of purpose-built Ethernet solutions in enabling scalable, sustainable AI growth.

With Spectrum-X, Nvidia continues to position itself at the centre of the data centre transformation – offering a networking architecture engineered for the era of giga-scale AI.