How Meta & Oracle Boost AI with Nvidia Spectrum-X Ethernet

Nvidia has confirmed that both Meta and Oracle are set to deploy its Spectrum-X Ethernet networking technologies.

This move is intended to advance the performance of their large-scale artificial intelligence data centres and improve training efficiency across their worldwide networks.

The adoption of Spectrum-X signals a major expansion of Nvidia's networking influence among hyperscale companies, allowing for greater efficiency and accelerated model training within these extensive computing environments.

Meta and Oracle are incorporating Nvidia Spectrum-X Ethernet switches into their infrastructure to help build next-generation AI facilities that can connect millions of graphics processing units (GPUs).

These complex systems are engineered to handle trillion-parameter models, which are large-scale machine learning systems that require high levels of data throughput and stability to function correctly.

Driving giga-scale AI performance

The development of AI is pushing data centres to evolve into what can be described as giga-scale AI factories.

"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 power this change in industry." 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 points to an increasing emphasis among hyperscalers on networking architectures specifically designed for AI tasks.

Spectrum-X provides an open and accelerated Ethernet solution that could reduce deployment times and increase overall efficiency for AI developers.

Oracle's AI supercomputer integration

Oracle plans to integrate Spectrum-X Ethernet switches within its Oracle Cloud Infrastructure (OCI) as a component of new AI supercomputers that will be powered by the Nvidia Vera Rubin architecture.

This partnership enhances Oracle’s cloud services to address the growing international demand for generative and reasoning AI.

Mahesh Thiagarajan, Executive Vice President of Oracle Cloud Infrastructure, explains that "Oracle Cloud Infrastructure is designed from the ground up for AI workloads, and our partnership with Nvidia extends that AI leadership. 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 integration of Nvidia Spectrum-X into OCI data centres is a strategic move toward establishing AI-ready network environments at a large scale.

Meta and open networking at scale

Meta is preparing to deploy Spectrum Ethernet switches within its Facebook Open Switching System (FBOSS), which is the company's software platform for managing network switches in its large data centres.

This integration aims to improve network efficiency, accelerate AI model training and preserve the flexibility of Meta’s open networking strategy.

"Meta’s next-generation AI infrastructure requires open and efficient networking at a scale the industry has never seen before," says 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."

The Nvidia Spectrum-X Ethernet platform is presented as the first Ethernet solution developed specifically for AI data centres.

According to Nvidia, it has achieved 95% data throughput in large-scale AI supercomputers by using its congestion-control technology.

In comparison, standard Ethernet typically reaches around 60% throughput under similar conditions.

This performance difference could represent a considerable advancement in the economics and design of AI-scale networking.

As AI workloads continue to grow, hyperscale operators are increasingly focused on networking innovation to manage performance and operational costs.

The adoption of Spectrum-X by major companies like Meta and Oracle shows the rising importance of purpose-built Ethernet solutions for enabling scalable and sustainable AI development.