By using NVIDIA’s Spectrum-X Ethernet switches, both companies can speed up AI model training while managing network load more effectively.
Meta and Oracle are adopting NVIDIA’s Spectrum-X Ethernet switches to enhance their AI data centres as demand for high-performance infrastructure continues to rise. The move signals a broader shift among major cloud and social media companies towards networks designed specifically for large-scale AI workloads.
The NVIDIA Spectrum-X platform combines Ethernet switches and SuperNICs to deliver high data throughput and improved congestion control. It has demonstrated up to 95% network efficiency, compared with roughly 60% in standard Ethernet systems. The technology also enables large-scale interconnections between data centres across regions.
“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.
Integrating Spectrum-X into Meta’s Facebook platform to manage and control data traffic across its vast network of data centres. Meta focuses on training larger AI models while maintaining consistent performance.
“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.
Oracle is deploying Spectrum-X within its Oracle Cloud Infrastructure (OCI) to connect millions of GPUs and build large-scale AI supercomputers. The company aims to accelerate AI model training and deployment for its global enterprise customers.
“By adopting Spectrum-X Ethernet, we can interconnect millions of GPUs without network speed delay efficiently, so our customers can more quickly train, deploy and benefit from the next wave of generative and reasoning AI.” said Mahesh Thiagarajan, executive vice-president of Oracle Cloud Infrastructure.
The adoption of Spectrum-X by Meta and Oracle reflects a growing focus across the technology industry on purpose-built AI networking, as hyperscalers expand infrastructure to support the next generation of generative and reasoning AI systems.
