Meta and Oracle are upgrading their AI data centre infrastructure to the NVIDIA Spectrum-X Ethernet platform to boost performance and efficiency as they scale up to support trillion-parameter AI models.
Designed for industrial-scale AI workloads, NVIDIA Spectrum-X enables hyperscalers to connect millions of GPUs and accelerate training and deployment of generative and reasoning AI applications.
Unlike standard Ethernet, which struggles with congestion and throughput losses at scale, Spectrum-X’s congestion control technology achieves up to 95 percent network utilisation. This is crucial for unlocking the full potential of large-scale data centres powering the next generation of AI innovations.
“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,” said Jensen Huang, Founder and CEO of NVIDIA.
Meta will integrate Spectrum-X Ethernet switches into its Facebook Open Switching System to enhance the deployment speed and reliability of its AI-powered services for billions of users worldwide.
“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, Vice President of Networking Engineering at Meta.
Oracle is leveraging Spectrum-X to build giga-scale AI supercomputers to interconnect vast numbers of NVIDIA GPUs for advanced cloud-based AI solutions.
“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,” said Mahesh Thiagarajan, Executive Vice President of Oracle Cloud Infrastructure.
Entelechy Asia 