TLDR
- NVIDIA surges as Meta, Oracle build next-gen AI data centers on Spectrum-X
- Meta, Oracle power massive AI upgrades with NVIDIA’s Spectrum-X Ethernet
- NVIDIA climbs after Meta and Oracle adopt Spectrum-X for AI expansion
- Big tech turns to NVIDIA Spectrum-X for faster, smarter AI networking
- NVIDIA gains as Meta, Oracle deploy ultra-efficient AI data infrastructure
NVIDIA’s stock rose by 2.49% to $187.61, continuing a steady upward trend throughout the session.
The gain followed early declines but shifted as major AI infrastructure news broke. Meta and Oracle announced large-scale integration of NVIDIA’s Spectrum-X Ethernet platform into their data center networks.
Meta Standardizes AI Infrastructure on Spectrum-X Ethernet
Meta began deploying NVIDIA Spectrum-X Ethernet switches to accelerate its internal AI workloads across large-scale data centers. This move supports the expansion of its Facebook Open Switching System (FBOSS) with next-gen Minipack3N switches. As a result, Meta aims to improve network control, reduce latency, and handle larger AI models more efficiently.
With AI models growing exponentially in size, Meta faces increased network demands across infrastructure. The new switches offer improved data throughput and reduced congestion, leading to higher reliability. The integration boosts Meta’s AI training capabilities, allowing faster deployment of AI-powered features across its platforms.
By advancing this network transformation, Meta sets the stage for deploying hyperscale AI solutions. Its infrastructure now supports scalable AI model training and inference across billions of interactions. This also aligns with its long-term roadmap for open networking and distributed computing.
Oracle Builds Giga-Scale AI Factories with NVIDIA Ethernet
Oracle announced it will construct new giga-scale AI factories using NVIDIA’s Spectrum-X Ethernet networking technology. These factories will leverage the Vera Rubin architecture, designed for massive AI workloads at high efficiency. Oracle plans to connect millions of GPUs using this system, offering ultra-low latency and high bandwidth.
This strategy supports Oracle Cloud Infrastructure’s focus on generative AI services for enterprise clients. By adopting Spectrum-X, Oracle ensures fast model training and deployment across geographically distributed facilities. This architecture allows seamless data flow and consistent performance under high-volume AI loads.
The company emphasized Spectrum-X’s adaptive routing and congestion control as critical to this transformation. These technologies reduce bottlenecks and enhance processing stability across global operations. Oracle’s move signals a commitment to scaling next-generation AI applications with reliable, purpose-built networking.
Spectrum-X Platform Sets New Benchmark in AI Networking
NVIDIA Spectrum-X Ethernet platform is purpose-built for AI data centers handling trillion-parameter models. It features Ethernet switches and SuperNICs optimized for high-throughput, low-collision networking. The platform allows hyperscalers to form massive GPU clusters with unprecedented efficiency.
NVIDIA reported that Spectrum-X delivers up to 95% data throughput using advanced congestion control. In comparison, traditional Ethernet often drops to 60% efficiency due to collision-prone flow management. These improvements mark a new benchmark in AI networking performance and cost-effectiveness.
The Spectrum-X platform complements NVIDIA’s broader stack, including GPUs, CPUs, NVLink, and software solutions. Together, they deliver a tightly integrated compute-to-network pipeline for AI-scale operations. This synergy helps organizations deploy faster and more predictable AI systems at industrial scale.
