Revolutionizing AI Networking With Co-Packaged Optics
At GTC 2025, NVIDIA unveiled the Spectrum-X™ Photonics Ethernet and Quantum-X Photonics InfiniBand switches, marking a major breakthrough in AI networking. Designed to support AI factories at unprecedented scale, these co-packaged optics networking switches enable seamless connectivity across millions of GPUs while drastically reducing energy consumption and operational costs.
By integrating silicon photonics directly into switches, NVIDIA has redefined hyperscale networking, delivering 3.5x greater power efficiency, 10x improved network resilience, and 1.3x faster deployment compared to traditional networking architectures.
NVIDIA Spectrum-X and Quantum-X Photonics
The NVIDIA Spectrum-X Photonics Ethernet platform sets a new standard for AI networking, offering:
- 1.6 Terabits per Second (Tb/s) per port, achieving higher bandwidth density for hyperscale AI factories.
- Configurations ranging from 128 ports of 800Gb/s to 2,048 ports of 200Gb/s, delivering up to 400 Tb/s of total throughput.
- 63x greater signal integrity, minimizing data loss across complex AI infrastructure.
The NVIDIA Quantum-X Photonics InfiniBand switch brings next-generation scalability to AI compute fabrics with:
- 144 ports of 800Gb/s InfiniBand based on 200Gb/s SerDes, optimized for high-performance AI workloads.
- 2x Faster Speeds and 5x Higher Scalability compared to previous generations.
- Liquid-cooled design, efficiently managing heat for maximum performance and longevity.
AI Factories Demand Next-Generation Networking
As AI factories grow in complexity and scale, traditional networking solutions struggle to keep pace. NVIDIA’s silicon photonics networking architecture ensures that multi-GPU AI workloads operate with lower latency, higher efficiency, and improved reliability.
Key benefits of NVIDIA’s co-packaged optics approach:
- 4x Fewer Lasers, reducing power consumption while maintaining signal integrity.
- Optimized for AI and HPC, ensuring efficient GPU-to-GPU communication at hyperscale.
- Industry-leading integration of optics and networking, paving the way for the next wave of exascale AI compute fabrics.
Build AI-Optimized Infrastructure With AMAX
As an NVIDIA Elite Partner, AMAX specializes in end-to-end AI infrastructure deployment, including racking, cabling, integration, and on-site installation for large-scale AI and HPC environments. Our expertise ensures seamless deployment of NVIDIA’s latest networking and compute technologies, optimizing performance for next-generation workloads.
Visit AMAX at GTC 2025
Visit AMAX at GTC 2025, Booth #2018, from March 17-21, showcasing the latest advancements in AI and liquid cooling technologies. Attendees can experience a live pumped two-phase liquid cooling demonstration and learn how to build a secure on-prem AI chatbot with NVIDIA AI Workbench, presented by AMAX’s engineering team.