Data Center Network Scalability | Corning GlassWorks AI Solutions | Corning

Frequently Asked Questions

• How can AI data center scaling be made easier with advanced technologies like Hopper, Blackwell, and Rubin?

  Innovative fiber optic solutions are the key to managing the increasing density and complexity of AI networks while scaling with GPUs like Hopper, Blackwell, and Rubin. Corning shuffle cables and passive adapter boxes simplify scaling by efficiently breaking out fibers, splitting 400G inputs into 2×200G outputs, and connecting GPUs across multiple switches. Smaller OD cables and MMC connectors optimize rack space and minimize pathway congestion, making scaling seamless and cost-effective.

• How can large-scale AI data center deployments be streamlined?

  Preterminated assemblies, smaller OD cables, and MMC connectors streamline deployments, reducing installation time and complexity. As fiber density increases with AI workloads, deploying Corning cables with Flow Ribbon technology simplifies tray management and installation, and reduces cable thickness by up to 60%. Shuffle solutions enhance deployment efficiency by meshing fibers and mapping polarity to accommodate larger cluster sizes.

• How can costly disruptions caused by port flapping be reduced?

  Corning’s high-quality, preterminated fiber assemblies and MMC connectors ensure precise connections to reduce the risk of port flapping. Passive adapter boxes with optical flex circuit technology help manage fiber polarity and routing, eliminating common errors and minimizing disruptions in high-density networks.

• What’s the most cost-effective way to shuffle fiber to scale out a network?

  Corning offers two scalable shuffling methods: manual fiber shuffling between connectors and passive adapter boxes utilizing optical flex circuits. These solutions split 400/800G inputs into multiple 200G outputs, efficiently meshing fibers across switches and GPUs. Smaller connectors, like MMCs, further reduce labor and space costs, ensuring cost-effective scale-out for dense AI networks.

• How can a data center be scaled while meeting environmental goals?

  Corning’s smaller OD cables and Flow Ribbon technology reduce material use and carbon footprint while maximizing space efficiency. Sustainable packaging and eco-conscious cable designs cut transportation waste and support environmentally responsible scaling. As multicore fiber innovations continue to emerge, further reductions in material and tray layers can be achieved. Corning’s in-region manufacturing capabilities also reduce environmental impacts from shipping.

• How does regional manufacturing support scalability for AI data centers?

  Corning is a highly regarded global fiber manufacturer with the resources and expertise in-region to support the unprecedented demands of AI infrastructure. In-region manufacturing ensures faster delivery and a reliable supply chain, enabling quicker project execution to support the scalability of your AI data center, along with reduced shipping costs and environmental impact.