AI and GPU clusters
A perfect building block for small to medium AI installations. It delivers predictable, low-latency performance and full wire-speed switching, keeping GPU nodes and storage traffic flowing without bottlenecks. The compact design and low power usage make it suitable for AI labs, research environments, and smaller rack deployments.
Future proofing for existing networks
The CRS804 DDQ can act as a high speed core or aggregation switch even in older racks. Each 400G QSFP-DD port is built from eight independent data lanes. This enables flexible breakout options, including two 200G links for AI workloads or lower-speed connections such as 100G or 40G. You can deploy the CRS804 DDQ in your existing network today and upgrade to higher speeds over time without replacing the switch. With the right breakout cables, the CRS804 DDQ can interconnect virtually any device with SFP or QSFP ports, regardless of generation.