Page 11 - Edgecore Product Catalogue 2026

Rail-Optimized Topology for Best Optimized Programmable Switch

Rail-Optimized Topology for best optimized performance of
DDN/LLM Performance of AMD MI325X 256 GPU AI Cluster
AMD MI325X 256 GPU AI cluster The EXP series is powered by the Xsight Labs X2 programmable switching silicon, delivering
high-performance, low latency connectivity for AI fabrics, cloud data centers, and HPC networks.
Combined with open networking software and DPU AIC solutions, the EXP series provides a scalable,
intelligent infrastructure with advanced traffic management and telemetry for next-generation AI workloads.
Spine Switch
AIS800 800G
TH5 (800G)

Rail#1 Rail#2 Rail#3 Rail#4 Rail#5 Rail#6 Rail#7 Rail#8
Leaf Switch
AIS800
TH5 (800G)
EXP800-16O EXP400-32X EXP100-32X
400G
16 × 800G OSFP800 ports, each 32 × 400G QSFP112 ports, each 32 × 100G QSFP28 ports, each
can breakout to 2×400G, 4×200G can breakout to 2 × 200G, can breakout to 4 × 25G,
AMD MI325X or 8×100G or 4 × 100G or 4 × 10G
GPU Servers
Xsight Labs X2 Xsight Labs X2 Xsight Labs X2
Intel® Xeon® D-1713NTE Intel® Xeon® D-1713NTE Intel® Xeon® D-1713NTE
NIC X8 (400G)

GPU Server #1 GPU Server #2 GPU Server #3 GPU Server #32
Option 2: AIS800-32 series switch 800G to 2x 400G breakout cables

HPC Switch

The HPC series is designed for high-performance computing, delivering ultra-low latency Edgecore DPU AIC
and high-bandwidth connectivity. Engineered for scientific computing, simulation, modeling, AIN400-2QS Add-In-Card (AIC)
and large-scale parallel applications, the HPC series provides a resilient and scalable foundation for World's First 800G DPU
modern HPC networks. Dual 400GbE connectivity and PCIe Gen5 ×16 performance
64-core Arm® Neoverse N2 compute engine with up to 64GB DRAM
Accelerated AI/GPU communications with RDMA and KV Cache support

AIS800-64OTU1
64 × 800G OSFP800 ports
51.2 Tbps
Broadcom® Tomahawk Ultra
Intel® Xeon® Processor D-1713NTE

10 11