Huawei Atlas 960E SuperPoD with 4,096 NPUs and 8 EFLOPS FP8
Huawei Atlas 960E SuperPoD: First NPO/Hi-ONE Optical Node, 4096 NPUs, 8 EFLOPS FP8
Optical co-packaging via Hi-ONE NPO points to where large-scale AI cluster interconnects are heading, trading discrete pluggable modules for integrated optics with measurable savings in parts count and power. Hardware engineers planning training clusters and procurement teams evaluating alternatives to NVIDIA-based superpods will find the scale and efficiency figures directly comparable.
Huawei has launched the Atlas 960E SuperPoD, built on its UnifiedBus interconnect architecture and the Hi-ONE NPO optical node.
A single cluster scales up to 4,096 NPUs with FP8 compute reaching 8 EFLOPS.Huawei says the design cuts roughly 48,000 800G optical modules compared with conventional setups and reduces power consumption by 550 kW.
Huawei Atlas 960E SuperPoD: First NPO/Hi-ONE Optical Node, 4096 NPUs, 8 EFLOPS FP8
Huawei's Atlas 960E SuperPoD pairs UnifiedBus with Hi-ONE NPO optics for up to 4,096 NPUs, 8 EFLOPS FP8, and a claimed 550 kW power cut versus about 48,000 800G optical modules.
This report is categorized under AITOP's "Products" desk, focusing on China's AI industry and domestic foundation model ecosystem. Within our cross-regional analytical framework, it directly maps to US-China foundation model benchmarks and commercialization comparison and AI developer toolchains, coding agents, and engineering workflows. Key entities and technical topics under continuous tracking include Huawei, Atlas 960E, SuperPoD, NPU.
Primary source verification: published directly by pandaily with first-party authority.
