Huawei unveiled the Atlas 960E SuperPoD and accelerated the launch of its Ascend 960DT AI chip to the first quarter of 2027, three quarters earlier than originally planned, according to Huawei. The Ascend 960PR is expected in the third quarter, with a new Ascend generation planned each year and the Ascend 970 and 980 scheduled for 2028 and 2029. The Atlas 960E combines Ascend 960 chips with Huawei's UnifiedBus interconnect and Hi-ONE near-packaged optics, scaling to 4,096 NPUs and delivering 8 EFLOPS of FP8 compute performance with up to 1 petabyte of high-bandwidth memory. Huawei said the system uses 5,500 Hi-ONE optical engines instead of the 48,000 800G optical modules traditionally needed, cutting power consumption by more than 550 kilowatts and providing 99.8% system availability. Multiple SuperPoDs can connect into a SuperCluster supporting up to 512,000 NPUs, with a multi-rail topology potentially extending to 1 million NPUs. Reuters reported Huawei is struggling to produce enough AI computing equipment to satisfy demand in China, and rotating chairman Eric Xu said the company has no plans for large-scale overseas expansion for now. Limited chip supply and Nvidia's widely used CUDA software remain major barriers to broader adoption.