At MWC 2026 Barcelona, Zhang Xiwei, the President of Huawei's Computing Product Line, made a groundbreaking announcement by introducing several super-node products and solutions internationally for the inaugural time. These include the Atlas 950 SuperPoD intelligent computing super-node, the Atlas 850E super-node server, the industry's pioneering general-purpose computing super-node TaiShan 950 SuperPoD, along with a series of products such as TaiShan 500 and TaiShan 200. The Atlas 950 SuperPoD, leveraging Lingqu interconnection technology, features a single cabinet with 64 cards as its foundational unit and supports high-speed interconnection of up to 8,192 NPU cards, catering to the needs of ultra-large-scale AI training and massive concurrent inference scenarios. The Atlas 850E super-node server is designed to be compatible with existing air-cooled data centers and allows for flexible expansion from 8 to 1,024 cards, enabling operators to incrementally upgrade their services from small-scale inference to cluster inference. The TaiShan 950 SuperPoD general-purpose computing super-node boasts capabilities such as sub-100-nanosecond ultra-low latency, terabyte-scale ultra-high bandwidth, and memory pooling, effectively tackling the challenges of high latency, substantial data transfer overhead, and low collaboration efficiency in general-purpose computing scenarios. Additionally, Huawei introduced the TaiShan 500 and TaiShan 200 series general-purpose computing products, which cater to high-, medium-, and low-computing-power scenarios, thereby meeting the diverse business requirements of operators. Huawei remains steadfast in its commitment to full open-source and openness, fostering developer innovation, promoting the flourishing of the computing industry ecosystem, consistently contributing to the openEuler open-source operating system, fully open-sourcing the CANN heterogeneous computing architecture, supporting mainstream open-source communities and projects within the industry, and enhancing the usability for developers.