In the future, CloudMatrix super nodes can build large clusters of more than 10,000 slices to provide computing power,
For AI computing, the Atlas 950 SuperPoD, powered by UnifiedBus, integrates 64 NPUs per cabinet and can scale up
This paper describes the "AI and Communication" scenario and the general principles for performance definition and provides an
supernode Key Points Supernodes = system efficiency: supernodes tightly integrate thousands of GPUs to prioritize
Huawei''s recently-announced Atlas 950 SuperPoD, Nvidia ''s DGX SuperPOD, and AMD ''s upcoming Instinct MegaPod
The same day, Huawei announced that its Ascend AI supernode offered “day zero” full support for DeepSeek V4,
At HUAWEI CONNECT 2025 in Shanghai, Rotating Chairman Eric Xu detailed the company''s newest high-performance computing
The AI Server Market is set to skyrocket, expected to reach USD 1.84 trillion by 2033 from USD 126.34 billion in 2024,
Yang explained, "Huawei has developed the groundbreaking SuperPoD architecture based on our UnifiedBus
Huawei executives said surging demand driven by agentic AI is accelerating the need for large-scale computing
Huawei revealed the vital parameters of the latest AI chip cluster. It says the CloudMatrix 384 uproots on the super
To unlock the full potential of SuperPod-scale infrastructure, we need a new system that runs disaggregated workloads, maintains
Import AI runs on lattes, ramen, and feedback from readers. If you''d like to support this, please subscribe. Subscribe
uring HUAWEI CONNECT 2023, David Wang, Huawei''s Executive Director of the Board, Chairman of ICT
Welcome to Import AI, a newsletter about AI research. Import AI runs on lattes, ramen, and feedback from readers. If
Ascend AI Cloud Service offers instant access to immense yet cost-effective AI computing power, a reliable platform for training and
Atlas 500 Pro AI Edge Server Maintenance and Service Guide (Model 3000) 18 About This Document Introduction Components
What to read next Atlas 950 runs on Huawei''s Ascend AI chips and works with CANN, its open-sourced compute
The implications for the development of China''s AI infrastructure are substantial. By creating an open ecosystem
This supernode uses 384 Ascend 910C NPUs and 192 Kunpeng CPUs as its core, connected via Huawei''s self
The rapid evolution of large language models (LLMs), driven by growing parameter scales, adoption of mixture-of
Huawei shipped Pangu Ultra MoE with 718B parameters trained entirely on Ascend AI silicon. Analysis of
Last September, Huawei disclosed for the first time details of its chip development roadmap, including the Ascend
Huawei positions its Atlas 950 SuperCluster to support training and inference workloads for AI models with hundreds
Purpose This document describes the Atlas 500 Pro AI edge server (model 3000), Atlas 500 Pro (model 3000) for short, in terms of
And at Huawei, Ascend chips are the foundation of our AI computing strategy.” He explains that Huawei''s goal is to
Contact us today for product inquiries, custom kits, or integration support