Liquid cooled full rack and SuperPod: UNIS develops liquid-cooled full rack products, accelerating the progress of liquid cooling technologies verification and commercialization, to meet the rising computing power demand and higher power consumption requirement. The company's SuperPods product aims for computing clusters with over 1,000 GPUs. The company plans to co-develop with CSP customers, leveraging company's hardware and networking capabilities. 13 December 2025 | 12:05AM HKT Equity Research UNIS (000 ...

Core Insights - Huawei Technologies has introduced new AI chip technology, specifically its SuperPod technology, to enhance computing power and compete with Nvidia's market dominance [1] - The company operates a super cluster with approximately 1 million graphic cards, indicating significant infrastructure investment [1] - Huawei's founder acknowledged the company's current limitations in single chip output compared to the US but emphasized the effectiveness of cluster-based computing to achieve desired results [2] Group 1: Huawei's Technological Developments - Huawei's SuperPod technology can link up to 15,488 graphic cards with Ascend AI chips, showcasing its ambition to challenge Nvidia [1] - The company is focusing on bundling multiple semiconductors together to compensate for the lower power of individual AI chips [6][7] - A new lineup of AI chips is planned for release over the next three years, including the Ascend 950PR, 950DT, 960, and 970 [7] Group 2: Market Context and Competition - Chinese tech stocks have seen a surge due to progress in developing homegrown AI and chips, with companies like Alibaba and Baidu gaining important clients [3] - The Chinese government has advised major tech firms against using Nvidia's RTX Pro 6000D and H20 chips, aiming to reduce reliance on Nvidia and promote domestic alternatives [4][5] - This creates a market opportunity for domestic chipmakers like Huawei and Cambricon to fill the gap left by Nvidia's restricted presence in China [6]