Group 1 - The 2025 World Artificial Intelligence Conference (WAIC2025) held in Shanghai showcased advancements in computing power infrastructure and applications, indicating a high level of industry prosperity driven by AI technology [1] - Huawei's Ascend 384 super node was a highlight at the conference, featuring twelve computing cabinets and four bus cabinets, optimizing performance for model training and inference, marking a new development stage for China's AI computing infrastructure [1] - Suiruan Technology demonstrated its AI commercial applications and infrastructure, with its "Suiruan® S60" AI inference card being widely applied in various scenarios such as chatbots and online meeting summaries [1] Group 2 - The rapid development of large models is transforming the industry ecosystem, driving the systematic and clustered development of computing power infrastructure, with significant potential for collaborative innovation [2] - Despite a decrease in AI large model companies at the conference, computing power companies maintained high participation, indicating strong demand for computing power, with a focus on design, production processes, and cooling technologies [2] - A report by IDC and Inspur indicates that China's intelligent computing power is expected to grow by 43% in 2025 compared to 2024, reaching 1460.3 EFLOPS by 2026, which is double that of 2024 [2] Group 3 - The rapid growth of intelligent computing power reflects China's strong momentum in the digital economy and AI sectors, supported by policies, technological advancements, and market demand [3] - The future development of the computing power industry is viewed positively, with AI being a key driver of a new technological revolution and the rise of domestic AI computing power seen as an inevitable trend [3] - As digital transformation deepens, the demand for data collection, storage, processing, and analysis will increase, further driving the need for computing power [3]

Core Viewpoint - The article emphasizes the role of domestic computing power in enabling innovative internet applications, highlighting the achievements of Suiyuan Technology in AI computing infrastructure and commercialization [2][5]. Group 1: AI Computing Infrastructure - Suiyuan Technology showcased its latest achievements in AI computing infrastructure at WAIC 2025, focusing on the theme "The Fire of Chips Spreads" [2]. - The company has established intelligent computing centers in various locations, including Qingyang, Wuxi, and Yichang, which support the development of domestic AI [5]. - By the end of 2024, Suiyuan Technology built the first 10,000-card inference cluster in Qingyang, providing robust support for the "East Data West Computing" initiative [5]. Group 2: Product Offerings - The exhibition featured the "Suiyuan® S60" AI inference card, which has been widely applied in various internet scenarios, including chatbots and online meeting summaries [2]. - The DeepSeek integrated machine series, set to launch in early 2025, supports domestic CPU platforms and offers scene optimization capabilities for efficient AI business deployment [3]. Group 3: Commercialization and Partnerships - Since 2020, Suiyuan Technology has collaborated with Tencent to explore the commercialization of domestic computing power across different business scenarios, delivering tens of thousands of cards for high-demand applications [5]. - The company has successfully deployed inference cards for the "AI dressing" feature of Meitu's beauty camera, ensuring stable performance during peak usage [6]. - Suiyuan Technology is actively converting domestic AI computing power into actual commercial value through deep collaboration with various clients [7]. Group 4: Industry Insights - The rapid development of large models is transforming the industry ecosystem, driving the systematic and clustered development of computing infrastructure [9]. - Suiyuan Technology aims to provide inclusive, efficient, and reliable AI infrastructure solutions, fostering an open-source domestic AI ecosystem in collaboration with industry partners [9].