明敏 发自 凹非寺 量子位 | 公众号 QbitAI 华为云算力再迎重大突破！ 刚刚落幕的华为全联接大会2025，一系列新进展发布—— 这距离CloudMatrix384超节点2025年4月正式发布仅半年，期间其 能力持续进化 ： 现阶段， AI行业内依旧被算力焦虑笼罩 。硅谷大厂近期在算力、芯片领域动作频频： OpenAI一边和博通自研AI芯片，一边向甲骨文抛出3000亿美元买算力；马斯克百天建成万卡超算集群，还计划向百万卡规模冲击，同时悄悄 布局芯片；Meta、AWS等企业也在积极获取更多算力资源……但算力的发展并非一蹴而就，它需要在单点技术上极致突破，还涉及芯片、硬 件、架构、软件、网络、能源乃至整个产业生态的协同演进。 放眼全球，能够输出澎湃算力的供应商，都离不开十数年、数十年的沉淀积累。 华为云作为其中一员，探索路径因所处产业阶段而显得尤为深刻：不仅需要在技术"无人区"重新定义算力运行规则；还需把握AI发展时机，通 过快速迭代响应产业海量需求。一步步成长为今天的"算力黑土地"。 AI算力云服务升级， 基于华为云刚刚发布的最新AI服务器规划， CloudMatrix的云上超节点规格将从384卡升级到未 ...

Core Viewpoint - Huawei Cloud is undergoing significant organizational restructuring and strategic focus to enhance its competitiveness in the AI era, with a particular emphasis on its "computing power black land" strategy, which is crucial for the company's future and its position in the AI landscape [2][29]. Group 1: Organizational Changes - A large-scale organizational adjustment is taking place within Huawei Cloud, involving the merger and restructuring of multiple core departments, indicating a strategic contraction as the company aims to address its ongoing losses [2][4][12]. - The restructuring has led to the removal of over 20 products since June, with significant impacts on key teams, affecting potentially thousands of employees [4][9]. - The focus of Huawei Cloud's business layout has shifted to "3+2+1," which includes core areas such as general computing, intelligent computing, and storage, alongside AI PaaS and security [7][12]. Group 2: Strategic Focus on AI and Computing Power - Huawei Cloud is betting heavily on computing power and AI, with plans to build a robust AI-native cloud infrastructure that includes AI databases and development tools [12][18]. - The company aims to leverage its Ascend chips and the Pangu model to create a competitive edge in the AI market, with a significant increase in the number of clients using its AI cloud services [18][23]. - The introduction of a new billing model based on "Tokens" allows clients to pay only for the computing power they use, lowering barriers for small and medium enterprises to access AI services [21][23]. Group 3: Market Position and Challenges - Despite achieving a revenue of 38.523 billion yuan in 2024, Huawei Cloud remains in a loss-making position, facing intense competition from rivals like Alibaba Cloud [9][24]. - The Chinese computing power market is characterized by both surplus and shortage, leading to aggressive pricing strategies among competitors, which Huawei Cloud must navigate to maintain its market share [4][24]. - Huawei Cloud's credibility has been challenged due to allegations regarding the originality of its Pangu model, which could impact client trust and procurement decisions [24][25]. Group 4: Future Outlook - The success of Huawei Cloud's "black land" strategy is critical for its long-term viability and profitability, as it seeks to transform AI demand into sustainable revenue [27][28]. - The company has established a global presence, serving over 170 countries and regions, which supports its strategic initiatives in the AI domain [29].

Core Viewpoint - The importance of chips is acknowledged, but the ability to provide the required computational results for customers is emphasized as more critical [1] Group 1: Huawei Cloud's Developments - Huawei Cloud is undergoing significant organizational restructuring, with a focus on enhancing computational power through its Ascend AI cloud services and Tokens services [1] - The CloudMatrix384 super node, launched in April, integrates 384 Ascend NPUs and 192 Kunpeng CPUs, achieving a computational scale of 300 PFlops [2] - The Tokens service has been integrated with the CloudMatrix384 super node, achieving a maximum of 2400 TPS and 50 ms latency, surpassing industry standards [2][3] Group 2: Performance and Growth Metrics - Huawei Cloud's overall computational capacity has increased by nearly 250% compared to the previous year, with the number of Ascend AI cloud service customers rising from 321 to 1714 [5] - The CloudMatrix384 super node can support training of large models, with the capability to connect 432 super nodes to form a 160,000-card AI cluster [2] - The deployment of over 40 CloudMatrix384 super nodes in Guizhou is part of a strategy to create a national computational network [5] Group 3: Market Position and Strategic Focus - Huawei Cloud ranks second in China's cloud service market with an 18% share, while Alibaba Cloud holds a 33% share [6] - The market demand is shifting from "cloud adoption" to "AI integration," prompting Huawei Cloud to streamline its operations and focus on maximizing the advantages of its Ascend AI and Pangu model combinations [6][7] - The organizational restructuring aims to concentrate resources on core areas that can leverage AI capabilities effectively [6][7]

Core Insights - The China International Big Data Industry Expo highlighted Huawei's commitment to building a "computing black land" to meet the exponential growth in computing power demand over the next decade [1][3] - Huawei Cloud's overall computing power has increased by nearly 250% compared to the same period last year, with the number of clients using Ascend AI cloud services rising from 321 to 1714 [3][4] Group 1: Computing Infrastructure - Huawei Cloud is developing a national computing network centered around key hubs in Gui'an, Ulanqab, Hohhot, and Wuhu to support global AI computing needs [3] - The CloudMatrix384 super node, deployed in Gui'an, is the largest of its kind, serving as a benchmark for the East Data West Computing project [3][4] - Huawei Cloud's supercomputing capabilities can reach 300 PFlops, and it can support training of large models with up to 1300 concurrent applications [4] Group 2: AI Services and Data Management - Huawei Cloud emphasizes the importance of high-quality datasets for AI model effectiveness and is working on creating AI-native data platforms [5] - The Tokens service from Huawei Cloud demonstrates significant performance advantages, achieving 2400 TPS with a latency of 50ms [5] - Huawei Cloud supports various mainstream open-source large models, enhancing their performance on the Ascend cloud [5] Group 3: Market Position and Strategy - Huawei Cloud has achieved leading market shares in key sectors such as government, industry, finance, and automotive, and has maintained a record of zero major incidents for 756 days [6] - The company advocates for an AI-native mindset to fully leverage AI's potential, emphasizing the need for businesses to adapt their applications, data, processes, and personnel around AI [6]

Core Insights - The demand for computing power is expected to grow exponentially in the next decade, potentially by tens of thousands of times, driven by advancements in AI and large models [3][4] - Huawei Cloud is committed to building a "computing power black land" to support this demand, leveraging its connectivity technology and data center resources [3][5] Group 1: Computing Power Infrastructure - Huawei Cloud aims to create a national computing power network centered around key hubs in Gui'an, Ulanqab, Hohhot, and Wuhu, establishing itself as a global AI computing power provider [3][4] - The overall computing power scale of Huawei Cloud has increased by nearly 250% compared to the same period last year, with the number of clients using Ascend AI cloud services rising from 321 to 1714 [3][4] - The CloudMatrix384 super node, deployed in Gui'an, can achieve a computing power scale of 300 PFlops, and can be expanded to form a 160,000-card AI cluster for training large models [4] Group 2: AI Services and Data Management - High-quality datasets are crucial for AI model effectiveness, and Huawei Cloud is working to create an AI-native data foundation that supports knowledge extraction and preparation for large models [5][6] - The Tokens service from Huawei Cloud demonstrates significant performance advantages in high-throughput scenarios, achieving 2400 TPS with a latency of 50ms [5] - Huawei Cloud has established a strong presence in key domestic industries, ranking first in sectors such as government, industry, finance, and automotive [6] Group 3: Strategic Vision and Innovation - Huawei Cloud emphasizes the importance of adopting an AI-native mindset to fully leverage AI's potential and drive innovation in business models [6] - The company has maintained a record of zero major incidents for 756 days, highlighting its commitment to security, stability, and continuous innovation [6]