12月9日，"昇腾CANN开源开放沙龙·上海站"成功举办。本次活动由华为技术有限公司主办，上海昇腾 人工智能生态创新中心（以下简称"创新中心"）、中国计算机学会上海活动中心承办，上海昇思生态创 新中心、上海市人工智能学会、上海市漕河泾新兴技术开发区发展总公司等单位共同协办。活动聚焦 CANN开源开放的技术路径与生态共建，吸引了众多华东地区的开发者、企业代表及专家学者积极参 与，现场交流氛围热烈，共同探讨算力技术的深度落地与产业融合。 共话开源生态，展望智能未来 中国计算机学会（CCF）上海副主席丁炎在开场致辞中表示，上海作为中国人工智能产业高地，展现出 强劲的创新活力与丰硕成果。他呼吁共同构建开源开放的AI基础软硬件体系，并建议以开发者为中 心，依托创新中心推动区域技术生态建设，为人工智能发展注入新动能。 中国计算机学会（CCF）上海副主席丁炎 华为上海战略与业务发展部部长董浩在致辞中指出，昇腾扎根上海以来，已联合300多家生态伙伴孵化 超1000个AI解决方案，培养5000余名开发者。他表示，昇腾将持续践行"硬件开放、软件开源、使能伙 伴、发展人才"战略，坚持全面开源开放，以开发者为中心加速创新。 华为上海 ...

Core Insights - The "Ascend CANN Open Source Salon - Zhejiang Station" was successfully held, focusing on the open-source strategy of CANN to enhance AI computing capabilities in Zhejiang [1][10] - The event gathered over 260 partners from government, industry, academia, and research to promote the development of the Ascend ecosystem [1][3] Group 1: Event Overview - The event was co-hosted by Huawei Technologies Co., Ltd. and Hangzhou High-tech Innovation Group, with support from various local institutions [1] - Key figures from the local government and Huawei emphasized the importance of building a leading industrial ecosystem in Zhejiang [3] Group 2: CANN Open Source Strategy - Huawei's Ascend ecosystem aims to facilitate the application of AI in Zhejiang and foster talent development through the CANN open-source strategy [3][4] - CANN's architecture allows developers to optimize and develop at multiple levels while ensuring performance and flexibility [3] Group 3: CANN Ecosystem Co-Creation Plan - The "CANN Ecosystem Co-Creation Plan" was launched, which includes five core support areas: providing 240P high-performance computing power, hosting over 10 industry summits and competitions, training 1,000 developers, creating 20+ benchmark application scenarios, and forming 20+ expert teams [5] - The first phase of the plan will focus on innovation in the power industry [5] Group 4: Developer Experience Sharing - Developers shared experiences on building AI-native applications based on the Ascend CANN open-source ecosystem [7] - A technical discussion highlighted practical challenges in optimizing inference engine performance and operator development [7][9] Group 5: Practical Experience Zones - The event featured hands-on experience zones for developers to engage with different operator development methods and platforms [9] - Participants reported improved understanding of performance tuning and scenario adaptation through practical exercises [9] Group 6: Future Prospects - The Ascend CANN open-source salon will continue to expand to more cities, connecting developers and providing more application cases and open-source resources [10] - The initiative aims to promote the prosperity and development of the AI hardware and software ecosystem [10]

Investment Rating - The report maintains a positive outlook on the domestic AI chip industry, particularly highlighting the advancements of leading companies such as Huawei, Cambricon, and Moore Threads [4]. Core Insights - The domestic AI computing industry is undergoing significant changes in both supply and demand, with a rapid increase in Token consumption driving AI capital expenditure growth [14][5]. - Moore Threads, founded in 2020, focuses on the development of fully functional GPU chips and has a strong team with backgrounds from NVIDIA, positioning itself as a leading player in the domestic AI chip market [4][25]. - The company's products are designed to meet diverse computing needs, integrating AI acceleration, graphics rendering, and other capabilities into a single GPU chip [38][25]. Summary by Sections 1. Domestic AI Chip Industry - The domestic AI Token consumption has increased by 300 times over 1.5 years, with daily consumption surpassing 30 trillion by mid-2025 [14]. - Major internet companies are significantly increasing their AI-related capital expenditures, with Alibaba's projected spending exceeding 380 billion over three years [15][19]. 2. Moore Threads: Ecosystem, Architecture, and Clusters - Moore Threads has launched four generations of GPU architectures, covering AI computing, cloud computing, and personal computing [25][38]. - The company’s AI computing products are primarily delivered in cluster form, with significant revenue growth expected from AI computing cards starting in 2024 [49][50]. 3. Company Positioning and Product Development - Moore Threads' products are positioned as fully functional GPUs, supporting various computing precisions to cater to different application scenarios [38][39]. - The company has a robust product matrix that includes AI computing cards, professional graphics acceleration products, and desktop graphics cards [26][39]. 4. Competitive Landscape - The leading domestic AI chip manufacturers include Huawei, Cambricon, and Moore Threads, with their latest products nearing NVIDIA's H100 performance levels [4][19]. - The report emphasizes the importance of software ecosystems and interconnectivity protocols being developed by companies like Huawei and Haiguang to enhance the overall AI chip ecosystem [6][7]. 5. Financial Performance and Projections - Moore Threads is expected to achieve significant revenue growth, with total revenue projected to reach 7.01 billion in the first half of 2025, a 60% increase from the previous year [42][43]. - The company maintains a stable gross margin of around 70%, with expectations of turning profitable as AI computing cards gain market traction [43][44].

Core Viewpoint - Huawei Cloud is undergoing a structural adjustment to focus on key business areas and aims to achieve profitability by 2025 after reporting losses in 2024 [1][3]. Group 1: Organizational Changes - Huawei Cloud has announced a restructuring that involves the product department, public cloud services department, and R&D department, affecting dozens of lower-level departments and organizations [1]. - The adjustment will concentrate on a "3+2+1" business model, which includes three areas: general computing, intelligent computing, and storage; two areas: AI PaaS and databases; and one area: security [1]. Group 2: R&D and Innovation - The restructuring is likely a response to internal management issues, technical disputes, and research efficiency problems within the Noah's Ark Lab, as well as controversies surrounding the Pangu large model [3]. - Huawei Cloud aims to enhance R&D management and overall innovation efficiency through this adjustment [3]. - Huawei Cloud plans to increase investment in the Pangu large model and support various industries in utilizing their large models effectively [3]. Group 3: Strategic Focus - Huawei Cloud has been optimizing its business structure in recent years, transitioning from the Cloud and Computing BG to a standalone Cloud BU, which plays a crucial role as a new growth engine for Huawei [4]. - The cloud business is seen as critical in the context of the AI wave, presenting both opportunities and challenges for the company [4].

Group 1 - Huawei announced the full open-source of Ascend hardware enabling CANN, along with the Mind series application enabling suite and toolchain, aimed at accelerating innovation among developers [1] - CANN is positioned against NVIDIA's CUDA, providing a high-performance operator library and various development methods, while the MindSpore open-source framework supports efficient development [1] - The "2+1+X" model of MindX is designed to lower industry development barriers, promoting Ascend computing as a comprehensive AI infrastructure [1] Group 2 - The construction of an ecosystem is seen as an inevitable trend, with a need to enhance the compatibility efficiency with CUDA and reduce performance loss [1] - The deep collaboration between domestic frameworks like MindSpore and chips is essential to form a self-sustaining loop of "chip-framework-application" [1] - Ascend's strong order fulfillment capability is expected to drive continuous expansion of the industry chain, with the Ascend 384 showcasing system-level performance advantages [1] Group 3 - Advanced Communication has officially launched the Ascend A800I A2 large model integrated machine [2] - Tuo Wei Information has established a comprehensive strategic partnership with Huawei, focusing on "Kunpeng/Ascend AI + industry large models + Harmony" [3]