2月26日，全市纵深推进融入服务全国统一大市场建设暨深化要素市场化配置综合改革、优化营商环境 会议召开，深入贯彻党的二十届四中全会精神和习近平总书记在河南考察时重要讲话精神，落实党中央 有关重大决策部署和全省纵深推进融入服务全国统一大市场大会要求，动员全市上下全面深化"三标"活 动，紧抓快干，奋战开局，加快建设国内国际双循环枢纽和战略支点、要素市场化配置综合改革示范城 市和国家营商环境标杆城市。省委常委、市委书记安伟出席并讲话。 市委副书记、市长庄建球主持会议，市委常委、常务副市长陈宏伟部署重点工作。 市领导周富强、杜新军、陈明、王治安、虎强、樊惠林、李小虎、石秀田、陈红民、王滔、马志峰、张 艳敏、翟政、夏扬等出席会议。 安伟指出，构建全国统一大市场是畅通国内大循环、构建新发展格局，以高质量发展全面推进中国式现 代化的必然要求，推进要素市场化配置综合改革是我市融入服务全国统一大市场建设的核心支撑和转型 发展重大机遇，一流的营商环境是要素集聚的"强磁场"、市场联通的"润滑剂"。要提高站位、深化认 识，坚持改革引领，强化环境保障，在全省融入服务全国统一大市场建设中担重任、走在前，为培育发 展新质生产力、提升城市能 ...

Core Insights - Shenghe Jingwei Semiconductor Co., Ltd. has successfully passed the IPO approval for the Sci-Tech Innovation Board, marking it as the first company to achieve this in the Year of the Horse [1][2] - The company is the first and only enterprise in mainland China to achieve large-scale production of 2.5D silicon-based chip packaging technology, showcasing its strategic rise supported by Kunyuan Asset's FOF ecosystem [1] Industry Trends - The AI boom is reshaping the global technology landscape, with computing power becoming a focal point in the semiconductor industry, as evidenced by IDC's prediction that China's computing power will reach 1,460.3 EFLOPS by 2026, doubling from 2024 [3] - The demand for high-performance chip manufacturing is surging, with industry experts noting that computing power demand is doubling every 3 to 4 months, necessitating new evolution paths in the semiconductor sector [4] Company Performance - Shenghe Jingwei is recognized as one of the largest and fastest-growing advanced packaging enterprises globally, ranking as the 10th largest and 4th largest in mainland China according to Gartner [6] - The company has the largest 12-inch bumping capacity in mainland China and is the first to provide 14nm advanced process bumping services, filling a critical gap in the high-end integrated circuit manufacturing chain [6] - In the 2.5D packaging market, Shenghe Jingwei holds an impressive 85% market share in mainland China, demonstrating its leading position in this technology [7] Financial Aspects - Shenghe Jingwei plans to raise 4.8 billion yuan through its IPO to invest in three-dimensional multi-chip integration packaging projects and enhance its manufacturing capacity for advanced packaging technologies [7] - The company’s revenue growth rate from 2022 to 2024 is the highest among the top ten global enterprises in the advanced packaging sector [6] Strategic Vision - Kunyuan Asset emphasizes the importance of "patient capital" in supporting hard technology breakthroughs, fostering a collaborative investment ecosystem that nurtures leading tech companies [8] - The successful IPO of Shenghe Jingwei is seen as a testament to the deep trust and mutual achievement between capital and technology, highlighting the potential for significant industrial advancement driven by the synergy of computing power, core algorithms, and intelligent terminals [9]

对于成绩单和目标，上海市人大代表们怎么看？又有何建议？ 不确定性与技术变革下，机遇与挑战并存 在今年的上海两会上，上海市市长龚正作政府工作报告时表示，2025年，上海市生产总值达到5.67万亿 元、增长5.4%，好于预期。综合各方面因素，建议2026年全市生产总值预期目标增长5%左右。 然而，面对巨大的变化及严峻复杂的经济形势，各行各业的市场主体在实际经营中仍然不乏挑战，对于 现实的体感也不尽相同。 在吴坚看来，面对复杂的内外部环境，上海的外贸和外资不可避免地遭受到了冲击。内部转型方面，上 海作为超大城市，也正面临着产业结构深度调整的压力，如何在传统动能减弱的情况下，快速培育新动 能，是全年工作的核心突破口，也是值得持续关注的重点。 以朱柯丁所在的行业为例，他提出，一方面，需求增长很快，但系统建设的复杂度明显提高。算力相关 用电不是"多拉几条线"就能解决的，而是对电网结构、备用能力、调峰能力提出系统性要求。另一方 面，技术和工程标准提升，但项目回报周期偏长。大多数项目，前期投入大、专业门槛高，但回报更多 体现在长期运行效率和综合效益上，企业在资金和风险承受上压力不小。另外，在新业态的快速发展 中，规则体系却仍 ...

1月27日，在2026"智算申城"高峰论坛上，"上海市智算产业高质量发展倡议"正式启动。 该倡议由上海仪电联合多家龙头企业共同发起，旨在协同推进新一代智算中心建设，构建高水平智算云服务体系，攻坚高能效服务器与高密度机柜技 术、共建开放协同的自主可控产业生态，致力于将智算产业打造成为上海经济高质量发展的新引擎。 上海仪电（集团）有限公司党委副书记、总裁、上海市算力网络协会理事长张建明致辞时表示，国内算力产业正从"设施规模"向"综合效能"转变，形 成"硬件—智算云—模型—应用"四级联动架构。 张建明说，上海仪电作为上海数智化建设的国资主力军，正全力推进智算基础设施迭代升级与国产自主可控生态建设，包括：持续扩容超万卡级先进性 能算力集群，推动高效制冷技术规模化落地；构建仪电智算云与算力调度服务平台，促进算力资源在多元场景中的便捷接入与高效流通；推动国产AI芯片 与仪电智算云深度适配，构建自主可控光互连GPU超节点等。市算网协会将持续促进行业协作与产学研融合，共同为上海"2+3+6+6"现代产业体系构建和经 济社会高质量发展注入强劲动能。 中国信息通信研究院党委副书记王晓丽致辞时表示，算力是数字经济的核心生产力，是推 ...

人民财讯1月27日电，1月27日，在2026"智算申城"高峰论坛上，上海"算力经济赋能共创体系"正式发 布。该体系通过"验证池"与"商用池"双轨联动，着力破解算力资源到经济价值的转化难题。 ...

Summary of Conference Call Notes Industry Overview - The conference call discusses the **AIDC (Artificial Intelligence Data Center)** industry, highlighting significant growth and changes in demand for domestic computing power cards and large models in China [1][3][4]. Key Points and Arguments 1. **Growth in Domestic Card Shipments**: - Domestic card shipments are expected to grow rapidly in 2026, with an increase in overall deployment rates. The sector is currently at a low valuation but shows signs of fundamental improvement supported by various event-driven plans [1][4]. 2. **Market Performance**: - Companies like Century Interconnect and WanGuo Data in the US, along with domestic firms like Runze, have seen stock price increases, partly due to the US lifting restrictions on the H200 card. However, the primary demand is shifting towards domestic large models and cards [1][4]. 3. **Government Support**: - Government policies are crucial for the development of domestic computing power. Beijing and Shanghai are set to launch large-scale subsidies for projects exceeding 100 million yuan, with a 20% funding rate for projects that meet specific criteria [5][1]. 4. **Market Dynamics**: - The AIDC industry is experiencing a significant growth phase, with increased bidding activity expected by the end of 2025. The market is likely to evolve towards large-scale park development to meet customer expansion and stable delivery needs [3][6]. 5. **Competitive Landscape**: - Major players like ByteDance, Alibaba, Baidu, Huawei, and Tencent continue to dominate the market. ByteDance plans to deliver approximately 300-400 MW of computing power in 2026 [4][20]. 6. **Cost and Pricing Trends**: - The price per kilowatt is currently stable at around 280 yuan, with significant regional variations. Short-term market competition is intense, and no significant price turning point is expected in the next one to two years [7][9]. 7. **Liquid Cooling Technology**: - Demand for liquid cooling technology is increasing, with design capacities reaching 170 kW per cabinet. However, profit margins remain limited despite slight cost increases [10][12]. 8. **Profitability Challenges**: - Despite increased bidding volumes, the market remains focused on volume rather than profitability. Head companies are concentrating resources, which limits expansion and keeps costs high [16][19]. 9. **Regional Insights**: - Areas like Shaoguan and regions with lower electricity prices (e.g., Inner Mongolia and Xinjiang) have potential advantages, but overall project numbers are limited [17][19]. 10. **Future Demand Trends**: - There is a noticeable increase in demand for edge computing nodes and urban-level inference computing nodes, with high-cost performance solutions becoming mainstream [6][18]. Other Important Insights - **Storage Costs**: Rapid increases in storage costs (over 40%) are affecting project budgets, particularly for small and medium enterprises that need to focus on cost control [2][5]. - **Market Supply and Demand Mismatch**: There is a national oversupply issue, with scattered projects lacking a cluster effect. However, the market will still be dominated by large clusters from leading companies [19][24]. - **Energy Approval Processes**: Energy approval processes remain slow in major cities, impacting new project developments [22][23]. This summary encapsulates the key insights and trends discussed in the conference call, providing a comprehensive overview of the current state and future outlook of the AIDC industry.

Core Insights - Guizhou is transforming into a "digital highland" by leveraging its leading computing power, which is becoming a key indicator of economic and social development in the digital economy era [1][2] - The province has achieved a fundamental leap in its digital economy during the 14th Five-Year Plan period, with an annual growth rate of 16.2% in the intelligent industry, reflecting strategic optimization of computing power structure [2] - The collaboration between Guizhou and Shenzhen exemplifies a successful model of resource synergy, enhancing the efficiency of computing power utilization and supporting technological innovation in the Greater Bay Area [3][4] Industry Development - Guizhou's computing power capacity is projected to reach 150 EFLOPS, with over 97% of its computing power now being intelligent computing, positioning it as one of the strongest regions in the country for intelligent computing capabilities [2] - The province is actively promoting the integration of computing power into the real economy, leading to rapid growth in the digital industry and significant revenue increases in software and information technology services [2] - Guizhou is not only focusing on resource output but is also advancing the market-oriented allocation of data elements, creating a new ecosystem for data value extraction and enhancing the scale of the computing power market [4] Strategic Initiatives - The "East Data West Computing" initiative is being implemented to facilitate regional coordination and development in the digital age, showcasing a new logic for regional collaboration [1][2] - Guizhou has introduced policies such as the "computing power voucher" to lower the barriers for enterprises, enabling efficient cross-province resource matching and fostering a collaborative path of "Eastern demand + Western supply" [3][4] - The province's efforts in data collaboration and application innovation are expected to provide solid support for optimizing and reorganizing industrial and innovation chains across a broader scope [4]

Core Insights - The domestic GPU industry is poised for historic development opportunities by 2025, driven by technological breakthroughs and a growing ecosystem, reflecting strong market recognition of domestic "hard technology" [1][2] - The rapid development of domestic GPUs not only supports cutting-edge industries like artificial intelligence but also enhances supply chain security, contributing to high-quality economic development [1] Industry Development - The demand for computing power has surged due to the global rise of AI models, with China accelerating the construction of robust computing infrastructure [2] - By 2025, major tech companies in China, including Alibaba, Tencent, and ByteDance, are expected to invest approximately 450 billion yuan in AI computing power, with 30% allocated for domestic chip validation and adaptation [2] - The domestic GPU industry is transitioning from technological exploration to commercialization and scaling, with leading companies innovating architectures and integrating ecosystems to catch up with international giants [2][3] Technological Innovations - Domestic GPU manufacturers are striving to capture market share through technological innovation and differentiation strategies, with some companies achieving significant breakthroughs [3] - The introduction of supernodes as core computing units in intelligent computing centers is a key focus, with Huawei showcasing its Ascend 384 supernode, achieving a total computing power of 300 PFLOPS, surpassing international competitors [3] Market Dynamics - The emergence of open-source domestic models like DeepSeek is driving explosive growth in AI applications, with future demand for inference computing power expected to be significantly higher than training power [4] - The domestic GPU sector faces challenges in general computing and high-performance scenarios due to technological gaps and ecological barriers, particularly in comparison to NVIDIA's established CUDA ecosystem [5] Ecosystem Development - To address ecological shortcomings, domestic companies are exploring paths such as chiplet technology, collaboration with model manufacturers, and enhancing software ecosystems [5][6] - Strategic collaborations among companies like Wallen Technology, Jiyue Star, and Shanghai Yidian Smart Computing aim to create a comprehensive ecosystem for domestic chips, models, and computing power [6] Capital Market Trends - The domestic GPU industry is experiencing a notable IPO boom, with significant fundraising activities, including the record-setting IPO of Moore Threads on the STAR Market [7][8] - The market's enthusiasm for domestic AI chip companies is fueled by high valuations of international competitors like NVIDIA and AMD, alongside the successful turnaround of companies like Cambricon [8] - Analysts highlight that the accelerated IPO processes reflect strong market recognition of the domestic GPU sector, with funds raised primarily aimed at R&D and ecosystem development [8]

Core Insights - The report indicates that AI is transitioning from the "tool era" to the "partner era" by 2025, with a clearer development trend for 2026 [1] Infrastructure - The report emphasizes the importance of computing power and chips, identifying the computing economy as the primary engine of the intelligent industry, with unprecedented global demand for AI computing power driving the construction of large-scale data centers [3] - These data centers are evolving from traditional server hosting to AI company-led powerhouses integrating massive computing, storage, and network resources [3] - Cloud computing vendors are shifting investments from general computing resources to dedicated computing infrastructure that meets AI demands, leading to strategic partnerships with AI companies [3] - The rise of AI-native demands is reshaping chip innovation, with GPU dominance being challenged by NPU and growth in ASIC/FPGA technologies [3] - China is accelerating the construction of a self-controlled computing ecosystem, with domestic "chip + SDK + framework" solutions validated in trillion-level model training [3] Model Innovation - Pre-training determines the hierarchy of large models, while architectural innovation influences pre-training levels, with hybrid expert models becoming mainstream under computational constraints [4] - In 2025, large models will enter the "inference time," with breakthroughs in multi-modal deep inference and adaptive reasoning [4] - The report highlights a surge in research and development in physical AI and embodied intelligence, with world models and VLA frameworks becoming focal points [4] - A new order collaboration was announced by UBTECH, involving humanoid robot sales exceeding 50 million yuan, showcasing integration with AI large models [4] Application Landscape - AI is reshaping traffic entry points, transitioning from "people finding services" to "services finding people," marking a new interaction paradigm [7] - AI agents are developing closed-loop capabilities in perception, planning, decision-making, and execution, gradually replacing traditional apps [7] - The new generation of AI systems can process and understand multiple information types simultaneously, enhancing performance in complex scenarios and opening new possibilities for creative content generation and intelligent interaction [7] - The report predicts that as technology matures in the next 2-3 years, AI will become a standard tool across various industries, transitioning from a competitive advantage to a necessity [7] - The AI hardware sector is also gaining attention, with lightweight models and edge computing technologies driving AI capabilities to mobile devices, cars, and IoT devices [7] - More smart devices are gaining local AI processing capabilities, addressing data privacy, network latency, and cost efficiency issues [7]

Core Insights - The report indicates that AI is transitioning from the "tool era" to the "partner era" by 2025, with clearer development trends emerging for 2026 [1] Group 1: Infrastructure and Computing Power - The report emphasizes that computing power is the primary engine of the intelligent industry, with unprecedented global demand for AI computing power driving the construction of large-scale data centers [5] - Cloud computing vendors are shifting investments from traditional general computing resources to dedicated computing infrastructure that meets AI demands, evolving their partnerships with AI companies into deep strategic collaborations [5] - The demand for AI is reshaping chip innovation, with a significant market opening up for new opportunities in chip technology, challenging the dominance of GPUs and promoting the adoption of NPUs and custom chips [6] Group 2: Model Innovation - Pre-training determines the hierarchy of large models, while architectural innovation influences pre-training levels, with hybrid expert models becoming mainstream due to their cost-effective design [7] - By 2025, large models will enter a "reasoning time," with continuous breakthroughs in multi-modal deep reasoning and adaptive reasoning frameworks [7] Group 3: Embodied Intelligence and Applications - There is a growing trend in the research and development of physical AI and embodied intelligence, with significant orders being placed for humanoid robots, indicating a shift towards broader application exploration [8] - The report highlights that AI is reshaping traffic entry points, transitioning from "people finding services" to "services finding people," which is becoming the next generation of interaction paradigms [10] - The emergence of lightweight models and edge computing technologies is facilitating the proliferation of AI capabilities in devices such as smartphones, cars, and IoT devices, addressing key issues of data privacy and cost efficiency [10]