Core Viewpoint - Huawei's latest AI chip, Ascend 950PR, does not currently provide a significant advantage over NVIDIA products for domestic cloud providers, but it has achieved a major upgrade by becoming compatible with the CUDA ecosystem [1] Group 1: Key Breakthroughs - The Ascend 950PR features a significant breakthrough in CUDA compatibility, with the upgraded CANN Next software stack introducing a SIMT programming model that closely aligns with CUDA's native functions [2] - CANN Next is designed to create a development environment that can nearly seamlessly replace CUDA, allowing developers to maintain their existing programming habits while optimizing for Ascend chips [2] Group 2: Market Demand and Production Capacity - Major cloud companies like ByteDance and Alibaba are reportedly planning to purchase the Ascend 950PR in bulk, with Huawei aiming for a production capacity of 750,000 units this year [3] - The chip supports low-precision computing formats, with peak performance of 1 PFLOPS for FP8 and 2 PFLOPS for FP4, and features a memory bandwidth of 2 TB/s [3] Group 3: Industry Context - There is a pressing need for domestic alternatives to NVIDIA's computing products in the Chinese market, especially among leading cloud providers facing strict regulatory constraints on purchasing NVIDIA chips [3] - Huawei is leveraging the CANN Next software stack and Ascend 950PR chip to strengthen its position in the domestic AI industry, although challenges remain regarding production capacity and the readiness of clients for large-scale deployment [3]

Core Viewpoint - Nvidia's market share in China has plummeted to 0%, with no sales of its AI chips, as confirmed by CEO Jensen Huang [1][3]. Group 1: Market Dynamics - Nvidia once held a 95% market share in China, but it has now been completely overtaken by domestic competitors [1]. - The Chinese AI chip market is dominated by local companies such as Huawei, Alibaba, and Cambricon, which have developed competitive products that outperform Nvidia's offerings [3][5]. - Huawei's latest AI accelerator card, Atlas 350, features the Ascend 950PR processor, which is 2.87 times more powerful than Nvidia's H20 [3]. Group 2: Competitor Landscape - Alibaba's self-developed GPU chip, Zhenwu 810E, has shipped over 470,000 units and is used by more than 400 domestic enterprises, matching the performance of Nvidia's H20 [3]. - Cambricon has achieved profitability in 2025, earning over 2 billion yuan, and has significantly replaced Nvidia's chips in the market [5]. - Haiguang Information has developed its own DCU chip, which also matches the performance of Nvidia's H20 and has been widely adopted by domestic companies [5]. Group 3: Emerging Players - The "GPU Four Little Dragons" including Moore Threads, Muxi Technology, Biran Technology, and Tian Shu Zhixin have successfully gone public, securing substantial funding for research and development [7]. - Numerous other GPU companies are entering the AI sector, contributing to the trend of domestic substitution of Nvidia's chips [8]. Group 4: Industry Trends - The tightening of U.S. export controls on AI chips has led to a strong emphasis on "domestic substitution" within China's AI chip industry, with more companies opting for local solutions over Nvidia's products [8].

Investment Rating - The semiconductor industry maintains a "Buy" rating for companies such as 天数智芯 (Tianshu Zhixin) and 航天电器 (Aerospace Electric) [2][6] Core Insights - The semiconductor price increase is impacting end products, with mobile brands like OPPO and vivo adjusting their prices due to rising semiconductor and storage costs, initiated by AI demand [3] - Huawei has launched the Ascend 950PR computing acceleration card, which significantly outperforms previous models and is expected to drive demand in various AI application scenarios [3] - The semiconductor industry is experiencing a shift in pricing power towards leading brands with scale and supply chain advantages, as consumer electronics sales face pressure [3] Industry Dynamics - The semiconductor sector saw a weekly decline of 1.78% in the Shenwan Semiconductor Index, closing at 7,309.41 on March 20 [12] - The semiconductor materials sector experienced the largest drop of 6.01%, while discrete devices saw a rise of 5.17% during the same period [15] - The top ten stocks in the semiconductor sector saw significant weekly gains, with 源杰科技 (Yuanjie Technology) leading at 26.80% [22] Company Focus and Earnings Forecast - Tianshu Zhixin (9903.HK) is forecasted to have an EPS of -5.45 in 2024, with a "Buy" rating [6] - Aerospace Electric (002025.SZ) is expected to have an EPS of 0.76 in 2024, also rated as "Buy" [6] - 芯原股份 (Xinyuan Co., Ltd.) remains unrated with an EPS forecast of -1.20 in 2024 [6] Global Semiconductor Sales - Global semiconductor sales reached $82.54 billion in January 2026, with a year-on-year growth of 46.1%, and China accounted for $22.82 billion of this total [34] Semiconductor Equipment Sales - In Q3 2025, semiconductor equipment sales in mainland China reached $14.56 billion, marking a year-on-year increase of 12.61% [37] Memory Chip Market - The price of NAND chips has been rising due to increased AI demand, with the average price for 512Gb TLC wafers at $23.13 as of March 9, 2026 [47] - DRAM prices have also seen fluctuations, with DDR5 prices reaching $39.17 as of March 20, 2026 [48] Strategic Moves - 和顺石油 (Heshun Petroleum) is entering the semiconductor sector by acquiring a stake in 奎芯科技 (Kuixin Technology) for 540 million yuan, indicating a strategic diversification into semiconductors [56][57] - 美光 (Micron) and 三星 (Samsung) are resuming long-term contract negotiations for memory chips, reflecting a shift in market dynamics as prices stabilize [60] AI and Semiconductor Demand - The demand for AI-related products is driving significant growth in the semiconductor sector, with companies like 平头哥 (Pingtouge) reporting substantial GPU deliveries and revenue growth [61] - The overall semiconductor market is expected to benefit from the increasing integration of AI technologies across various industries [61]

Core Viewpoint - Huawei has officially launched the AI training inference acceleration card Atlas 350, powered by the new Ascend 950PR processor, marking the commercial phase of the Ascend 950 series [1][3][7]. Group 1: Product Launch and Features - The Atlas 350 is equipped with the Ascend 950PR processor and has been showcased at the Huawei China Partner Conference 2026, receiving significant attention from the industry [1][7]. - The Atlas 350's single-card computing power is 2.87 times that of NVIDIA's H20, with a FP4 precision computing power of 1.56P and a bandwidth of 1.4TB/s, while its power consumption is 600W, which is 1.5 times that of H20 [9][11]. - The Atlas 350 supports multiple precision formats, including FP4, FP16, and FP8, enabling it to handle larger models and lower latency inference, making it suitable for high-concurrency scenarios like short videos and e-commerce [11][12]. Group 2: Industry Partnerships and Applications - Seven core partners of Huawei, including Kunlun and Baode, have launched server products based on the Atlas 350, indicating a collaborative effort to enhance AI capabilities [3][5]. - The Ascend 950 series aims to support various AI applications across industries, with a focus on lightweight deployment and rapid implementation, addressing the complexities of intelligent deployment [12]. - Huawei has collaborated with 20 leading industry partners to release AI application solutions for 2026, covering key scenarios such as smart customer service and electronic medical records, with a significant market presence in integrated machines [12].

Group 1 - The core point of the article is that Chinese large models have surpassed American models in token usage, marking a significant milestone in the AI industry [1][2][13] - From February 9 to 15, 2023, the token call volume for Chinese models reached 41.2 trillion, surpassing the U.S. models at 29.4 trillion, and further increased to 51.6 trillion in the following week, a 127% rise [1] - The newly released MiniMax M2.5 achieved a token call volume of 45.5 trillion, becoming the monthly champion on OpenRouter [1][2] Group 2 - The rise of domestic computing power is breaking the monopoly of Nvidia, with significant investments in production capacity from local wafer manufacturers [3] - HW Ascend is accelerating its product launches, with the Ascend 950PR and 950DT expected in Q1 and Q4 of 2026, respectively, enhancing the capabilities of the Atlas 900 A3 SuperPoD [3] - The integration of domestic models, computing power, and China's electricity supply forms a competitive advantage that is difficult to replicate [3][4] Group 3 - The essence of AI is power consumption, which is fundamentally linked to chip computation and electricity supply [4] - China's leading position in power infrastructure and clean energy supports the growth of computing power, which in turn drives the iteration of large models [4] - The collaboration between HW Ascend and domestic manufacturers enhances the competitive edge of the domestic ecosystem [5] Group 4 - HW Ascend's public testing of the CodeArts AI development tool lowers the entry barrier for AI development, increasing participation in the ecosystem [7] - HW Ascend is actively defining global AI standards by joining the Linux Foundation's AAIF, positioning its chip architecture within global technology norms [7] - Nvidia's recent financial report showed strong revenue but resulted in a significant stock drop, attributed to market concerns over its growth sustainability and competition from emerging players [8][12] Group 5 - The "halo effect" in the AI industry is driven by strong demand for AI infrastructure and the rapid evolution of AI applications, impacting the software sector [10] - Key investment opportunities are identified in four areas: AIDC cloud services, domestic computing power, core segments of the global AI computing industry, and the "optical-electrical-material" triangle in AI infrastructure [10][12] - The "optical-electrical-material" triangle represents a high-demand segment, with increasing requirements for optical communication and power supply as AI computing needs grow [10][12] Group 6 - The overall trend indicates that the global AI industry landscape is being restructured, with China emerging as a significant player rather than merely a follower [13] - The era of domestic large models and computing power is just beginning, highlighting the importance of these developments in the global AI context [13]

Summary of Conference Call Notes Company and Industry Involved - The focus is on **Zhongtai Communication** and the **Huawei Ascend** series, particularly the **Ascend 950PR** chip. Core Points and Arguments - **Ascend 950PR Launch**: The Ascend 950 chip is expected to be launched soon, featuring a four-die package structure that includes two computing dies and two I/O dies, connected via CoWoS-L technology [1] - **Yield Improvement and Performance**: Recent improvements in yield rates and actual performance scores may exceed expectations [2][3] - **Surge in Computing Demand**: The demand for computing power is anticipated to surge due to rapid growth in TOKEN demand, with 2026 projected to be a significant year for domestic computing power [4] - **Domestic Internet Giants' Strategy**: Major domestic internet companies are expected to adopt a dual strategy of self-research and external procurement, with Huawei likely to capture a substantial market share due to its ecosystem [4] - **Investment Focus Areas**: Suggested companies to watch within Huawei's full industry chain include: - **Connectors**: Huafeng Technology, Luxshare Precision, Yihua Co., Ltd. - **Optical Modules**: Huagong Technology, Guangxun Technology - **Computing Power**: Huasheng Tiancai, Digital China, Fenghuo Communication - **Liquid Cooling**: Yingweike, Chuanrun Co., Ltd., Shenling Environment, Gaolan Co., Ltd. - **Switch Chips/Switches**: Shengke Communication, Ruijie Networks - **Power Supplies**: Taijia Co., Ltd., Jiewate [4] - **Investment Logic**: The successful launch of the Ascend 950, combined with the strong demonstration effect from leading internet companies, is expected to transition domestic computing power from "usable" to "highly usable" and "essential for large-scale applications" [4] Other Important but Possibly Overlooked Content - **High Demand Certainty**: The competition for computing power among major companies is likened to the historical competition for prime real estate, indicating a high level of demand certainty [5] - **Risk Factors**: Potential risks include product development not meeting expectations and slower-than-expected order placements from the internet sector [4]

Core Insights - The report highlights the strong performance of the communication resources sector, leading the A-share market's upward trend, with a focus on investment opportunities in telecommunications and technology sectors [5][8][22] - The semiconductor industry is experiencing significant growth, driven by increased capital expenditure from major cloud companies and rising demand for AI-related hardware [31][34][35] - The agricultural sector, particularly in pig farming, shows signs of price stabilization and potential upward trends due to supply constraints [21][20] Domestic Market Performance - The Shanghai Composite Index closed at 4,117.41, with a slight increase of 0.87%, while the Shenzhen Component Index rose by 1.36% to 14,291.57 [3] - The A-share market is characterized by a mixed performance across sectors, with telecommunications, coal, and precious metals showing strength, while sectors like tourism and gaming lagged [5][9][10] International Market Performance - The Dow Jones Industrial Average closed at 30,772.79, down by 0.67%, while the S&P 500 and Nasdaq also experienced slight declines [4] - The international market reflects a cautious sentiment, with various indices showing mixed results, indicating potential volatility [4] Industry Analysis - The report discusses the strong performance of the AI sector, with significant advancements in AI models and applications expected to disrupt traditional industries such as media and software [17][19] - The telecommunications industry is projected to benefit from increased capital expenditures and technological advancements, with a focus on 5G and cloud integration [22][26] - The semiconductor industry is on an upward trajectory, with substantial growth in sales and capital investments anticipated, particularly in AI infrastructure [31][33][34] Investment Recommendations - Investors are advised to focus on sectors such as telecommunications, AI, and semiconductors, which are expected to yield strong returns due to ongoing technological advancements and market demand [19][34][35] - The report suggests a balanced investment strategy, emphasizing the importance of monitoring macroeconomic indicators and policy changes that could impact market dynamics [5][11][12]

Summary of Key Points from the Conference Call Industry Overview - The document discusses the AI chip industry in Shenzhen, particularly focusing on Huawei's role as a leading player in the semiconductor sector following the release of the "Artificial Intelligence+" advanced manufacturing action plan on February 12 [1][2]. Core Insights and Arguments - The Shenzhen government has officially recognized AI chips as a breakthrough in the semiconductor industry, indicating a strategic focus on domestic alternatives, particularly targeting Huawei's Ascend series [1]. - The policy aligns closely with Huawei's technology roadmap, emphasizing the development of AI terminal SoC main control chips, automotive-grade intelligent driving chips below 14nm, and new architectures for integrated computing and storage, which correspond with Huawei's Ascend series and Kirin automotive chips [1]. - The document highlights that the policy is tailored to support Huawei's ecosystem, particularly in AI smartphones, AI glasses, and intelligent robots, indicating a strong governmental backing for Huawei's products [1]. Quantitative Backing - The Shenzhen Development and Reform Commission has quantitatively endorsed the Ascend 910 series, stating it meets international mainstream standards [2]. - A significant project in Guangming Science City has allocated 155 million yuan for intelligent computing, explicitly stating that the Ascend 910C will be the core architecture, rejecting imported alternatives, which shows the policy's implementation is moving into the budget execution phase [2]. - Huawei's roadmap indicates that by Q1 2026, the Ascend 950PR will achieve a computing power of 1 PFLOPS (FP8), representing a more than 2.5 times improvement over the 910C [2]. Industry Dynamics - The strategy of pushing for chip iteration through terminal applications is expected to unlock the commercial potential of the Ascend series [3]. - The Ascend chip is noted as the only domestic AI chip that achieves full coverage across "cloud-edge-end" scenarios with a self-developed instruction set [3]. - The essence of the current policy is to use Shenzhen as a testing ground to establish a domestic AI computing replacement system centered around Huawei [3]. - The industry benefits are anticipated to extend beyond individual chips to encompass the entire supply chain, including EDA tools, advanced packaging, and computing center construction [3]. Valuation Shift - The valuation logic for Huawei and its supply chain is transitioning from being event-driven to being fundamentally driven, indicating a more stable and predictable growth outlook [4].

Group 1 - The core point of the article is that after a significant drop in stock price, Cambricon issued a statement denying rumors about its revenue guidance and reaffirming its operational stability and R&D progress [1][3][4] - On February 3, Cambricon's stock price fell over 14%, reaching a low of 1063 yuan per share, which raised concerns about competition and revenue expectations [3][4] - Following the statement, Cambricon's stock experienced a slight recovery, with the decline narrowing to within 10% [4] Group 2 - Cambricon's 2025 revenue forecast is between 6 billion to 7 billion yuan, representing a growth of 410.87% to 496.02% year-on-year, with a projected net profit of 1.85 billion to 2.15 billion yuan [6] - The company attributes its expected revenue growth to the increasing demand for computing power in the AI industry and its competitive product offerings [6] - However, Cambricon's Q4 net profit is expected to be between 245 million to 545 million yuan, which is below market expectations [6] Group 3 - The competition in the domestic AI chip market is intensifying, with major internet companies ramping up their efforts [8] - Alibaba's self-developed AI chip "Zhenwu" has been launched, achieving significant deployment and surpassing Cambricon in shipment volumes [9] - Huawei and Baidu are also advancing their AI chip technologies, with Huawei planning to release several new chips by 2028 and Baidu launching new Kunlun chips for large-scale AI tasks [12][13]

Core Viewpoint - The article discusses the recent developments in the Ascend industry chain, particularly focusing on Huawei's potential orders and product launches related to AI computing cards in South Korea, indicating a significant market interest and potential growth in domestic alternatives for technology products [2][3][4]. Group 1: Huawei's Orders and Product Launches - There are rumors about Huawei receiving an order of 50 billion, but it has been confirmed that a previous order of around 10 billion was placed, with further cooperation expected, although the exact amount remains unclear [3]. - Regardless of the exact figures, it is evident that the orders for Ascend products are substantial, reflecting a positive outlook for domestic replacement chains in the upcoming year [4]. Group 2: Ascend Product Roadmap and Supply Chain - Huawei announced its product roadmap at the All-Connect Conference on September 18, detailing the launch of the 950PR in Q1 2026, along with other series products planned for 2026-2028 [5]. - The 950PR will focus on upgrading interconnect bandwidth, enhancing cluster and training capabilities, with significant value added in the inter-chip connections [5]. Group 3: Connector Component Analysis - The market value for the Ascend 8-card machine is estimated to be between 55,000 to 60,000, with potential for higher value in super nodes like CM384 [7]. - The discussion around "optical replacing copper" does not apply to the copper connections used in Ascend servers, as the high-speed backplane connection cables are essential for short distances, indicating a coexistence of copper and optical connections in the future [11]. Group 4: Strategic Partnerships in Gas Turbines - Jerry's announcement of a strategic partnership with Kawasaki Heavy Industries aims to provide efficient, low-carbon, and intelligent power solutions for global AI data centers and industrial manufacturing [13]. - This partnership is expected to alleviate concerns regarding Jerry's resource availability in gas turbines, especially in a market with high demand and stringent policy requirements [13].