智通财经2月26日讯（编辑 胡家荣）港股三大指数集体下挫。截至收盘，恒生指数跌1.44%，报26381.02点；科技指数跌2.87%，报5109.32点；国企指数跌 2.44%，报8814.29点。 中信建投研报指出，人工智能数据中心(AIDC)建设已进入爆发式增长通道。测算显示，2025–2028年美国AI发展催生的电力容量需求年均复合增长率 (CAGR)约55%，未来三年累计需求超150吉瓦。 同时北美地区电力缺口持续扩大，"自建电源"成为科技巨头刚需。燃气轮机凭借快速启停、高功率适配、成本优势及运行可靠性，正成为AIDC主电源首选 方案，带动发电设备产业链价值重估。 机构看多加密市场 金涌投资涨超29% 今日市场 今日盘面呈现"新质生产力主线强势突围，传统周期板块承压回调"的鲜明特征，其中电力设备、加密货币、光通信等代表技术迭代与全球产业趋势的板块集 体走强；而光伏、建材水泥、锂电池、黄金珠宝等受制于供需结构和股东行为影响的板块则明显回调。 AIDC引爆全球电力基建浪潮 东方电气涨超15% 截至收盘，东方电气(01072.HK)涨15.52%、哈尔滨电气(01133.HK)涨7.15%、金风科技(022 ...

智通财经APP获悉，鸿腾精密(06088)再涨超10%，截至发稿，涨8.82%，报6.54港元，成交额3.13亿港 元。 财通证券发布研报称，鸿腾精密实现从"精密零件"到"系统级互联解决方案"的升级。AI侧，公司受益于 全球光模块市场高增长，公司布局224G高速互联、800G/1.6T光模块及CPO解决方案，产品已导入北美 核心客户。EV侧，随800V平台普及与自动驾驶等级提升，连接器单车价值量跃升。公司通过并购整合 Voltaira与Auto-Kabel，构建"传感-连接-电源管理-高压"垂直整合布局，并成功开拓了沙特等新兴蓝海市 场。 国投证券国际此前发布研报称，市场长期将鸿腾精密视为传统的"消费电子组装/线缆厂"，随着公司在 AI服务器互连及电动车业务的实质性突破，其营收结构正向高毛利、高壁垒的"连接器与核心组件厂"转 型。公司在光模块、CPO技术、车用高压连接器等高毛利新产品的客户认证及量产进度，将是优化产品 结构、推动整体毛利率持续改善的关键。2026年高毛利相关业务预计迎来业绩放量。 ...

国投证券国际此前发布研报称，市场长期将鸿腾精密视为传统的"消费电子组装/线缆厂"，随着公司在 AI服务器互连及电动车业务的实质性突破，其营收结构正向高毛利、高壁垒的"连接器与核心组件厂"转 型。公司在光模块、CPO技术、车用高压连接器等高毛利新产品的客户认证及量产进度，将是优化产品 结构、推动整体毛利率持续改善的关键。2026年高毛利相关业务预计迎来业绩放量。 财通证券（601108）发布研报称，鸿腾精密实现从"精密零件"到"系统级互联解决方案"的升级。AI侧， 公司受益于全球光模块市场高增长，公司布局224G高速互联、800G/1.6T光模块及CPO解决方案，产品 已导入北美核心客户。EV侧，随800V平台普及与自动驾驶等级提升，连接器单车价值量跃升。公司通 过并购整合Voltaira与Auto-Kabel，构建"传感-连接-电源管理-高压"垂直整合布局，并成功开拓了沙特等 新兴蓝海市场。 鸿腾精密(06088)再涨超10%，截至发稿，涨8.82%，报6.54港元，成交额3.13亿港元。 ...

在产业协同实践中，Kevin Pan深刻认识到AI训练集群的数据吞吐量挑战——单集群日均可达10PB，直 接推动800G/1.6T光模块商业化，并将链路质量要求提升至极限：误码率必须低于10 。他推动建立了覆 盖硅光器件供应商、光模块厂商及DSP厂商的协同创新机制，围绕"眼图质量—误码表现"的稳定相关 性，聚焦TDECQ与系统级误码率的闭环优化取得重要突破。针对800G模块，通过TDECQ算法优化和大 量模块与码型实验，使多数厂商能够将TDECQ控制在2以下，显著提升链路质量的可迁移性与稳定性。 同时通过DCA测试精度提升，完善算法适配与跨仪器一致性，实现理论拟合度与可重复性的大幅提 升。在1.6T模块方面，团队联合产业链上下游在编码、工艺与算法侧协同优化，最终通过与DSP厂商合 作，降低底噪、提升灵敏度，建立稳定相关性，为1.6T量产奠定基础。 技术引领：突破算力时代互联瓶颈 随着生成式AI驱动算力需求爆发式增长，大型语言模型迭代对训练吞吐与推理时延提出极致要求，数 据中心由此迈入"超高带宽、超低时延、高密度互联"的新阶段。Kevin Pan敏锐把握这一趋势，指出硅 光技术将成为突破互联瓶颈的关键。在他的推 ...

Group 1 - The article emphasizes the importance of timely and professional information interpretation in investment decision-making [1] - HBM+ chip packaging company has started mass production of HBM2e and is advancing the tape-out of HBM3/3e, with applications in CPU, GPU, AI, and automotive high-performance chips [1] - PCB products are utilized in humanoid robots and AI servers, with the company achieving delivery and certification of 800G/1.6T optical modules [1]

Economic Growth Target - The GDP growth target for 2026 is set at around 5%, which is considered a significant figure given the current economic challenges [3][30] - The target is primarily driven by the need to create jobs, as failing to meet growth expectations could lead to employment issues [5][30] - Structural constraints such as aging population and diminishing returns on capital are evident, but advancements in AI may help boost potential economic growth [5][30] Policy Measures - A coordinated approach between fiscal and monetary policies is essential, with an emphasis on active fiscal measures and reasonable debt arrangements [9] - Monetary policy is expected to be moderately accommodative, with potential interest rate cuts and liquidity support for key sectors [9] - Stability in the RMB exchange rate is crucial to avoid significant fluctuations [9] Domestic Demand Challenges - There is a notable lack of consumer confidence, leading to reduced spending by households and investment hesitance from businesses [11][30] - The real estate market is facing weakened demand, with changing demographics affecting purchasing power and urbanization slowing down [11][30] - Local government debt poses a significant constraint on growth, necessitating careful management of fiscal resources [11][30] Consumer Spending Dynamics - The marginal propensity to consume among Chinese residents is relatively low compared to international standards, with significant disparities across income groups [14][16] - Addressing income distribution and improving social security systems are critical to enhancing consumer spending [16][17] - Reducing mortgage interest rates could significantly release consumer spending potential [17] Technological Focus - The global competition in technology underscores the importance of self-sufficiency and domestic substitution [20][30] - There are substantial opportunities in AI infrastructure, including demand for servers, semiconductors, and optical technologies [21][23] - The advancement of new technologies is expected to enhance total factor productivity, which is vital for economic growth [24][30] Investment Opportunities - Key areas for investment include expanding domestic demand, technological self-reliance, and sectors experiencing a rebound in supply-demand dynamics [28][30] - The economic growth trajectory is anticipated to show a pattern of lower growth in the early quarters, followed by gradual recovery [28][30]

Core Viewpoint - The report from Caitong Securities initiates coverage on Hongteng Precision (06088) with a "Buy" rating, highlighting the company's leading position in the connector industry and its growth potential in the "AI computing power + electric vehicles" sectors [1] Group 1: Core Competence Leap - The company has upgraded from "precision parts" to "system-level interconnection solutions," leveraging 40 years of connector technology accumulation and strategic acquisitions to cover a full product line in optics, electricity, and vehicles [1] - With a global R&D and manufacturing network, the company has transitioned from a single component supplier to a system solution provider, offering customized capabilities from concept to mass production [1] Group 2: Growth Momentum - AI and automotive businesses have become the dual engines of growth, with the implementation of a "3+3" transformation strategy yielding significant results [2] - By the first half of 2025, revenue from cloud network facilities and automotive sectors is expected to exceed traditional consumer electronics for the first time, accounting for 35.3% of total revenue, driven by the rapid growth of AI and EV businesses at rates of 35.7% and 102.3%, respectively, compared to the overall company revenue growth of 11.53% [2] Group 3: Technological Upgrade - The company is seizing opportunities in the AI computing infrastructure and automotive neural network markets [3] - On the AI side, the company benefits from the high growth of the global optical module market (CAGR > 30%) and has developed 224G high-speed interconnects, 800G/1.6T optical modules, and CPO solutions, with products already introduced to key customers in North America [3] - On the EV side, as the 800V platform becomes more widespread and autonomous driving levels improve, the value of connectors per vehicle is increasing. The company has integrated Voltaira and Auto-Kabel through acquisitions to establish a vertically integrated layout for "sensing-connection-power management high voltage," successfully entering emerging markets like Saudi Arabia [3]

Group 1: Core Trends in Technology Industry - Goldman Sachs identifies ten key trends in the technology sector for 2026, focusing on AI servers, optical communication, foldable iPhones, semiconductors, smart driving, and satellite communication, revealing structural investment opportunities driven by technological innovation and supply chain changes [1] - The AI server market is expected to see explosive growth, with shipments projected to increase from 19,000 units in 2025 to 50,000 units in 2026, driven by the rising penetration of ASIC chips, which is expected to reach 40% by 2026 [2] - The optical communication sector will benefit from the expansion of AI infrastructure, with demand for optical transceivers expected to surge as data centers upgrade from 400G to 800G/1.6T [3] Group 2: Key Developments in Specific Technologies - Liquid cooling technology is anticipated to see a significant increase in penetration, particularly in the ASIC AI server domain, as supply chains adapt to higher thermal challenges from increased computing power [4] - ODM manufacturers with strong commitments or capacity plans in the U.S. are expected to outperform the market, with companies like Hon Hai, Wistron, and Wiwynn favored due to their robust R&D capabilities and vertical integration [5] - The PC market faces significant challenges in 2026, with only global leaders like Lenovo expected to maintain resilience due to stronger supply chain bargaining power and exposure to high-end products [6] Group 3: Consumer Electronics and Semiconductor Insights - Apple's upcoming foldable iPhone is projected to ship between 11 million and 35 million units in 2026, becoming a strong catalyst in the smartphone market, with high-end foldable models driving growth for related component suppliers [7] - The PCB market remains solid, particularly for high-end CCL and PCB suppliers, benefiting from the growth in AI server shipments and increased ASIC penetration, with average selling prices expected to rise by 20-30% annually in 2026 and 2027 [8] - The Chinese semiconductor industry is expected to continue its growth, driven by local leaders' expansion plans and the rise of domestic GPU suppliers, with AI technology innovation and new demands from edge devices being key growth drivers [9] Group 4: Emerging Technologies and Future Outlook - The smart driving trend is expected to deepen in 2026, with the proliferation of urban navigation-assisted driving and Robotaxi services driving growth for chipsets, software, and sensor suppliers [10] - The low Earth orbit satellite industry is entering an acceleration phase, with increased rocket payload capacities and reduced launch costs expected to speed up satellite launches, alongside upgrades in satellite specifications [11]

Core Insights - The demand for optical modules is surging due to the diversification of AI chips, with a shift towards ASIC architectures expected to increase their market share to 45% by 2025-2027, requiring 3-5 optical modules per chip compared to 2-3 for GPUs [1] - The market is transitioning to 800G and 1.6T optical modules, with 800G expected to dominate by 2025 and 1.6T already in deployment, providing the company with a significant first-mover advantage and a cycle of "technological leadership leading to market share increase and profit optimization" [1] - The adoption of silicon photonics technology is replacing EML, offering higher gross margins and mitigating the 10%-20% price decline pressure in the industry, while export control impacts are gradually easing [1]

Core Viewpoint - The U.S. government has adjusted its export control policy on advanced AI chips to China, allowing NVIDIA to deliver its H200 chip to approved commercial clients, but with significant restrictions [2][3] Group 1: Policy Changes - The U.S. government, through President Trump, announced that NVIDIA can sell its H200 chip to vetted Chinese clients, while excluding the more advanced Blackwell architecture and next-generation Rubin chip from this permission [3] - A requirement has been established that 25% of the sales revenue from these chips must be paid to the U.S. government [3] Group 2: Technical Specifications - The H200 chip is based on the complete Hopper architecture, featuring 141GB of HBM3E high-bandwidth memory and a memory bandwidth of 4.8TB/s, with floating-point performance (FP64) and AI training (FP8 Tensor Core) capabilities at the leading level in the international market [3] - In contrast, the previously designed "China-specific" H20 chip has significantly reduced memory bandwidth, interconnect speed, and core computing performance, resulting in much lower overall AI computing power compared to the H200 [3] Group 3: Implications for AI Development - The limited approval of the H200 chip is seen as a positive signal for domestic AI development, alleviating pressure on leading institutions during large model training and accelerating model iteration and application [4] - This move is interpreted as a sign of a shift towards "managed competition" in the U.S.-China tech rivalry, reducing short-term risks of a hard decoupling in the global AI supply chain [4] Group 4: Long-term Strategic Considerations - The stringent conditions attached to the H200 chip approval may reinforce China's commitment to developing its own computing power system, as the 25% sales share acts as a long-term "technology tax" that could erode profits and increase costs [4] - The exclusion of the most advanced architectures like Blackwell indicates a deliberate effort to maintain at least a generational technology gap, suggesting that reliance on external licenses for computing power is not a sustainable strategy [4] Group 5: Investment Recommendations - Companies in the AI industry chain and domestic computing power industry chain are recommended for attention, including AIDC-related stocks such as Runjian Co., Dataport, and Runze Technology [5] - Other relevant companies include those involved in liquid cooling, optical modules, optical devices, optical chips, and optical engines [5]