硬科技企业主导上市潮。 港交所锣声频响，国际资本重返香江，硬科技公司成为上市主角。2026年1月，香港IPO市场延续了2025年的活跃态势。截至1月26日，已有12家 公司在港交所上市，总募资额达347.47亿港元，同比激增480.87%。目前，港交所排队企业超过300家，为市场持续活跃提供了充足后备资源。 2026年初上市的公司主要集中在半导体、AI和生物医药等硬科技领域，包括国产GPU企业壁仞科技、AI大模型公司智谱和MiniMax。这些企业 大多通过港交所的18C特专科技公司上市通道实现上市。 国际资金参与度显著提升。国际长线资金对港股IPO项目的参与率从2024年初的约10%-15%跃升至目前的85%-90%。中东主权财富基金、欧洲家 族办公室等机构投资者频频现身基石投资。多家市场机构预测，2026年港股IPO募资规模有望超过3000亿港元。市场将呈现"两头大、中间分 化"的特征，大型项目和行业龙头更易获得资金支持。 ...

在全球流动性环境边际改善、科技周期持续推进的背景下，外资机构对2026年的行业主题判断 逐步趋于一致。 在多家外资公募机构看来，围绕科技创新展开的结构性机会，仍将是中期配置的重要主线之一， 其核心不在于短期题材轮动，而在于产业演进方向与企业所处位置的变化。 贝莱德基金权益、量化及多资产首席投资官王晓京表示，具体到行业主题方面，结构性行情在 2025年已充分展现。在美联储降息带来全球流动性溢出且地缘矛盾缓和的背景下，2026年多项主 题行情具备延续基础。 人工智能作为重塑生产力与生产关系的科技革命，目前仍处于进行时。贝莱德基金看好AI本身的 结构性行情在2026年持续，并逐步从硬件基础设施向更多应用AI技术的传统与创新行业扩散。对 于AI等科技成长板块，投资者需密切关注估值水平与公司基本面现金流改善的匹配度，甄别"真 增长"。尽管2025年部分标的估值已偏高，但由于AI对各行各业的改造远未完成，2026年内恐怕 难以完全证实或证伪支撑其高估值的远期现金流预期。因此，即使存在波动风险，AI主题行情预 计仍将活跃。 从更长期视角看，AI并非单一市场的独立叙事。在宏利基金看来， AI作为全球范围内的重要产 业发展趋势 ...

在全球流动性环境边际改善、科技周期持续推进的背景下，外资机构对2026年的行业主题判断逐步趋于一 致。 多家外资公募认为，人工智能仍是最具确定性的中长期主线之一。尽管部分细分领域在2025年已出现估值抬 升，但从生产力重塑和产业渗透进程看，AI行情尚未走完，结构性机会具备跨年度延展的基础，主题活跃度 预计仍将维持。 在此基础上，围绕科技创新主线，外资机构普遍强调，应结合产业链自主可控的发展方向，优先关注在关键环 节具备引领力和整合能力的核心企业，其在产业体系中的位置，正成为判断长期投资价值的重要依据。 人工智能主题行情仍有延展空间 在行业主题层面，外资机构的判断整体偏向"趋势未完"。在宏观环境与流动性条件发生边际变化的背景下，部 分结构性机会被认为具备跨年度延续的基础。 贝莱德基金权益、量化及多资产首席投资官王晓京表示，具体到行业主题方面，结构性行情在2025年已充分展 现。在美联储降息带来全球流动性溢出且地缘矛盾缓和的背景下，2026年多项主题行情具备延续基础。 人工智能作为重塑生产力与生产关系的科技革命，目前仍处于进行时。贝莱德基金看好AI本身的结构性行情 在2026年持续，并逐步从硬件基础设施向更多应用 ...

Core Insights - Global AI capital expenditure is growing at an unprecedented rate, with the AI industry entering a second phase characterized by diverse applications and infrastructure development [1] - By 2026, AI capital expenditure is projected to reach $600 billion, which is still below the historical peaks of previous productivity revolutions [1] - The infrastructure scale for AI is expected to surge to $3 trillion by 2030, with a compound annual growth rate (CAGR) of 31% from 2024 to 2030, marking it as the fastest infrastructure cycle in history [1] Group 1: Investment Focus Areas - North American computing infrastructure is evolving from a focus on GPU to a system that integrates computing, memory, storage, and connectivity, with an emphasis on storage as a core capability [1] - Domestic computing capabilities are catching up, with a focus on the penetration of domestic GPUs/ASICs and the supporting industries such as servers and optical modules [2] - The expansion of the AI application ecosystem is anticipated, particularly with multi-agent systems that can autonomously execute tasks and enhance decision-making density [2] Group 2: Market Dynamics - The penetration rate of domestic AI computing is accelerating, with increasing acceptance of domestic GPUs in sectors like government, finance, and manufacturing [2] - China's AI computing network is entering a scaling phase, with local AI computing centers being established across provinces, providing advantages such as lower latency and localized data [2] - The AI ecosystem in China and the U.S. is developing in parallel, with the U.S. leading in large model research while China shows stronger competitiveness in industry model applications [2]

Core Insights - The rise of domestic AI computing chips represents not only a technological competition but also a significant strategic move towards industrial autonomy and security [1] - The Chinese AI computing chip industry is currently in a critical phase of catching up and breakthrough, transitioning from "engineering usable" to "scalable replaceable" [2] Industry Development - Domestic GPU manufacturers have made significant achievements in chip design, architecture innovation, and performance optimization, with some products meeting mainstream application needs and even surpassing in specific areas [2] - National policies and top-level designs have been pivotal in driving the growth of the domestic GPU market, with ongoing infrastructure development and massive demand for AI applications creating a strong resonance with capital [2][3] Policy Support - The government has provided comprehensive support for domestic chip companies through investment funds, innovative listing standards, and preferential procurement in key industries [3] - The introduction of clear domestic goals has created substantial market access opportunities for domestic GPUs in critical sectors [3] Software Ecosystem - The software ecosystem is crucial for the AI chip industry, serving as a competitive moat and determining whether domestic GPUs can become mainstream [3][4] - A full-stack, self-controllable ecosystem from hardware to applications is essential, with the flexibility of GPU ecosystems being more advantageous than the long-established CPU ecosystems [3][4] Market Dynamics - Domestic GPU companies are challenging Nvidia's pricing through a combination of usable performance and significant price advantages, gradually eroding existing market structures [5] - The automotive sector is identified as a key area for scaling domestic AI computing chips, with increasing demands for real-time processing and high performance [5] Future Outlook - The domestic AI computing chip industry is expected to transition from "usable" to "well usable" over the next five years, focusing on practical application performance, stability, usability, and overall cost-effectiveness [6] - Emerging trends such as embodied intelligence are anticipated to drive new growth in AI computing chip demand, necessitating advancements in performance and latency [6] Strategic Positioning - The domestic AI computing chip sector is on a path of autonomous innovation and steady growth amidst open competition and security demands, with the potential to become globally competitive in key application areas [7]

Group 1 - The semiconductor chip stocks experienced a rally, with the Shanghai Stock Exchange Sci-Tech Innovation Board chip index rising by 0.70% as of 09:54 on January 20, 2026, driven by significant gains in individual stocks such as Baiwei Storage (up 7.42%) and Longxin Zhongke (up 5.62%) [1] - The DRAM market continues to face supply tightness, with the price of mainstream DDR4 models increasing by nearly 10% week-over-week, primarily due to suppliers and traders adopting a strategy of withholding sales and stockpiling inventory [1] - Micron reported that AI demand now accounts for 50%-60% of the DRAM market, serving as the main driver for the sustained increase in storage chip prices [1] - The enterprise SSD pricing is also under upward pressure, with some manufacturers like SanDisk requiring customers to make full prepayments to secure NAND allocations for the coming years [1] Group 2 - High-performance computing (HPC) chips have become the core engine driving TSMC's advanced process business growth, with HPC revenue accounting for 55% of TSMC's total revenue, significantly surpassing the smartphone segment at 32% [1] - The HPC business continues to show strong growth resilience compared to the seasonal fluctuations of the smartphone business, suggesting a robust investment opportunity in the HPC chip supply chain, particularly in domestic GPU and advanced process foundry sectors [1] - As of December 31, 2025, the top ten weighted stocks in the Sci-Tech Innovation Board chip index include SMIC, Haiguang Information, and Cambrian, collectively accounting for 57.76% of the index [2] - The Sci-Tech Chip ETF (588200) tracks the Sci-Tech Innovation Board chip index, providing a convenient tool for investors to gain exposure to the domestic chip sector [2] - Investors without stock accounts can access the domestic chip investment opportunities through the Sci-Tech Chip ETF linked fund (017470) [2]

1. Report Industry Investment Rating No relevant information provided. 2. Core View of the Report - On Thursday, the main indices of the two markets oscillated and adjusted to repair technical indicators, with the semiconductor equipment sector leading the gains. International investors are accelerating the allocation of Chinese assets, and multiple foreign - funded institutions are optimistic about the performance of Chinese assets in 2026. The Chinese stock market has a high probability of rising in 2026, and the growth - type indices are expected to end the adjustment and resume the upward trend on Friday or early next week. Long - term index futures positions should be held, and investors can choose to buy out - of - the - money long - term call options on the CSI 1000 Index [1][2][3] 3. Summary by Relevant Catalogs 3.1 Market Review - On Thursday, the trading volume of the two markets was 2.90 trillion yuan, showing a rapid contraction. The CSI 500 Index closed at 8223 points, down 4 points or - 0.05%; the CSI 1000 Index closed at 8240 points, down 16 points or - 0.20%; the SSE 50 Index closed at 3105 points, down 6 points or - 0.21%; the SHSZ 300 Index closed at 4751 points, up 9 points or 0.20%. The funds in stock index futures for the CSI 1000, SHSZ 300, CSI 500, and SSE 50 indices had net outflows of 7.1 billion, 4.4 billion, 3.6 billion, and 1.3 billion yuan respectively [1] - Among industry and theme ETFs, the top gainers were semiconductor equipment ETFs, while the top losers were satellite industry ETFs. Among the sector indices of the two markets, the glass - fiber, rubber, and other sectors led the gains, and the industrial Internet, aerospace equipment, and other sectors led the losses [1] 3.2 Important Information - The central bank lowered the interest rates of various structural monetary policy tools by 0.25 percentage points and will lower the minimum down - payment ratio for commercial housing mortgages to 30%. There is still room for reserve requirement ratio and interest rate cuts this year [1] - China's "demand shortage" is mainly "consumer demand shortage", with the consumption - to - GDP ratio about 20 percentage points lower than the global average, and this gap needs to be corrected [1] - More than 10 billion yuan of funds continuously entered the market through ETFs. From January 9th to 13th, the net subscription amount of equity ETFs exceeded 12 billion yuan for three consecutive trading days, with a total net subscription amount exceeding 47 billion yuan. On January 14th, the trading volume of many broad - based ETFs increased significantly, and many newly issued funds announced the early end of fundraising [1] - China shows strong R & D potential and a clear technology iteration path. The relaxation of the H200 export policy by the US indicates that it can no longer block China's technological take - off [1] - Nvidia's new AI inference context memory storage (ICMS) architecture is expected to significantly exacerbate the global shortage of NAND flash memory, bringing additional demand equivalent to 2.8% and 9.3% of the global total NAND demand in 2026 and 2027 respectively [1] - The global explosion in AI chip demand is constrained by TSMC's production capacity, with a supply - demand gap of three times for its most advanced processes. Although TSMC is adjusting production lines and expanding production globally, the shortage cannot be alleviated in the short term, and the advanced packaging link has become a key bottleneck [2] - Deutsche Bank believes that the cost gap between space deployment and ground construction is shrinking rapidly, and in the next decade, the cost of building a space data center will approach that of ground construction [2] - OpenAI signed a three - year agreement with Cerebras, promising to purchase up to 750 megawatts of computing power, all using Cerebras' wafer - scale chips, with a transaction value exceeding $10 billion [2] - Trump launched fiscal, monetary, and credit stimulus, which may lead to debt out - of - control, financial risk accumulation, and a future debt crisis and market crash [2] - Wall Street financial institutions are entering the prediction market, and this emerging market has evolved into a sports - contract - based betting platform [2] - Citi's research report indicates that the commodity market is at a turning point. Crude oil is driven by geopolitics in the short term with a target of $70, but faces long - term oversupply pressure. Precious metals are bullish, with a target of $100 for silver and $5000 for gold. Among industrial metals, the target for aluminum is $3400 - 3500, and the copper price is expected to reach $14000, but January may be the annual high [2] 3.3 Market Logic - International investors are accelerating the allocation of Chinese assets. Many foreign - funded institutions are optimistic about the performance of Chinese assets in 2026. The continuous improvement of corporate profits, continuous technological innovation breakthroughs, and increasing valuation attractiveness provide a solid foundation for the continuous rise of Chinese assets [2] - In 2025, the stock market had a net inflow of 2.26 trillion yuan. In 2026, insurance, wealth management, and pensions are expected to be the three major sources of incremental funds, with institutional incremental funds in the stock market reaching 3.1 trillion yuan, and the scale of public fixed - income + products at least doubling [2] - More international funds are turning their attention to the AI track outside the US. China's technology sector, with its valuation advantages, complete industrial ecosystem, and large - scale manufacturing capabilities, is becoming a new destination for global funds in the AI field [2][3] 3.4 Market Outlook - The trading - type investors are actively increasing their positions in Chinese assets, and the allocation - type investors are optimizing the weight of Chinese assets in the global portfolio. The application for satellite frequency and orbit resources has risen to the national strategic level, and the upward trend of the Chinese market is expected to continue in 2026 [3] - The risk of a significant rise in the Chinese stock market in 2026 is much higher than that of a significant decline. The acceleration of AI applications, anti - involution, and the re - allocation of domestic liquidity from deposits to the stock market are positive factors [3] - The return of the US to the Monroe Doctrine will accelerate the flow of global funds to the Chinese capital market. The Fed's interest rate cut and balance - sheet expansion, along with the appreciation of the RMB, will lead to the return of a large amount of US dollars held by foreign trade enterprises overseas, and funds will flow from enterprise accounts to resident accounts and then to securities accounts [3] - China's application for 200,000 low - orbit satellites has shocked the market, indicating that the Sino - US space infrastructure competition has become a key area of technological competition. The adjustment of growth - type indices is expected to end on Friday or early next week, and the upward trend will resume [3] 3.5 Trading Strategy - For stock index futures directional trading, the policy hopes for a bull market but also a slow one. With a large amount of off - market funds still flowing in, the upward trend driven by market funds remains unchanged. Long - term stock index futures positions should be held [3] - For stock index option trading, investors can choose to buy out - of - the - money long - term call options on the CSI 1000 Index [3]

Group 1: AI Market Outlook - The probability of an AI bubble in China is significantly lower than in the US, with no clear signs of an AI bubble emerging in the short term [2][3] - Chinese leading AI firms rely on cash flow from parent companies for R&D, making funding sources more sustainable compared to the US [2] - Capital expenditure by major Chinese internet companies is pragmatic and cautious, with a projected capital expenditure of approximately 400 billion RMB in 2025, only one-tenth of that of their US counterparts, yet achieving similar model capabilities [2][3] Group 2: Semiconductor Industry - The semiconductor market is expected to reach over $700 billion by 2025, with projections of $1 trillion by 2026, driven by AI demand [5][6] - Semiconductor equipment investments are anticipated to grow by 10% in 2026, benefiting from advanced process production demands [6] - Domestic semiconductor companies are increasingly listing in Hong Kong, which may enhance their valuation and attract overseas talent [6] Group 3: Humanoid Robotics Sector - Global shipments of humanoid robots are projected to reach 30,000 units by 2026, with potential market size reaching $1.4 to $1.7 trillion by 2050 [7] - The industry is still in its early stages, facing challenges such as the lack of large-scale datasets for training and the absence of specialized AI models for humanoid robots [7][8] - The upstream supply chain, including components like screws, sensors, and core chips, is expected to benefit first from the growth in humanoid robotics, while midstream manufacturers face cash flow pressures [7][8]

围绕H200芯片的这场博弈，验证了自主创新才是打破封锁、赢得战略主动权的唯一正解。近日， DeepSeek和中国高校团队再发技术论文，"提出了一种绕过GPU内存限制的新模型训练方法"，或许将为 AI发展开辟一条更自主和可持续的全新路径。时至今日，随着技术的厚积薄发，中国不再满足于在别 人的赛道上跟跑，而是致力于开辟新的技术路线与产业生态。中国科技的崛起也不仅仅是为了自身的突 围，更是为了打破垄断，为世界提供差异化发展道路，帮助更多全球南方国家跨越数字鸿沟，共享科技 文明成果。 在当前全球科技竞争与合作交织的复杂背景下，大洋彼岸传来的一则消息颇具深意。当地时间1月13 日，美国政府正式批准英伟达H200人工智能（AI）芯片有条件销往中国，虽然在安全审查、销量规模 等方面施加了严格限定，但这依然被视为中美科技领域互动的微妙信号。有分析认为，华盛顿正在"遏 制中国AI能力"与"避免倒逼中国技术跃迁"之间寻求战略平衡。而这项"权衡利弊"后的政策调整，恰恰 是中国科技自立自强成果的生动写照。 英伟达H200芯片的"有条件放行"，绝非美方突如其来的"善意"。这项决策经过了精致的计算：允许输华 的H200芯片，在性能上落后 ...

与此同时，我们也看到，之前美国更低算力的芯片都被禁止对华出口，美方这次有条件解禁H200确实 是一项重大调整。这是双方在激烈的竞争博弈中努力管控分歧、维持经贸联系的现实映射，也反映出， 即便在被华盛顿视为"战略高地"甚至"必争之地"的算力与AI领域，违背经济规律的彻底"脱钩断链"也注 定行不通。产业链供应链的深度互嵌，是"合作才是最大安全"的生动注脚。 事实反复证明，人为构筑科技壁垒，最终结果是反噬自身的创新生态，让本国企业在封闭中失去活力。 美国过去针对中国的芯片"断供"行为，并未扼杀中国的高科技发展，而是激发了新型举国体制优势下的 强力攻关。从华为昇腾芯片的迭代，到国产GPU在特定场景的应用突破，再到全产业链的补链强链，中 国显示出强大的研发潜力与清晰的技术迭代路径。放宽H200出口政策，是美方已从行动上承认： 靠"堵"和"封"都已经不足以阻挡中国科技的腾飞。 围绕H200芯片的这场博弈，验证了自主创新才是打破封锁、赢得战略主动权的唯一正解。近日， DeepSeek和中国高校团队再发技术论文，"提出了一种绕过GPU内存限制的新模型训练方法"，或许将为 AI发展开辟一条更自主和可持续的全新路径。时至今日，随 ...