Market Performance - Global markets experienced a slight increase last week, with MSCI Global up by 1.0%, MSCI Developed Markets up by 1.0%, and MSCI Emerging Markets up by 1.1% [8][9] - Among developed markets, the Nasdaq Index showed the strongest performance with a rise of 2.2%, while the Australian S&P 200 was the weakest, declining by 1.0% [8][9] - In the emerging markets, the ChiNext Index performed best with a gain of 2.3%, while the Shanghai Composite Index fell by 1.3% [8][9] Trading Sentiment - Trading volume in European and American stock markets significantly increased, while the short-selling ratio in Hong Kong stocks dropped to a historical low of 11.8% [17][21] - The Hang Seng Index's trading volume decreased to 177 billion shares with a total turnover of 918 billion USD, while the S&P 500's trading volume increased to 62 billion shares with a turnover of 82,150 billion USD [17] - Investor sentiment in Hong Kong improved, reaching historical highs, while sentiment in the US declined but remained elevated [17] Earnings Expectations - Earnings expectations for Hong Kong stocks were revised upward, with the Hang Seng Index's 2025 EPS forecast increased from 2066 to 2068 [59][60] - In contrast, the S&P 500's earnings expectations remained flat at 269, while the Eurozone STOXX50's earnings expectations also held steady at 337 [59][60] - Sector-wise, the energy sector in Hong Kong saw the largest upward revision in earnings expectations, while the consumer staples sector experienced the most significant downward adjustment [59][60] Economic Expectations - Economic expectations in Europe and the US improved, with the Citigroup Economic Surprise Index for the US rising due to the Fed's rate cut and tech giants accelerating AI infrastructure investments [8][59] - Conversely, China's Economic Surprise Index declined, impacted by US-China trade negotiations and weaker-than-expected economic data for August [8] Fund Flows - Following the Fed's rate cut, market speculation regarding future rate cuts intensified, with futures markets indicating an expectation of 1.8 rate cuts by the end of the year [46][52] - Global liquidity trends showed significant inflows into India, Europe, Hong Kong, and South Korea, with Hong Kong experiencing a net inflow of 292 billion HKD last week [53][58] - The inflow included 145 billion HKD from stable foreign capital, while flexible foreign capital saw an outflow of 130 billion HKD [53]

制造成长周报（第 29 期） 优于大市 Figure 融资超 10 亿美元，微软 300 亿美元投资英国 AI 基建 重点事件及点评 事件 1-人形机器人：Figure 完成超 10 亿美元融资：据澎湃新闻消息，2025 年 9 月 16 日，AI 机器人初创公司 Figure 宣布完成 10 亿美元承诺资金 的 C 轮融资，投后估值达 390 亿美元。 证券研究报告 | 2025年09月23日 行业动态&政府新闻&公司动态 行业动态：人形机器人-9 月 20 日，华为官宣昇腾 AI、具身智能新动向；9 月 18 日，汉威科技动捕服开启人机交互新篇章；9 月 17 日，智元灵犀 X2 完 成韦伯斯特空翻；9 月 17 日智元机器人 Genie Trailblazer 计划全球招募； 9 月 17 日，舞肌科技发布第一代 Wuji Hand 灵巧手；9 月 16 日，Figure 完 成超 10 亿美元 C 轮融资；9 月 16 日，日本商社兼松最早将于 2027 年在全球 首次量产球形齿轮；9 月 15 日，宇树发布开源世界模型。3C-9 月 18 日，Meta 推出旗下首款内置显示屏的智能眼镜。AI 基建- ...

Macroeconomic Insights - The number of financing platforms in China decreased by over 60% compared to March 2023, and the scale of financial debt dropped by over 50% as of the end of June [2] - The current monetary policy stance in China is supportive, with a focus on moderately easing monetary policy to reduce overall financing costs [2] - As of the end of August, various long-term funds held approximately 21.4 trillion yuan in A-share market value, a 32% increase compared to the end of the 13th Five-Year Plan [2] - The balance of inclusive loans for small and micro enterprises reached 36 trillion yuan, which is 2.3 times that of the end of the 13th Five-Year Plan [2] Industry Developments - The Ministry of Industry and Information Technology issued a notice for the "Steel Industry Growth Stabilization Work Plan (2025-2026)," targeting an average annual growth of around 4% in the steel industry's added value [4] - The plan includes precise control of capacity and output, prohibiting new capacity additions, and enhancing the supply and price stability of raw materials like iron ore and coking coal [4] - The State Sports General Administration released guidelines to promote a high-quality development of sports for health, aiming to establish a collaborative service system by 2030 [4] Company Dynamics - OpenAI and NVIDIA announced a strategic partnership, with NVIDIA planning to invest up to 100 billion dollars in OpenAI [6] - BYD's Li Yunfei responded to Warren Buffett's divestment, noting that stock investment involves both buying and selling [6] - Pfizer is nearing a deal to acquire obesity drug developer Metsera for 7.3 billion dollars, with potential additional payments based on performance targets [6] - Longchuan Technology reported a net profit of 827 million to 877 million yuan for the first three quarters, representing a year-on-year increase of 131% to 145% [12]

看似特朗普的财政部部长贝森特看美联储不顺眼？但实际上，两个掌握着经济命脉的部门，真到事上，还是"亲密无间"。 就是在这样向好的预期之下，美联储降息了25个基点，对于降息的原因解释说得也比较明白，主要是因为就业比原本预期的要差。 按照美联储的解释，这个差不只是说移民导致的供给减少，而是岗位需求减少更快，导致失业率是缓慢上升的，虽然目前的失业率还是在健康状态，但是 要提前预防式降息，来刺激企业岗位需求，防止就业进一步恶化。 所以核心两点：经济没问题情况下的预防式降息。 但更有意思的是，在FOMC的经济预测更新上，FOMC委员们调高了近三年的经济增长预期、2026的通胀预期，但调低了政策利率和失业率。 也就是说，2025年除了GDP更好之外，其他不变；2026年开始通胀稍微抬头，但GDP增长预期和失业率预期都是在转好的。 | 2025/9/17 | 2025E | 2026E | 2027E | 2028E | 长期 | | --- | --- | --- | --- | --- | --- | | GDP | 1.6% | 1.8% | 1.9% | 1.8% | 1.8% | | PCE | 3.0% | 2. ...

9月19日，上交所上市审核委员会定于9月26日审议摩尔线程首发事项。 今年6月30日，摩尔线程IPO申请在获上交所受理。随后，上交所进行了两轮问询，9月5日和9月18日， 摩尔线程针对两轮问询进行了回复。 *免责声明：文章内容仅供参考，不构成投资建议 为了解决这一问题，公司高管向部分股东透露，未来五年内，OpenAI计划在租用云服务商的备用服务 器上花费约1000亿美元。这是此前预计到2030年间为租用服务器所支出的3500亿美元之外的额外开支。 综合来看，民生证券认为，当前AI基建海外关注点已从云厂商开支侧向AIDC云计算中心侧扩散，具备 GPU卡先发优势厂商如Oracle、Coreweave、Nebius、AppliedGigital、Iris Energy等。回顾国内，当下AI 叙事逻辑仍不变，重点聚焦AI基建侧。 公司方面，据上市公司互动平台表示， 东华软件：公司与摩尔线程等的合作正深入推进，具体围绕软件适配、市场推广两大核心方向展开，同 时同步推进技术、市场、服务三方面协同。 圣元环保：公司通过认购中原前海股权投资基金的基金份额3亿元人民币间接参与了摩尔线程的投资， 该基金投资摩尔线程金额为人民币2 ...

Group 1 - The automotive market is increasingly prioritizing design and aesthetics, with a call for manufacturers to focus on creating visually appealing vehicles to differentiate their products [1] - Leading solar companies Longi Green Energy and JinkoSolar have reached a settlement regarding global patent disputes, establishing a model for intellectual property cooperation in the industry and shifting the focus from price competition to technology-driven strategies [1] Group 2 - Research from Bank of America and Morgan Stanley indicates that major US tech companies are heavily investing in AI infrastructure, with capital expenditures nearing the peak levels seen during the internet bubble, potentially leading to a price war in cloud services by 2027 if supply and demand become imbalanced [2] - Huawei has proposed a three-step approach to industry intelligence, which includes integrating AI with core business functions, transforming data into industry-specific models, and scaling the deployment of AI agents to reshape key business processes [2]

Investment Rating - The report maintains a "Recommended" rating for several companies in the AI infrastructure sector, particularly those with a first-mover advantage in GPU cards, such as Oracle, Coreweave, Nebius, AppliedGigital, and Iris Energy [3]. Core Insights - The global AI infrastructure investment is accelerating, with significant contracts being signed, such as Nebius securing a deal with Microsoft worth nearly $20 billion, which could reach up to $19.4 billion by 2031 [2]. - Major companies like Oracle and Meta are expanding their cloud businesses, with Oracle projecting a revenue growth of 14% to 16% year-over-year for the second fiscal quarter of 2026, driven by strong cloud performance [1][2]. - The report highlights that the focus of AI infrastructure investment is shifting from cloud vendor spending to AIDC (AI Data Center) cloud computing centers, indicating a broader market trend [3]. Summary by Sections Company Performance - Oracle's first fiscal quarter cloud infrastructure revenue reached $3.3 billion, a 55% year-over-year increase, with a remaining performance obligation of $455 billion, up 359% year-over-year [1]. - Meta's projected spending for 2025 is between $66 billion and $72 billion, primarily for expanding generative AI infrastructure [2]. Market Trends - Global cloud infrastructure service spending is expected to reach $95.3 billion in Q2 2025, reflecting a 22% year-over-year growth [2]. - Microsoft announced an annual AI capital expenditure of $120 billion for the current year, significantly up from $88 billion the previous year [2]. Investment Recommendations - The report suggests focusing on companies within the AIDC chain, such as Runze Technology and Runjian Co., as well as optical module companies like Dekoli and Shijia Photon [3].

Investment Rating - The report maintains a "Recommended" rating for several companies in the AI infrastructure sector, particularly focusing on those with a first-mover advantage in GPU cards [4][5]. Core Insights - The AI infrastructure focus has shifted from cloud vendor spending to AIDC cloud computing centers, indicating a growing investment trend in this area [4]. - Microsoft is building the world's most powerful AI data center in Wisconsin, with a total investment of $4 billion, which will house hundreds of thousands of NVIDIA GB200 chips [1][2]. - The Fairwater data center is designed to handle complex AI workloads with a massive cluster of interconnected NVIDIA GB200 servers, featuring millions of computing cores and extensive storage [3]. Summary by Sections Investment Opportunities - Companies such as Oracle, Coreweave, Nebius, Applied Digital, and Iris Energy are highlighted as having a first-mover advantage in GPU technology [4]. - Domestic companies to watch include Runze Technology, Aofei Data, and Runjian Co., with a focus on the AIDC chain [4]. Key Company Forecasts - The report provides earnings per share (EPS) and price-to-earnings (PE) ratios for key companies, with several receiving a "Recommended" rating based on their projected performance [5]. - For example, Zhongji Xuchuang is projected to have an EPS of 8.58 yuan in 2025 with a PE ratio of 49, while Xinyi Sheng is expected to have an EPS of 8.78 yuan with a PE of 40 [5]. Market Trends - The report notes that AI data centers represent significant capital projects, amounting to hundreds of billions of dollars in investments and involving tens of thousands of advanced AI chips [2][4]. - The integration of data centers, GPU clusters, and network design into a cohesive system is emphasized as a key development in supporting unprecedented AI training and inference workloads [2].

Group 1 - The company is actively engaged in the development of AI infrastructure and has initiated projects such as the third phase of its headquarters and the "Tenglong Huan Niao" smart workshop in Zhejiang [1] - The company confirms that it is involved in these projects, with some engineering already underway, focusing on high-end pipeline research and manufacturing as well as the upstream and downstream supply chain [1] - The implementation of robotic equipment in the workshop is aimed at enhancing new productive forces and promoting the digitalization of the company's manufacturing processes [1]

Investment Rating - The industry investment rating is maintained at "Overweight" [1] Core Views - The report highlights that Oracle has secured an additional 300 billion computing power orders, indicating sustained high demand for global computing power [4][5] - The electronic industry has shown a recovery trend in consumer electronics, with continuous releases of foldable screen smartphones and advancements in AI technology driving demand for AI infrastructure [7] Market Performance - The electronic industry index rose by 6.15% during the week of September 8 to September 12, 2025, outperforming the CSI 300 by 4.77 percentage points [9] - The semiconductor sector reported a 6.52% increase, while consumer electronics rose by 5.17% [2] Valuation - The electronic sector's PE (TTM, excluding negative values) is 61.40X, which is an increase of 3.46X week-on-week, and is at the 43.93 percentile of the past 10 years [3][10] - The PB (LF) stands at 4.90X, also reflecting a week-on-week increase of 0.29X, placing it at the 59.36 percentile historically [3][10] Industry Dynamics - Oracle's significant new orders validate the high demand for global computing power, driven by AI inference and training needs, as well as intense competition among leading AI companies [5] - The report emphasizes the investment opportunities in the North American computing power and domestic computing industry chains [5] Investment Recommendations - The report suggests focusing on investment opportunities in AI infrastructure, edge SOC, and the supply chain for foldable smartphones, maintaining an "Overweight" rating for the electronic industry [7] - Specific companies to watch in the AI infrastructure sector include Cambrian, Chipone, and Aojie Technology; in the edge SOC sector, recommended companies are Rockchip, Hengxuan Technology, and Lattice Semiconductor [7]