你有没有发现，这两年美国那套谈判套路越来越像逼供？一边笑着说合作，一边手里举着刀。最近那场谈崩的消息，我看完只觉得，这回中 国是真不想忍了。 美国这次动手有点狠。直接在芯片、金融、供应链上连出几招，还动用了长臂管辖那一套。据路透社报道，美国商务部在9月宣布进一步扩大 对中国企业的出口限制名单，新增了13家涉及人工智能和半导体领域的公司。理由还是那句老话：国家安全。可问题是，这"安全"到底是谁 的安全？这话题我越看越火大。 我查了下，美国的芯片出口限制其实早在2022年10月就开始，那时候就已经卡死了中国拿到高端GPU的渠道。结果两年过去，中国居然硬是 靠自研和"国产替代"扛了下来。据商务部最新数据，2025年上半年，中国半导体设备进口同比下降了19%，但国内设备出货量增长了近40%。 这说明啥？说明封锁逼出来的，不是倒下，而是自我进化。 这次谈崩，其实在很多人意料之中。因为早在7月底中美上海磋商的时候，美国就已经不断追加所谓"验证机制还要求中方开放特定技术审查 权。这个要求，放哪国都不可能答应。要主权就得有底线，要利益就得算账。美国看中国不配合，就翻脸。然后中国也不演了。 现在的态势很明显。谈判桌已经没意义了 ...

Group 1 - The electronic industry is entering a peak season with the launch of innovative products such as AI smartphones, AI PCs, and AI glasses, leading to a potential recovery in demand due to seasonal factors and government subsidies [1] - The semiconductor sector is experiencing strong demand for computing infrastructure, benefiting advanced semiconductors, with high capital expenditure in cloud AI driving the demand for high-end GPUs, HBM storage, and advanced packaging [1] - The domestic substitution in the semiconductor industry is deepening, with sustained demand for computing power, indicating a high level of industry prosperity [1] Group 2 - The integrated circuit ETF (159546) tracks the integrated circuit index (932087), which selects listed companies involved in semiconductor design, manufacturing, packaging, testing, and related materials and equipment to reflect the overall performance of the integrated circuit industry [1] - The constituent stocks of the index are characterized by high technological content and growth potential, representing the development level and trends of China's integrated circuit industry, with a focus on electronic and information technology sectors [1] - Investors without stock accounts can consider the Guotai CSI All-Share Integrated Circuit ETF Initiator Link C (020227) and Guotai CSI All-Share Integrated Circuit ETF Initiator Link A (020226) [1]

Group 1 - The company, Banan Group, has announced its intention to voluntarily withdraw its shares from the Hong Kong Stock Exchange (HKEX) and will maintain its primary listing on the Singapore Exchange (SGX) after the withdrawal [1][2] - The board approved the proposal to delist from HKEX, with the application submitted on September 12, 2025, contingent upon certain conditions being met [1] - Shareholders will have the option to hold their shares post-delisting or to have their shares deposited with a central custodian for trading on SGX [1] Group 2 - As part of its growth strategy, the company has expanded its operations into new markets in the Asia-Pacific region, particularly Southeast Asia, and has relocated its headquarters to Singapore [2] - The establishment of a factory in Batam, Indonesia, and the primary listing on SGX are aimed at enhancing the company's strategic position in Southeast Asia and improving procurement flexibility for high-end GPUs [2] - The company acknowledges potential challenges from future trade restrictions and tariff changes that may impact its business, despite successfully relocating its headquarters and seizing new opportunities since listing on SGX [2]

今年以来中国互联网ETF(KWEB)上涨接近三成，主要是由市场情绪好转带动估值倍数提升所驱动。事 实上，电商巨头布局即时零售导致今年行业盈利预测下修14%，但大型互联网公司市盈率(PE)估值倍数 扩大至16倍左右。 他续指，由于内地外卖市场未来一两个季度的投资力度仍有较大不确定性，相对更喜欢盈利确定性较高 的游戏、旅游子版块。 格隆汇9月1日｜瑞银投资银行中国互联网研究主管方锦聪表示，尽管短期支出波动，内地主要互联网公 司全年资本开支展望基本维持不变。据行业调研，内地高端GPU库存足以满足未来几个季度的需求。 | 上证指数 | 深证成指 | 北证50 | | --- | --- | --- | | 3875.53 12828.95 1568.62 | | | | +17.60 +0.46% +132.80 +1.05% -5.62 -0.36% | | | | 科创50 | 创业板指 | 万得全A | | 1357.15 | 2956.37 | 6225.51 | | +15.84 +1.18% +66.25 +2.29% +49.77 +0.81% | | | | 沪深300 | 中证500 | 中证A50 ...

Core Viewpoint - A new bill proposed by Senator Tom Cotton aims to implement "geolocation tracking" features in high-end GPUs from companies like Nvidia and AMD to prevent these technologies from falling into the hands of competing nations [1][3]. Group 1: Bill Details - The bill targets not only AI chips but also high-performance gaming graphics cards [3]. - If passed, the measures will take effect six months after the bill's approval [3]. - Manufacturers of high-performance AI processors and graphics cards, such as Nvidia, Intel, and AMD, will be required to embed geolocation technology in their products to monitor the physical location of the hardware [5][10]. Group 2: Technical and Compliance Challenges - The implementation of geolocation tracking technology is not straightforward, especially for already designed high-end processors and graphics cards [12]. - Companies like Nvidia and AMD will face increased R&D costs and time due to the need to adjust production processes to incorporate tracking mechanisms [12][14]. - Exporting chip manufacturers will bear additional responsibilities, including tracking the location and usage of their products post-sale and reporting any unauthorized transfers [14]. Group 3: Impact on the Industry - Since 2022, the U.S. government has imposed strict export controls on advanced chips, particularly targeting AI and high-performance computing sectors [15]. - The recent export controls have significantly impacted companies, with AMD estimated to lose around $800 million in potential revenue and Nvidia facing losses of up to $5.5 billion [15]. - The bill also sets the stage for future regulatory upgrades, requiring annual assessments and potential new requirements based on technological advancements [18].

Core Viewpoint - The article discusses a new bill proposed by U.S. Senator Tom Cotton that mandates the integration of "geolocation tracking" technology in high-end GPUs and AI chips produced by companies like NVIDIA and AMD to prevent unauthorized access by foreign entities [2][4][7]. Group 1: Bill Details - The bill targets high-performance AI processors and gaming graphics cards, requiring manufacturers to embed geolocation technology in their products [4][7]. - If passed, the measures will take effect six months after the bill's approval [5]. - The primary goal of the bill is to ensure that strategic hardware is not used by unauthorized foreign entities [14]. Group 2: Implications for Manufacturers - The requirement to add geolocation tracking poses significant challenges for chip manufacturers, as it necessitates adjustments to existing designs and production processes, potentially increasing R&D costs and time [15][16]. - Manufacturers will be responsible for continuously tracking the location and usage of their products after export, with obligations to report any unauthorized transfers or tampering [19][21]. - NVIDIA has publicly stated its inability to track hardware post-sale, highlighting concerns about the feasibility of the new requirements [22]. Group 3: Regulatory Landscape - The bill sets the stage for future regulatory upgrades, including annual assessments and joint research by the Department of Commerce and the Department of Defense to explore additional protective measures [28][29]. - The evaluations will assess the latest security technology advancements applicable to export-controlled products, potentially leading to new requirements [30][31]. - The bill emphasizes the need to protect sensitive business secrets and intellectual property during the development and deployment of these technologies [32][33]. Group 4: Economic Impact - The recent export controls have already significantly impacted companies like AMD and NVIDIA, with AMD estimated to lose around $800 million in potential revenue and NVIDIA facing losses of up to $5.5 billion due to stringent restrictions on advanced chips [25].

Market Overview - The global AI server market is projected to reach $125.1 billion in 2024, $158.7 billion in 2025, and $222.7 billion by 2028, with generative AI servers expected to increase their market share from 29.6% in 2025 to 37.7% in 2028 [1] Major Players' Shipment Data - NVIDIA holds nearly 70% of the AI chip market share, with the new Blackwell platform expected to account for 82% of its high-end GPU shipments in 2025. The company plans to launch the B30 for the Chinese market in the second half of 2025, with the B300 and GB300 expected to contribute 60-65% of its total GPU shipments for the year [2][3] - AMD's high-end GPU shipments are forecasted to grow by 48% in 2025, reaching approximately 585,000 units. The MI325 series has a low adoption rate, while the MI350 series is expected to enter the market in the second half of 2025 [5][4] - Intel's Gaudi3 high-end AI chip is projected to have a shipment volume of around 100,000 units in 2025, targeting CSPs and IBM as primary customers [6] - Google is expected to lead in cloud service provider (CSP) shipments with approximately 2.2 million TPUs in 2025, while AWS's self-developed ASIC shipments are projected to reach over 1.8 million units, nearly doubling from previous figures [7][8] Chip Type and Shipment Forecast - NVIDIA's high-end GPU shipments are expected to total around 6.6 million units in 2025, with significant contributions from the Blackwell platform [3] - AMD's MI series is expected to see a total shipment of 585,000 high-end GPUs in 2025, with the MI350 series anticipated to compete directly with NVIDIA's offerings [5] - The Ascend (昇腾) ASIC is projected to reach 450,000 units in 2025, driven by domestic AI demand and trade restrictions [8]

Core Viewpoint - Increasing evidence suggests that AI training does not necessarily rely on high-end GPUs, which may slow down Nvidia's future growth [2][5][14] Group 1: Nvidia's Financial Performance - Nvidia's data center business has experienced strong growth, with revenue increasing by 216% in FY2024 and 142% in FY2025 [2] - Revenue growth rates for Nvidia are projected at 63% for FY2026, driven by a 70% increase in the data center segment, alongside a recovery in gaming and automotive markets [8][9] - The company's total revenue is expected to reach $430 billion in Q1 FY2026, with a slight fluctuation of 2% [6] Group 2: Competitive Landscape - Ant Group's research indicates that their 300B MoE LLM can be trained on lower-performance GPUs, reducing costs by 20%, which poses a significant risk to Nvidia's market position [2][5] - Major hyperscalers like Meta are developing their own AI training chips, reducing reliance on Nvidia's GPUs, with Meta's internal chip testing marking a critical milestone [5][14] - Custom silicon solutions from companies like Google and Amazon are emerging as attractive alternatives for AI training and inference [5] Group 3: Long-term Growth Challenges - Nvidia's high-end GPU growth may face increasing resistance as AI enters the inference phase and lower-cost models become more prevalent [14] - Analysts have revised growth expectations for Nvidia's data center business, projecting a slowdown to 30% growth in FY2027 and further declines to 20% from FY2028 to FY2030 [8][9] - The company's operating expenses are expected to grow by 19% from FY2028 to FY2030, impacting profit margins [9] Group 4: Capital Expenditure Trends - Major tech companies are significantly increasing capital expenditures, with a projected 46% year-over-year growth in 2025, which may boost demand for Nvidia's GPUs in the short term [12][13] - Nvidia has established its own custom ASIC division, potentially mitigating risks from competitors like Broadcom and Marvell [14]