A股三大股指集体高开，沪指涨0.36%，创业板指涨0.9%。盘面上，CPO、光刻机、算力租赁等板块涨 幅居前。 这种情况下，调仓的思路不是刻意回避AI叙事，而是尽量选择ROE底部向上趋势性抬升的品种，从这 个角度来看，适度增加对化工、有色、电新这样沉寂比较久且利润率和行业景气度在历史相对低点的行 业，是更优的选择。 另外，AI叙事只是影响了行情斜率而不是趋势。，如果未来AI叙事出现波动，的确可能会带来这些行 业剧烈的波动，但只要ROE能实现从底部开始向上的持续抬升，这种股价波动都只是短期的(相对于那 些ROE在历史高点的行业而言)，不会威胁到本金安全。 招商证券：有色、钢铁、建材是当前可以考虑布局的顺周期选择 机构看后市 中信证券：适度增加化工、有色、电新的仓位，是更优的选择 10月以来市场波动加大，但择时成功率并不高，背后的原因是增量资金的底层结构在发生变化，稳健绝 对收益型资金持续入市在降低传统激进策略择时的有效性。当前真正重要的变量还是企业出海环境的稳 定性以及AI，涉及到的是中美关系以及AI基础设施的投建进程。当前不仅是TMT板块，连有色、化 工、电新的上涨直接或间接都受到AI叙事的影响，而这些板块占 ...

Core Insights - The article discusses the rapid growth of the computing power leasing market in China, driven by the increasing demand for AI capabilities and the challenges posed by U.S. technology restrictions on high-end chips [1][7]. Group 1: Industry Overview - Computing power leasing allows businesses or individuals to rent necessary computing resources, including CPU, GPU, and FPGA, typically provided by specialized data centers or cloud service providers [1][2]. - The market for computing power leasing in China has experienced double-digit annual growth rates in recent years, with a projected market size of 377 EFlops in 2024, representing an 88.5% year-on-year increase [1][10]. Group 2: Market Dynamics - The demand for computing power has surged due to the rapid development of AI large models, with significant examples including OpenAI's GPT series and others, leading to an exponential increase in computing resource requirements [1][8]. - The internet industry is the primary demand driver for computing power leasing, accounting for 62% of the total demand, followed by government (14%), finance (6%), healthcare (5%), education (4%), and manufacturing (4%) [1][12]. Group 3: Pricing Trends - The rental prices for computing power have significantly decreased, with global market prices for H100 GPUs dropping from approximately $5 per GPU·hour in early 2024 to about $2.4 per GPU·hour by mid-2025. In China, prices fell from $13 per GPU·hour to around $5 per GPU·hour [1][17]. Group 4: Future Trends - The computing power leasing industry is expected to continue evolving, with trends towards cloud integration, regional collaboration, and increased emphasis on domestic alternatives in response to international challenges [1][18].

Summary of Key Points from the Conference Call of Xiechuang Data Company Overview - **Company**: Xiechuang Data - **Industry**: AI Computing and Data Storage Financial Performance - **Revenue**: In the first three quarters of 2025, revenue reached 77.49 billion, a year-on-year increase of 25.03%. In Q3 alone, revenue was 33.87 billion, up 86.43% year-on-year, with net profit attributable to shareholders at 2.66 billion, a 33.44% increase, marking a new high since the company went public [2][3] - **Total Assets**: By Q3, total assets reached 181.94 billion, an increase of nearly 150% since the beginning of the year [3] Research and Development - **R&D Investment**: R&D expenses for the first three quarters of 2025 amounted to 2.29 billion, with significant investment in Q3 aimed at enhancing competitiveness in the AI computing sector, particularly in computing leasing platforms [2][5] - **Focus Areas**: The primary focus of R&D is on AI computing platforms and related projects [5] Business Growth and Strategy - **Storage Business**: The storage business is expected to recover to pre-spin-off levels by Q1 2026, with continued high growth anticipated in the coming quarters. A strategic partnership with SanDisk aims to secure 20% of a major internet client's storage needs next year [2][7] - **Server Procurement**: The company disclosed server procurement contracts not exceeding 122 billion and initiated H-share project financing to raise approximately 100 billion for future investments, aiming for a growth scale of at least 5 to 10 times by 2026 [4][10] - **Cloud Services**: The company offers cloud services based on computing rather than leasing, including large model scheduling and game rendering, with monthly billing [11] Market Outlook - **AI Investment Sentiment**: The market views the company's ongoing investment in AI as a positive signal, with expectations for significant output and competitive advantages [9] - **High-End Computing Demand**: The tightening demand for high-end computing presents opportunities for the company, especially in light of the ongoing US-China tensions [13][14] Future Projections - **Revenue Expectations**: The company anticipates a revenue rebound in Q4, projecting around 6 billion from two clusters, with a complete revenue cycle expected in Q1 2026 [22] - **Storage Business Growth**: The storage business is projected to double in revenue next year, with significant contributions from the partnership with SanDisk [17] Challenges and Responses - **US-China Relations**: The company has established assembly and repair factories in Thailand, Japan, and Australia to mitigate risks associated with US-China tensions [13] - **Market Demand**: There is a strong market-driven demand for recovery and remanufacturing services, with significant price increases observed in server procurement [18] Collaboration and Partnerships - **Partnerships**: The company collaborates with NVIDIA in the robotics simulation field and has distinct roles compared to its partnership with Alibaba, focusing on implementation rather than research [20][27] Conclusion Xiechuang Data is positioned for significant growth in the AI computing and data storage sectors, with strong financial performance, strategic partnerships, and a proactive approach to market challenges. The company's focus on R&D and cloud services, along with its response to geopolitical dynamics, underscores its commitment to maintaining a competitive edge in the industry.

Core Viewpoint - Zhiyun International Holdings (08521) has announced a leasing service agreement to purchase computing power from Pansemi (Singapore) Pte. Ltd, with a total payment of $3.9 million for the service period from November 1, 2025, to October 31, 2026 [1] Group 1 - The agreement involves a total payment of $3.9 million, which must be paid in full within three calendar days after signing the leasing service agreement [1] - The supplier will allocate and deliver a computing power of 897,726 trillion hashes per second to the buyer upon receipt of the full payment [1] - The service period for the computing power lease is set for one year, starting from November 1, 2025, to October 31, 2026 [1]

Core Insights - The recent failure of Qunxing Toys in the computing power rental sector marks another setback for traditional companies attempting to diversify into this field, highlighting the challenges faced by many A-share companies in pursuing new growth avenues amidst stagnating core businesses [1][2]. Industry Overview - The surge in generative AI since 2024 has led to an exponential increase in demand for computing power, creating a rapidly growing market that many A-share companies are eager to enter as they seek new growth opportunities [2][10]. - A diverse range of companies, from toy manufacturers to construction firms, have announced their entry into the computing power rental business, driven primarily by the need to overcome growth bottlenecks in their core operations [2][3]. Company Examples - Qunxing Toys reported a nearly 500% year-on-year revenue increase in 2024 but still faced significant losses, prompting its entry into the computing power sector through a planned acquisition of a computing service provider [2]. - Hainan Huatie, previously focused on construction equipment rental, announced a significant investment of 10 billion in computing power and secured a contract worth nearly 3.7 billion, indicating a strong push for transformation [3]. - Lianhua Holdings, despite facing losses in its computing power business, managed to achieve a breakeven point in the first half of 2025, although it still contends with rising interest expenses [7]. Market Reactions - The capital market has responded positively to announcements related to computing power, with stock prices of companies involved in this sector experiencing significant increases following such news [4][6]. - However, as the initial excitement wanes, a clear differentiation is emerging among companies based on their actual performance and the sustainability of their computing power ventures [6][11]. Future Outlook - The computing power rental market is projected to grow at a compound annual growth rate of 53% over the next three years, with the market size expected to reach 1,346 EFlops by 2027, supported by national strategic initiatives [10][12]. - Despite the promising outlook, the industry presents high barriers to entry, including the need for stable supply chains, strong operational capabilities, and effective financial management [11][12].

Core Viewpoint - The article discusses the impact of recent regulatory changes and market dynamics on the domestic AI chip industry, particularly focusing on the shift towards domestic chips and the evolving landscape of computing power rental and supply chain management. Group 1: Regulatory Changes and Market Dynamics - A new "window guidance" document mandates that projects with local subsidies must exclusively use domestic chips, prohibiting the use of foreign GPUs like H20 in new market-oriented projects [2] - Projects outside the "national hub computing power facility cluster" will not receive financial or electricity subsidies, leading to the cancellation of existing local policies [2] - The ban on foreign chips and the push for domestic alternatives create a favorable environment for domestic AI chip manufacturers [3] Group 2: Computing Power Rental Market - Major companies in East China are paying significantly higher rental prices for computing power, which benefits both the companies and their IDC partners by enhancing cloud business collaboration and easing financial pressures [5] - The rental market is seeing a shift, with a model company planning to lease 60 H200 servers, indicating a trend towards smaller, more flexible rental agreements that align with technology upgrade cycles [8] - GPU rental prices have drastically decreased, with H100 prices dropping from 60,000-80,000 yuan to around 40,000 yuan, while H200 is priced at 60,000 yuan, making it a more attractive option for companies [9] Group 3: Supply Chain and Vendor Management - Leading companies are tightening their supplier networks, imposing penalties exceeding 100 million yuan on non-compliant suppliers, indicating a move towards a more stable and closed supply chain [6] - The construction of computing centers is now focused on calculating potential losses rather than profits, reflecting a more cautious approach in project planning due to delayed subsidy disbursements [7] Group 4: Emerging Trends in Chip Technology - Domestic RISC-V chip companies are expected to aggressively enter the server market next year, with several firms already developing high-performance RISC-V server chips [10][11] - The storage chip market is experiencing price increases driven by strong demand from AI and high-performance computing sectors, with distributors adopting strategies to hold inventory until prices rise further [12]

Group 1 - A-shares indices declined on October 30, with the ChiNext index dropping over 1%, led by sectors such as telecommunications, media, and electronics [1] - New Yi Sheng reported a Q3 2025 revenue of 16.505 billion yuan, a year-on-year increase of 221.70%, and a net profit of 6.327 billion yuan, up 284.38% year-on-year [1] - The Q3 revenue showed a quarter-on-quarter decline of 4.97%, while net profit increased by 0.63% quarter-on-quarter, attributed to the impact of major customer delivery schedules and material shortages [1] Group 2 - Dongxing Securities noted that the "14th Five-Year Plan" draft suggests a shift in national investment towards efficient development of communication and computing networks, indicating continued expansion in related markets [2] - The cloud computing 50 ETF tracks an index with a high AI computing content, covering popular concepts such as optical modules, computing leasing, and data centers [2]

Core Viewpoint - The company, Qunxing Toys, has announced the termination of its fifth major asset restructuring attempt since its listing, primarily due to failure to reach consensus on key terms of the transaction [1][7]. Group 1: Company Background and History - Qunxing Toys was established in 1996 and initially focused on toy research and production, but has since attempted multiple cross-industry transformations, all of which have failed [1][8]. - The company’s revenue peaked at 493 million yuan in 2011 but plummeted to 54 million yuan by 2017, and further declined to 19 million yuan in 2018 [2][8]. - Wang Sanshou became the actual controller of Qunxing Toys in November 2018, but his tenure was marred by personal financial issues that worsened the company's situation [2][3]. Group 2: Restructuring Attempts - Qunxing Toys has made five unsuccessful attempts at cross-industry restructuring since its listing, targeting various sectors including mobile gaming, nuclear equipment, energy technology, and electronics [7][8]. - The latest attempt involved acquiring at least 51% of Hangzhou Tiankuan Technology Co., which was terminated due to disagreements on transaction pricing and terms [1][9]. Group 3: Financial Issues and Management Changes - Wang Sanshou's management led to significant financial mismanagement, with 327 million yuan misappropriated, representing 44% of the company's net assets as of 2019 [2][3]. - Zhang Jincheng took over as chairman and general manager in late 2020, successfully stabilizing the company and shifting its focus to alcohol sales, which contributed 337 million yuan to revenue in 2024 [3][5]. Group 4: Future Prospects and Strategic Moves - Following the restructuring failure, Qunxing Toys is exploring acquisitions in high-tech fields like artificial intelligence, despite skepticism about the viability of its current computing power business [1][10]. - The company has begun emphasizing its computing power leasing business, which has attracted new investors, although it currently operates at a loss [10][11].

Core Viewpoint - The company, Qunxing Toys, has attempted five mergers and acquisitions over the past decade, all of which have failed, with the latest attempt to acquire Tian Kuan Technology officially terminated due to disagreements on key terms [1][7]. Group 1: Mergers and Acquisitions Attempts - Qunxing Toys announced on October 20 that it would terminate its plan to acquire at least 51% of Tian Kuan Technology due to a lack of consensus on core issues [1]. - The company has a history of pursuing acquisitions in various sectors, including gaming, nuclear power, military, power batteries, and consumer electronics, but has not succeeded in any of these attempts since its listing in 2011 [1][2]. - The latest acquisition attempt was part of a strategy to enter the computing power leasing industry, which the company began exploring last year [2]. Group 2: Financial Performance and Business Transition - After its listing, Qunxing Toys experienced a decline in net profit from 52 million yuan in its first year to 14.87 million yuan in 2014, prompting a search for quality assets to enhance its business [3]. - The company has gradually shifted away from toy manufacturing to focus on other sectors, including alcohol sales, property leasing, and smart computing power leasing, as indicated in its 2024 annual report [8][9]. - The transition has led to periods where the company had zero revenue from its original toy business, highlighting the challenges of maintaining a coherent business strategy [9]. Group 3: Current Ownership and Control - Following multiple failed acquisitions, the original controlling shareholders, Lin Weizhang and Huang Shiqun, sought to transfer control to strategic investors, ultimately resulting in a change of control to Wang Sanshou in 2018 [6]. - As of September 2024, Qunxing Toys has no controlling shareholder or actual controller, with ownership highly dispersed among individual shareholders [9].

Core Viewpoint - The company, Qunxing Toys, announced the termination of its plan to acquire at least 51% of Hangzhou Tiankuan Technology due to failure to reach consensus on key terms such as transaction price and party arrangements [1][2]. Group 1: Acquisition Details - The acquisition was initially announced on February 26, leading to a surge in stock prices, but ultimately did not progress as planned [1][2]. - Qunxing Toys stated that the transaction was still in the planning stage and had not been submitted for board or shareholder approval [1]. - The company committed to not planning any major asset restructuring for at least one month following the announcement [1]. Group 2: Financial Performance - Qunxing Toys reported a revenue of 5.79 billion yuan for Tiankuan Technology in 2024, with approximately 58% of that revenue coming from its core business of AI computing center construction and operation [2]. - The company’s stock price has decreased by 24.84% over the past eight months, with a year-to-date decline of 14.44% [3]. - In the first half of the year, the AI computing business generated revenue of 18.11 million yuan, accounting for 10.32% of total revenue, but incurred a net loss of 18.61 million yuan [4]. Group 3: Historical Context - This acquisition attempt marks the fifth restructuring effort by Qunxing Toys since its listing, with previous attempts also failing despite targeting popular sectors [2]. - The company has previously attempted to diversify into various sectors, including mobile gaming and environmental technology, but these efforts have not been successful [2]. - The company’s main revenue source remains alcohol sales, which accounted for over 80% of its revenue, with a low gross margin of 2.32% [5].