证券代码：002230 证券简称：科大讯飞 公告编号：2025-047 公司本次向特定对象发行 A 股股票事项尚需深圳证券交易所审核通过，并获得中国 证券监督管理委员会同意注册后方可实施。最终能否通过深圳证券交易所审核并获得中国 证券监督管理委员会同意注册的决定及其时间仍存在不确定性。公司将根据上述事项的进 展情况，及时履行信息披露义务，敬请广大投资者谨慎决策，注意投资风险。 科大讯飞股份有限公司 关于2025年度向特定对象发行A股股票审核问询函回复及募集说明 书等申请文件更新的提示性公告 本公司及董事会全体成员保证信息披露的内容真实、准确、完整，没有虚 假记载、误导性陈述或重大遗漏。 科大讯飞股份有限公司（以下简称"公司"）于 2025 年 10 月 24 日收到深圳证券交易 所（以下简称"深交所"）上市审核中心出具的《关于科大讯飞股份有限公司申请向特定 对象发行股票的审核问询函》（审核函〔2025〕120041 号）（以下简称"《审核问询函》"）。 公司会同相关中介机构对《审核问询函》所提出的问题进行了认真研究和逐项落实后，对 《审核问询函》所列的问题进行逐项说明和作出回复，并对募集说明书等申请文件的内容 ...

Core Viewpoint - The article emphasizes that China needs "problem-solving AI" rather than pursuing AGI (Artificial General Intelligence), focusing on practical applications that address real-world issues in education, healthcare, and industry [4][5][6]. Group 1: AI Development and Strategy - By 2025, the global AI industry will enter a "dividend period," but the development paths of China and the U.S. are diverging significantly [5]. - Keda Xunfei's chairman, Liu Qingfeng, outlined four key strategies: self-controllable, integrated hardware and software, industry depth, and personalization [5][12]. - Keda Xunfei's Starfire X1.5 model achieved a training efficiency of 93% compared to NVIDIA A800, showcasing advancements in domestic computing power [10][11]. Group 2: Market Performance and Financials - In 2024, Keda Xunfei led the bidding market with 91 projects and a total bid amount of 936 million yuan, further expanding its lead in 2025 with 108 projects [10][11]. - Despite a 10% revenue growth in Q3 2025, Keda Xunfei's net profit increased by 202%, indicating improved business quality [16][17]. Group 3: Industry Applications - In education, Keda Xunfei developed a comprehensive error analysis system that has been validated by experts, enhancing teaching efficiency [17][18]. - In healthcare, the AI assistant has provided over 11 billion diagnostic aids, significantly improving diagnostic accuracy in grassroots medical settings [19][22]. Group 4: Competitive Advantages - Keda Xunfei's strategy emphasizes deep integration of technology and industry, focusing on long-term relationships with state-owned enterprises and government sectors [23][24]. - The company’s approach to AI is characterized by a focus on personalized services, leveraging user data to enhance AI interactions [25][26]. Group 5: Global Strategy and Positioning - Keda Xunfei's international strategy aligns with the "Belt and Road" initiative, offering sovereign AI models to countries seeking alternatives to U.S. technology [40][42]. - The company aims to provide localized solutions that meet the specific needs of Chinese users, contrasting with the broader AGI ambitions of Silicon Valley [43][44]. Group 6: Future Outlook - The article concludes that Keda Xunfei represents a stabilizing force in China's AI landscape, focusing on practical applications that serve national interests rather than speculative technological pursuits [45][49].

Core Viewpoint - The article emphasizes that China needs "problem-solving AI" rather than pursuing AGI (Artificial General Intelligence), focusing on practical applications that address national and social needs [2][3][43]. Group 1: AI Development Paths - By 2025, the global AI industry will enter a "dividend period," but the development paths of China and the US are diverging significantly [3]. - While US tech giants are losing billions annually in pursuit of AGI, Chinese companies like iFlytek are focusing on commercializing AI to solve real-world problems [3][43]. Group 2: iFlytek's Strategic Approach - iFlytek's chairman, Liu Qingfeng, outlined a four-dimensional strategy: self-controllable, integrated software and hardware, industry depth, and personalization [3][48]. - iFlytek's "Xinghuo X1.5" model is the first general model trained on fully domestic computing power, achieving training efficiency improvements over NVIDIA's A800 [4][6][8]. Group 3: Market Performance and Financials - iFlytek led the bidding market in 2024 with 91 projects and a total bid amount of 936 million yuan, further increasing to 108 projects in the first three quarters of 2025 [8][9][30]. - Despite a 10% revenue growth year-on-year in Q3 2025, iFlytek's net profit increased by 202%, indicating improved business quality [15][16]. Group 4: Industry Applications - iFlytek's AI solutions are deeply integrated into education and healthcare, with tools like the "Intelligent Medical Assistant" significantly improving diagnostic accuracy in rural healthcare settings [17][23]. - The company has developed a comprehensive error analysis system for education, enhancing teaching effectiveness through AI-driven insights [16][17]. Group 5: Competitive Landscape - iFlytek's competitive advantage lies in its long-term data accumulation and deep partnerships with hospitals, making it difficult for competitors to replicate its success [23][24]. - The shift from hardware to software services in AI projects is evident, with clients increasingly valuing ongoing AI capabilities over one-time hardware purchases [35][36]. Group 6: Global Strategy and Positioning - iFlytek's international strategy aligns with China's Belt and Road Initiative, offering "sovereign models" to countries seeking alternatives to US technology [38][40]. - The company aims to leverage its capabilities to serve not just the Chinese market but also to establish a presence in Southeast Asia and Africa [38][40]. Group 7: Conclusion on AI's Role in China - The article concludes that iFlytek represents a pragmatic approach to AI in China, focusing on real-world applications rather than abstract technological ideals [43][48]. - The company's strategy is seen as a response to the unique needs of the Chinese market, emphasizing practical solutions that enhance everyday life [48][49].

Core Insights - The article highlights the transition of AI from technical breakthroughs to practical applications, emphasizing the importance of creating sustainable social and economic value in real-world scenarios [15] Group 1: Company Strategy - iFlytek's "Artificial Intelligence + Education" strategy is exemplified by the Starfire Intelligent Grading Machine, which significantly reduces teacher grading time from 90 minutes to 15 minutes while providing detailed diagnostic feedback [1] - The company focuses on "self-controllable" and "soft-hard integration" strategies, aiming to balance between being an algorithm provider and a hardware manufacturer [4][5] - iFlytek's deep learning model, Starfire X1.5, features 293 billion total parameters and 30 billion active parameters, achieving a training efficiency increase from 30% to 84% [4] Group 2: Market Applications - iFlytek's AI applications extend to education, healthcare, and transportation, with products like the AI Grading Machine and AI Medical Assistant enhancing user experience and operational efficiency [8][9] - The AI Medical Assistant has been deployed in 801 districts, providing diagnostic pathways and health management services, achieving over 1.6 million AI health consultations with a 98% user satisfaction rate [9] - In the automotive sector, iFlytek's collaboration with Chery aims to create a smart cockpit that integrates emotional intelligence and situational awareness, enhancing user interaction [9] Group 3: Financial Performance - In Q3 2025, iFlytek reported revenue of 6.078 billion yuan, a year-on-year increase of 10.02%, with net profit exceeding 200% [12] - The company has positioned itself as a leader in the AI market, securing 60 projects with a total bid amount of 5.45 billion yuan, maintaining its status as the top bidder for five consecutive quarters [13] Group 4: Ecosystem Development - iFlytek is building an open technology ecosystem, modularizing core capabilities like speech recognition and machine translation, and engaging nearly 9.62 million developer teams [12] - The company's technology is being integrated into various regional industries, enhancing its global footprint and operational reach [12] Group 5: Future Outlook - The focus of AI competition is shifting from model size to practical application and value creation in various sectors, with iFlytek's approach serving as a model for "Chinese-style AI" [15]

Core Concept - The article discusses the launch of "AI that understands you better," aimed at creating a more open, multilingual large model application ecosystem to address industry needs and promote the inclusive sharing of artificial intelligence [2][3]. Group 1: AI Development and Applications - The current era of large models is transitioning from "potential benefits" to "realized benefits," with AI becoming a core force reshaping industries and daily life [2][4]. - AI applications in education include intelligent grading machines and AI learning assistants that provide personalized tutoring, enhancing the learning experience [3][4]. - In healthcare, the AI-powered app "Xiao Yi" integrates personal health data to offer tailored health management plans, acting as a "family doctor" [3]. Group 2: AI in Business Operations - AI is being integrated into various business functions such as legal, document writing, and market analysis, moving from "efficiency improvement" to "value reconstruction" [4][5]. - The digital assistant "Xiao Fei" showcases advanced multimodal interaction capabilities, enhancing user experience through emotional intelligence and personalized interactions [4][13]. Group 3: Technical Innovations - The newly launched deep reasoning model "Xunfei Spark X1.5" boasts a 100% improvement in reasoning efficiency compared to its predecessor, with a total parameter count of 293 billion [7][10]. - The model's training efficiency has significantly improved, reducing the time required for training on domestic computing power, which is crucial for achieving self-sufficiency in AI technology [8][10]. Group 4: Strategic Focus Areas - The article emphasizes four key areas for the realization of "AI that understands you better": self-control, integrated software and hardware, industry depth, and personalization [6][10]. - The importance of a fully domestic computing power platform is highlighted, as it ensures technological security and the autonomy of the AI industry [10][11]. Group 5: Ecosystem and Collaboration - The company aims to foster a prosperous AI ecosystem by collaborating with global partners in technology, industry, and education [20]. - The growth of the developer community, with over 968 million developers, indicates the vibrant potential of artificial intelligence [18].

Core Insights - The report highlights a potential rotation in consumer styles, driven by a recovery in CPI and favorable profit-valuation comparisons, suggesting that consumer sectors may experience a rebound [6][2][7] Company Summaries 1. China Eastern Airlines (600115.SH) - Notable inclusion in the November stock selection list, indicating positive sentiment towards the airline sector [1] 2. Baijiu Industry - The baijiu industry is experiencing a significant decline, with revenue and net profit growth rates of -5.8% and -6.9% respectively for the first three quarters of 2025. The third quarter saw a sharper decline with revenue and net profit growth rates of -18.5% and -22.1% respectively. The industry is transitioning from "over-competition" to "orderly competition" as companies reduce channel expenses to stabilize prices [8][7] 3. Baijun Medical (佰仁医疗) - The company reported a revenue of 382 million yuan for the first three quarters of 2025, a year-on-year increase of 30.58%. The net profit attributable to shareholders was 93 million yuan, up 57.93%. The third quarter alone saw a revenue of 134 million yuan, a 31.54% increase year-on-year, although net profit decreased by 9.39% [3][11][12] 4. Food and Beverage Sector - The food and beverage industry is expected to recover due to policies aimed at boosting consumer spending and improving macroeconomic data. The report notes that the core CPI has shown signs of recovery, with September and October figures at 1.0% and 1.2% respectively, indicating a positive trend in consumer prices [7][2][6] 5. Frozen Food and Beer Sectors - The frozen food sector is closely linked to the restaurant industry, with leading companies showing significant recovery as they adapt to market changes. The beer sector, while still facing challenges, is expected to benefit from the recovery in restaurant consumption in 2026 [9][8] 6. Consumer Spending - The report emphasizes that improving consumer spending is a key goal in the "14th Five-Year Plan," with a focus on enhancing domestic demand as a primary driver of economic growth. The contribution of final consumption expenditure to GDP growth has been higher than that of capital formation in recent years [7][6] 7. Research and Development in Baijun Medical - Baijun Medical has increased its R&D investment, with 118 million yuan spent in the first three quarters, accounting for 30.81% of its revenue. The company has several products in the approval process, which are expected to contribute to future growth [13][12][11]

Group 1 - A series of innovative educational products were showcased at the "Ning Innovation Products" event, highlighting advancements in AI-driven grading systems that can complete class assignments in just 20 minutes [1][2] - The AI grading device developed by iFLYTEK features capabilities such as efficient data collection, printing of exam records, and precise analysis of student performance, allowing for targeted teaching based on individual weaknesses [1] - Chengmai Technology presented a multi-screen display computer designed for educational settings, enhancing teaching efficiency through collaborative screen management and interactive feedback [1] Group 2 - Various companies, including Chengmai Technology, Phoenix Media, iFLYTEK, and Jiyu Information, introduced software and hardware solutions tailored for educational environments, facilitating discussions on topics like "teaching software adaptation" and "AI teaching platform integration" [2]

Core Insights - The article discusses the strategic choice of iFlytek to pursue a "long-termism" approach in AI hardware development, contrasting it with the prevalent "hit product mentality" in the industry [1][5] - iFlytek's "soft and hard integration" strategy aims to create a comprehensive technology system that deeply integrates AI capabilities into existing industrial scenarios rather than focusing solely on creating standout hardware products [1][3] AI Hardware Development - The current trend in the AI hardware sector shows a tendency to prioritize hardware over integration, with many products failing to achieve effective soft and hard collaboration [1] - iFlytek's solutions, such as the X5 smart office notebook and AI translation headset, demonstrate high accuracy rates in noisy environments, achieving 95.08%, 97.1%, and 98.69% respectively [1] Technological Innovations - iFlytek's digital assistant "Xiao Fei" showcases multi-modal interaction capabilities, enabling personalized recommendations and task completion based on conversation history [2] - Key technologies introduced include personalized memory and voice replication, allowing for high-fidelity voice recreation and a deeper understanding of user preferences [2] Market Positioning - Unlike international giants that view consumer AI hardware as the next entry point, iFlytek focuses on practical value realization by embedding AI capabilities into existing industries [3][4] - The company aims to leverage proven AI capabilities to quickly generate scale effects and commercial returns by integrating them into established scenarios like automotive, education, and healthcare [4] Industry Impact - iFlytek's AI products have already made significant inroads in education and healthcare, with the "Smart Medical Assistant" model improving diagnostic accuracy from 87% to 96% and halving medical record writing time [5] - The company's approach emphasizes the tangible realization of AI benefits across various sectors, moving beyond mere concepts to actual value delivery [5]

Core Insights - The core message of the articles revolves around the significant advancements and strategic vision of iFLYTEK in the AI sector, particularly with the launch of the "Xunfei Xinghuo X1.5" model, which emphasizes a more personalized AI experience [1][19]. Group 1: Product and Technology Development - iFLYTEK has launched the "Xunfei Xinghuo X1.5," a deep reasoning model, showcasing its capabilities in AI technology and product integration [1]. - The company has introduced various AI hardware solutions, including "AI + Microphone Array" and "AI + Speaker Array," enhancing multi-modal interaction experiences [1][14]. - The new model features personalized memory capabilities, allowing for long-term user profiling and customized interactions [14]. Group 2: Market Position and Achievements - iFLYTEK has established itself as a leader in the large model market, with 91 successful project bids in 2024, amounting to 848 million yuan, the highest among all vendors [6]. - By the third quarter of 2025, iFLYTEK maintained a dominant position in the public bidding market for large models among central state-owned enterprises, surpassing the combined market share of its closest competitors [7]. - The company reported a revenue of 6.078 billion yuan in Q3 2025, reflecting a year-on-year growth of 10.02%, and a net profit increase of 202.40% [23]. Group 3: Strategic Vision and Future Outlook - iFLYTEK's vision of "AI that understands you better" aims to redefine the relationship between humans and AI, aligning with global trends in AI development [3][19]. - The company emphasizes the importance of self-controlled technology and the integration of hardware and software to enhance user experience [10][13]. - With the increasing adoption of generative AI, iFLYTEK is positioned to capture significant market opportunities in the coming years, as indicated by industry forecasts [23].

Core Viewpoint - The A-share market experienced a strong rebound on November 13, particularly in the hard technology sector, with the Robot ETF Fund (159213) rising over 1% and attracting significant net inflows of 9 million yuan on the same day, following a total of over 30 million yuan in net inflows over the previous five days [1][3]. Group 1: Market Performance - The Robot ETF Fund (159213) ended the day with a gain of over 1%, marking the end of a three-day decline [1]. - The fund has seen continuous strong capital inflow, with a total of over 30 million yuan in net inflows over the last five days [1]. - Key component stocks of the Robot ETF Fund showed positive performance, with several stocks, including Dahua Technology and Zhongkong Technology, rising by over 2% and 1% respectively [3]. Group 2: Component Stocks - The top ten component stocks of the Robot ETF Fund include: - Huichuan Technology (300124) with a gain of 1.26% and an estimated weight of 10.22% - Keda Xunfei (002230) with a gain of 0.63% and an estimated weight of 9.25% - Dahua Technology (002236) with a gain of 2.20% and an estimated weight of 5.26% [4]. Group 3: Industry Developments - Yushu Technology launched a full-stack solution for humanoid robot data collection training, featuring the G1-D robot, which is designed for various applications [5]. - UBTECH announced the mass production of its Walker S2 humanoid robots, with a delivery target of 500 units this year, contributing to a cumulative order amount exceeding 800 million yuan for the Walker series [5]. - Industry experts predict that 2026 will mark a significant turning point for the humanoid robot industry, akin to the "iPhone moment," leading to large-scale production and benefiting companies across the supply chain [6]. Group 4: Future Outlook - The robot sector is expected to enter a phase of validation and consolidation after significant adjustments in October, with key developments from companies like Tesla and others supporting market expectations [7]. - Domestic robot companies are accelerating their capitalizations, with notable IPO plans from Yushu Technology and Leju Robotics, indicating a critical moment for the industry [8]. - Major tech players are intensifying their focus on humanoid robots, with expectations for significant production increases by 2026, potentially reaching a shipment volume of "ten thousand units" for domestic robots [8].