Financial Performance - In 2024, the company achieved total operating revenue of 1,505.77 million yuan, a decrease of 0.63% year-on-year[6] - The operating profit improved to -224.42 million yuan, an increase of 35.30% compared to the previous year[6] - The total profit was -228.94 million yuan, reflecting a growth of 33.78% year-on-year[6] - The net profit attributable to the parent company was -237.45 million yuan, a year-on-year increase of 30.72%[6] - The basic earnings per share improved to -0.52 yuan, a 31.04% increase compared to the previous year[8] Profitability and Efficiency - The gross profit margin increased from 11.11% to 20.74%, representing a rise of 9.63 percentage points[8] - The company reduced sales, management, and R&D expenses by 25.83 million yuan compared to the previous year, enhancing operational efficiency[8] Assets and Equity - Total assets at the end of the reporting period were 3,373.19 million yuan, a decrease of 11.28% from the beginning of the period[5] - The equity attributable to the parent company was 2,390.53 million yuan, down 9.15% from the beginning of the period[5] Strategic Focus - The company focused on high-quality development strategies, optimizing revenue structure and enhancing profitability through investments in AIGC and data center operations[8]

Summary of Conference Call Records Industry and Company Involved - The conference call primarily discusses the advancements and implications of **DeepSeek**, a new AI technology, within the **Chinese technology and AI industry**. It highlights the impact on various sectors, including cloud computing and chip manufacturing. Core Points and Arguments 1. **DeepSeek's Impact on China's Tech Position** DeepSeek has significantly enhanced China's standing in the global tech arena, potentially increasing the valuation of companies like Alibaba from 60 to 80 points, which will affect valuation systems across US, Hong Kong, and A-shares [2][4]. 2. **Open Source Model Benefits** The open-source nature of DeepSeek allows companies, especially state-owned enterprises and government units, to deploy AI solutions internally, ensuring data security and reducing costs, thus broadening AI application [2][5]. 3. **Acceleration of AI Application** DeepSeek's characteristics, such as low cost and private deployment, are accelerating the adoption of AI applications, with listed companies already conducting gray tests [3][5]. 4. **Chip Industry Growth** The domestic chip industry is on the verge of a significant breakthrough, with companies like Huawei and MiniMax actively adapting to large models, indicating a robust growth trajectory [3][7]. 5. **Cloud Computing Opportunities** Cloud computing companies are expanding their services to meet the rising demand for AI large models, providing distributed computing APIs and localized training services for financial institutions and government entities [3][9]. 6. **Shift in Demand for AI Resources** The demand for AI inference is expected to surpass training needs, with projections indicating that by 2025, the ratio of inference to training will be 9:1, highlighting a shift in resource allocation [3][10]. 7. **Client Demand Dynamics** There is a growing demand from three main client groups: enterprises replacing OpenAI APIs to cut costs, local governments seeking to optimize their resources, and enterprises requiring integrated AI solutions [11][10]. 8. **DeepSeek R1's Advantages** DeepSeek R1 excels in data organization, intelligence, and local deployment, significantly enhancing client willingness to adopt and integrate the technology into their operations [10][13]. 9. **AI Technology Acceptance in Enterprises** Acceptance of AI technology varies among different roles within companies, with technical staff more inclined to adopt AI, while mid-level employees may resist due to perceived job security concerns [15]. 10. **Future of AI Business Models** The AI business model landscape is still evolving, with a need for further exploration of profitability and operational strategies as the market matures [21][22]. Other Important but Overlooked Content 1. **Competition in Cloud Services** Future competitiveness in cloud services will hinge on capabilities related to large models, including deployment, training, and service capabilities, rather than just infrastructure [18][20]. 2. **Long-term Profitability Concerns** While short-term profitability in the cloud computing sector is expected to rise, long-term sustainability may be challenged by intense competition and unclear business models [20][23]. 3. **Emerging Players in AI** The AI landscape may see new entrants alongside established players like Peet and DeepMind, indicating a dynamic and competitive environment [21]. 4. **Open Source's Role in Leveling the Playing Field** Open source is diminishing the advantages of large corporations, allowing smaller firms to leverage cloud resources effectively, emphasizing the importance of neutrality and service quality [22]. 5. **Trends in the Chinese AI Market** The Chinese AI market is witnessing widespread adoption across various industries, mirroring trends seen in the US, which is a positive development for the sector [24].

Summary of Conference Call Notes Company and Industry Involved - **Company**: 优刻得 (UCloud) - **Industry**: AI and Cloud Computing, specifically focusing on domestic chip development and large model deployment Core Points and Arguments - **Investment Opportunities from DeepCache**: The emergence of DeepCache presents significant investment opportunities for domestic chips, driven by accelerated technological iterations, increased cloud computing demand, and rapid development of domestic chip R&D and deployment [3][4] - **Collaboration with Huawei**: The company is deepening its collaboration with Huawei to optimize large model operation efficiency by utilizing idle computing card resources from local state-owned assets [3][5] - **Local Data Center Development**: There is a surge in self-built data centers across regions, with various types of computing cards being utilized, including NVIDIA and Huawei cards, addressing challenges such as low utilization of idle computing power [3][6] - **Trend of Localized Deployment**: Localized deployment is becoming a trend, enhancing data security and processing efficiency by integrating large models with local government data [3][11] - **Advantages of Domestic Inference Cards**: Domestic inference cards have advantages in technical thresholds, memory adjustments, and cost control, particularly the H20 card, which offers high cost-performance ratio [3][10] - **Private Deployment Trends**: Enterprises are increasingly adopting private deployment and integrated machine models to ensure data security and service reliability [3][12] - **Rapid Adoption of Large Models**: The rapid adoption of large models in China is attributed to the emergence of domestic models, reduced inference costs, and the integration of open-source models with industry data [3][13] - **Support from Tencent**: Tencent is actively supporting the domestic large model deployment by ordering a significant number of computing cards for large-scale applications [3][14] - **Globalization Opportunities**: Chinese companies are exploring globalization opportunities, leveraging local sovereignty AI and providing low-cost services similar to OpenAI [3][15] - **GPU Card Utilization Potential**: There is potential for increased GPU card usage and profit margins by extending usage hours beyond typical operational times [3][16] - **Cloud Computing Demand**: There is a noticeable increase in cloud computing demand, with the company facing GPU resource shortages [3][17] - **New Business Models**: The company has successfully implemented new business models, including international projects utilizing GPU management software [3][30] Other Important but Possibly Overlooked Content - **Challenges in Local Government and State-Owned Enterprises**: Local governments and state-owned enterprises face challenges in data security and local deployment, which the company aims to address through proprietary cloud solutions [3][7] - **Data Center Expansion Plans**: The company plans to expand its self-owned resources and data centers in Inner Mongolia and Shanghai, with the latter nearing operational readiness [3][33] - **Liquid Cooling Technology**: The application of liquid cooling technology is still limited in China, with future adoption dependent on market demand and technological maturity [3][35] - **Collaboration with Local Governments**: The company is exploring partnerships with local governments to enhance resource utilization and promote local industry development [3][37] - **Cost-Effectiveness of Low-Power IDC Cabinets**: Transforming low-power IDC cabinets in first-tier cities may be cost-prohibitive, suggesting a shift to second-tier cities for deployment [3][38]

证券代码：688158 证券简称：优刻得 公告编号：2025-004 优刻得科技股份有限公司 关于股票交易异常波动暨严重异常波动公告 本公司董事会及全体董事保证本公告内容不存在任何虚假记载、误导性陈 述或者重大遗漏，并对其内容的真实性、准确性和完整性依法承担法律责任。 重要内容提示： 优刻得科技股份有限公司（以下简称"公司"）股票于 2 月 12 日、2 月 13 日、2 月 14 日连续 3 个交易日内收盘价格涨幅偏离值累计超过 30%。根据 《上海证券交易所科创板股票上市规则》、《上海证券交易所交易规则》、《上海 证券交易所科创板股票异常交易实时监控细则》的有关规定，属于股票交易异 常波动情形。 公司股票连续 10 个交易日内出现 3 次同向异常波动情形，连续 30 个 交易日内收盘价格涨幅偏离值累计超过 200%。根据《上海证券交易所科创板股 票上市规则》《上海证券交易所交易规则》《上海证券交易所科创板股票异常交 易实时监控细则》的有关规定，属于股票交易严重异常波动情形。 经公司自查，并书面征询公司控股股东、实际控制人及其一致行动人， 截至本公告披露日，公司控股股东、实际控制人及其一致行动人不存在影响公 ...

1、截至目前，作为贵公司的控股股东、实际控制人及其一 致行动人，不存在影响公司股票异常波动的重大事项，不存在应 披露而未披露的重大信息，包括但不限于并购重组、股份发行、 债务重组、业务重组、资产剥离和资产注入等重大事项。 《关于优刻得科技股份有限公司股票交易 异常波动暨严重异常波动的询证函》的回函 致：优刻得科技股份有限公司： 贵司发来的《关于优刻得科技股份有限公司股票交易异常波 动暨严重异常波动的询证函》，经认真自查，现就有关情况回复 如下： 控股股东/实际控制人及其一致行动人： | 季昕华 | 莫显峰 | 华 | 琨 | | --- | --- | --- | --- | | （季昕华） | （莫显峰） | （华 | 琨） | 2、本人及其一致行动人在贵公司本次股票异常波动暨严重 异常波动期间，不存在买卖贵公司股票的情况。 2025 年 2 月 14 日 特此回函。 以下无正文 （本页无正文,为控股股东、实际控制人及其一致行动人《<关于优刻得科技股份 有限公司股票交易异常波动暨严重异常波动的询证函>的回函》之签署页） ...

Core Viewpoint - On February 12, 2023, UCloud (688158) closed at 30.26 yuan, marking an increase of 8.46% with a trading volume of 36.60 billion yuan and a turnover rate of 31.01% [1] Trading Activity - The stock was listed on the daily trading leaderboard due to a turnover rate exceeding 30% [1] - The top five trading departments accounted for a total transaction of 4.29 billion yuan, with a buying amount of 2.04 billion yuan and a selling amount of 2.25 billion yuan, resulting in a net sell of 21.73 million yuan [1] - Among the top five buying departments, one institutional special seat was present, contributing 145.47 million yuan to the buying total of 1.89 billion yuan [1] - The leading selling department was Shenwan Hongyuan Securities Co., Ltd., with a selling amount of 82.64 million yuan [1] Fund Flow - The stock experienced a net outflow of 21.67 million yuan in main funds throughout the day [1] Historical Performance - Over the past six months, the stock has appeared on the trading leaderboard six times, with an average price increase of 13.14% the day after being listed and an average increase of 34.96% in the following five days [1]

Summary of Conference Call Notes Company and Industry Overview - The conference call primarily discusses the developments and strategies of **Youke De** in the **cloud computing** and **AI** sectors, particularly focusing on the implications of large model technologies and the demand for computing power. Key Points and Arguments Industry Dynamics - The price increase of Deepseek's official API and the limitation on new user registrations indicate a significant rise in user demand, both domestically and internationally, highlighting a robust market need for Deepseek applications and APIs [3] - DPC's limited commercialization efforts present substantial opportunities for cloud computing companies, as many enterprises are opting to utilize cloud services rather than expanding their server capacities [4][5] - There are three main customer demands for large model machines: public cloud services, internal private/exclusive models, and government/state-owned enterprise exclusive inference clusters, all of which are rapidly growing [6] Company Strategy and Focus - Starting in 2024, the company will focus on inference business due to its long duration, rapid growth, and strong multi-tenant concurrency capabilities [5] - The company plans to invest heavily in inference cards by 2025 to prepare for the upcoming inference era [5] - The compatibility of domestic cards has improved, creating new opportunities, although performance optimization and cost reduction are still necessary [5][16] Capacity and Resource Management - The company's AI production capacity utilization is nearly 100%, and it is accelerating capacity deployment through equipment procurement, resource leasing, and cooperative operations [5][13] - The company has over 30 data centers globally and plans to launch an overseas version of Deep Connect to support international clients [19] - The company has received significant revenue from collaborations with major model manufacturers, enhancing its inference service capabilities [5][25] Market Trends and Future Outlook - The AI computing power demand is expected to grow significantly, with the company preparing to meet this demand by purchasing inference chips [7][13] - The company anticipates adjusting its revenue targets due to unexpectedly high demand [14] - The cloud computing industry is viewed positively, with expectations of growth driven by advancements in large model technologies and GPU business [20] Challenges and Considerations - The company faces challenges in talent retention and acquiring compliant training and inference resources, which are crucial for model knowledge and cost control [9] - The company is also addressing API response time issues caused by high user volume and concurrent requests, emphasizing the need for system optimization [28] Customer Engagement and Service Offerings - The company has received inquiries from various sectors, including finance and startups, focusing on internal deployment of large models and API usage [27] - The company provides private cloud and private deployment services, with deployment times varying based on customer needs [33][36] Financial Performance and Profitability - The company operates under a token-based charging model in the inference era, with profitability influenced by the global user base [39] - In the private cloud model, the company profits from hardware sales, deployment services, and model tuning, with higher profit margins compared to general services [40] Technological Developments - The development of large models and GPUs is expected to enhance the competitiveness of Chinese technology companies on a global scale, improving their valuations [38] - The emergence of Tensor Processing Clusters (TPC) allows cloud computing companies to leverage open-source models for a broader customer base [38] Additional Important Insights - The company is actively working on optimizing the performance of domestic cards to improve their competitiveness against NVIDIA cards, which currently outperform them significantly [37] - The company is committed to expanding its global footprint, with a notable presence in various countries and plans to enhance its service offerings for international clients [19]

证券代码：688158 证券简称：优刻得 公告编号：2025-003 优刻得科技股份有限公司 关于股票交易异常波动暨 严重异常波动公告 本公司董事会及全体董事保证本公告内容不存在任何虚假记载、误导性陈 述或者重大遗漏，并对其内容的真实性、准确性和完整性依法承担法律责任。 重要内容提示： 优刻得科技股份有限公司（以下简称"公司"）股票于 2025 年 2 月 7 日、2 月 10 日连续 3 个交易日内收盘价格涨幅偏离值累计超过 30%，根据《上 海证券交易所科创板股票上市规则》、《上海证券交易所交易规则》、《上海证券 交易所科创板股票异常交易实时监控细则》的有关规定，属于股票交易异常波 动情形。 公司股票连续 10 个交易日内（2025 年 1 月 24 日至 2025 年 2 月 10 日） 收盘价格涨幅偏离值累计超过 100%。根据《上海证券交易所科创板股票上市规 则》《上海证券交易所交易规则》《上海证券交易所科创板股票异常交易实时监 控细则》的有关规定，属于股票交易严重异常波动情形。 经公司自查，并书面征询公司控股股东、实际控制人及其一致行动人， 截至本公告披露日，公司控股股东、实际控制人及其一致行动人 ...

《关于优刻得科技股份有限公司股票交易 异常波动暨严重异常波动的询证函》的回函 控股股东/实际控制人及其一致行动人： 致：优刻得科技股份有限公司： | 季昕华 | 莫显峰 | 华 | 琨 | | --- | --- | --- | --- | | （季昕华） | （莫显峰） | （华 | 琨） | 贵司发来的《关于优刻得科技股份有限公司股票交易异常波 动暨严重异常波动的询证函》，经认真自查，现就有关情况回复 如下： 2、本人及其一致行动人在贵公司本次股票异常波动暨严重 异常波动期间，不存在买卖贵公司股票的情况。 1、截至目前，作为贵公司的控股股东、实际控制人及其一 致行动人，不存在影响公司股票异常波动的重大事项，不存在应 披露而未披露的重大信息，包括但不限于并购重组、股份发行、 债务重组、业务重组、资产剥离和资产注入等重大事项。 特此回函。 以下无正文 （本页无正文,为控股股东、实际控制人及其一致行动人《<关于优刻得科技股份 有限公司股票交易异常波动暨严重异常波动的询证函>的回函》之签署页） 2025 年 2 月 10 日 ...

重要内容提示： 优刻得科技股份有限公司 关于股票交易异常波动公告 本公司董事会及全体董事保证本公告内容不存在任何虚假记载、误导性陈 述或者重大遗漏，并对其内容的真实性、准确性和完整性依法承担法律责任。 证券代码：688158 证券简称：优刻得 公告编号：2025-002 优刻得科技股份有限公司（以下简称"公司"）股票于 2025 年 2 月 5 日、2 月 6 日连续 2 个交易日收盘价格涨幅偏离值累计超过 30%，根据《上海 证券交易所科创板股票上市规则》、《上海证券交易所交易规则》、《上海证券交 易所科创板股票异常交易实时监控细则》的有关规定，属于股票交易异常波动 情形。 经公司自查，并书面征询公司控股股东、实际控制人及其一致行动人， 截至本公告披露日，公司控股股东、实际控制人及其一致行动人不存在影响公 司股票交易异常波动的重大事项；不存在其他应披露而未披露的重大信息。公 司目前生产经营正常，生产经营未发生重大变化。 DeepSeek 为开源大模型，各家厂商均可根据自身需求对其进行模型的 适配与本地化部署，任何用户可免费基于该模型开展相关业务。 公司近期也与 DeepSeek 进行了全系列模型适配工作，目 ...