最新数据显示，基于百度数字人技术已生产超过10万个数字人，应用于电商、教育、法律等各行各业， 帮助降低开播成本达80%，提升直播转化率31%。在6月15日百度优选罗永浩数字人首场直播中，运用 该技术打造的罗永浩与朱萧木数字人，成功完成长达6小时的超拟真直播，观播人数达1300万，GMV突 破5500万。 目前，该技术已在百度电商场景的明星、图书和健康直播上落地应用，帮助众多商家拓宽了销售渠道， 降低直播成本，提高运营效率。数字人可以24小时不间断直播，增加了商品的曝光时间和销售机会，进 一步提升了经济效益。 公开资料显示，自2016年以来，百度已凭借百度大脑、小度助手（DuerOS）、Apollo、飞桨、知识增 强大语言模型、文心智能体技术、剧本驱动多模协同的高拟真数字人技术等创新成果七次获得世界互联 网大会领先科技奖。 11月6日，2025世界互联网大会领先科技奖正式揭晓，百度"剧本驱动多模协同的高拟真数字人技术"获 奖，这也是百度大模型继2023年"知识增强大语言模型关键技术"和2024年"文心智能体技术"之后，连续 三年获得该奖项，百度也成为唯一连续三年获奖的人工智能公司。 据了解，世界互联网大会领先科 ...

来源：市场资讯 （来源：V观财报） "家人们，这个链接已经断货了！"你可能已经注意到，直播间里有些主播不间断直播，脸上却丝毫不见 疲惫。 这些24小时不断播的主播便是数字人。除了电商直播间外，你可能还接到过数字人打来的推销电话，也 向数字人客服反映过售后问题。 数字人应用越发广泛背后，有公司已经凭借数字人的商业化应用年入数亿。近期，南京硅基智能科技集 团股份有限公司（下称硅基智能），正式向港交所递交招股说明书。 8万"数字人"去年 这8万"数字员工"赚钱能力怎么样？根据财务数据，2022年到2024年及2025年上半年，公司分别收入2.2 亿元、5.3亿元、6.6亿元、3.3亿元，毛利率分别为38.5%、45.8%、34.3%、31.6%。同期，公司分别亏 损1.1亿元、1.0亿元、1.1亿元、829.0万元，3年半累计亏损超3亿元。经调整（加回开支、以权益结算以 股份为基础的付款开支、赎回负债之增值）后，2022年至2024年分别亏损4622.1万元、2941.2万元、 3524.2万元，2025年上半年扭亏为盈，净利为528.9万元。 招股书援引灼识咨询的资料称，根据2024年提供数字人智能体解决方案产生 ...

Core Insights - The digital human guide "Xiao Fei" was officially launched at the iFlytek 1024 Developer Festival on November 6, showcasing advanced multimodal interaction capabilities [1] - "Xiao Fei" surpasses simple Q&A limitations, enabling free dialogue among multiple users and multilingual communication [1] - The digital guide possesses personalized memory capabilities, allowing it to remember visitor history and provide thoughtful reminders [1]

Core Insights - Silicon-based Intelligence has submitted its prospectus to the Hong Kong Stock Exchange, aiming for an IPO as a leading player in the AI sector in China, with a revenue of 655 million yuan in 2024, capturing 32.2% of the domestic digital human market [2][4]. Group 1: Company Overview - Silicon-based Intelligence is the largest digital human company in China and ranks second globally, having delivered over 80,000 "Silicon-based labor forces" across various industries including telecommunications, finance, healthcare, education, and public services [4]. - The company’s revenue has grown significantly from 228 million yuan in 2022 to 655 million yuan in 2024, marking a 187% increase over two years [5]. Group 2: Revenue Sources - A substantial portion of the revenue growth is attributed to a single major client, China Unicom, which accounted for 90% of the revenue increase from 2022 to 2024, with its contribution rising from 36.9 million yuan in 2022 to 422 million yuan in 2024 [6][7]. - By the end of 2024, China Unicom is expected to represent 64.4% of Silicon-based Intelligence's total revenue [7]. Group 3: Client Strategy and Challenges - The company initially attempted to expand its client base by introducing a distribution model, increasing the number of distributors from 2 to 251 in 2023, but later reduced this number to 157 by the end of 2024 due to low retention rates among small clients [8][10]. - The average revenue per client, excluding China Unicom, was approximately 350,000 yuan, while the customer acquisition cost was 184,000 yuan, indicating a challenging profitability scenario [10][12]. Group 4: Market Dynamics - The company faces structural challenges in the AI B2B market, where small enterprises lack budgets for AI products, and large enterprises demand extensive customization, leading to compressed profit margins [17][18]. - The reliance on a few large clients limits pricing power and creates vulnerability in revenue streams, as evidenced by the declining gross margin from 45.8% in 2023 to 31.6% in the first half of 2025 [15][14]. Group 5: Broader Industry Implications - The situation of Silicon-based Intelligence reflects broader issues in the AI industry, where companies struggle to balance between serving small clients with high acquisition costs and large clients with demanding customization needs [18][19]. - The challenges faced by Silicon-based Intelligence highlight the difficulties in achieving sustainable profitability in the rapidly evolving AI landscape, despite being positioned in a high-growth sector [16][19].

近日，国内最大数字人智能体提供商——硅基智能正式递表港交所，开启赴港上市之路。 招股书显示，根据2024年提供数字人智能体解决方案产生之收入而言，硅基智能在中国所有数字人智能 体提供商中排名第一，在中国数字人智能体行业中拥有32.2%的市场份额，处于领导地位。而在全球数 字人智能体提供商中，硅基智能位列第二。 值得注意的是，硅基智能今年曾涉及一则"全员裁员"的传闻，不过随后硅基智能在8月3日发布了一份声 明，称"目前拥有一支稳定的产研与销售团队，且持续在全球范围内扩大招聘规模。2025年，我们将重 点布局杭州、嘉兴、香港等地，预计全年新增岗位数百个，2026年将达到新增数千人的扩张节奏"。 目前，证券时报记者发现，硅基智能微信公众号已删除上述声明。 严重依赖单一客户 自2017年成立以来，硅基智能始终站在人工智能技术创新的前沿，不断推动数字人智能体的职场能力边 界。硅基智能认为AI不仅是一种工具，它还可被视为一种新型的劳动力。硅基智能把这种数字劳动力 创新性地起名叫做"硅基劳动力"，使其区别于"碳基"人类劳动力。 招股书显示，硅基智能专注于数字人智能体以及拟人多模态智能交互技术之研发与商业化应用，技术范 畴涵 ...

进校园：讲授新生开学支付"第一课" 在山东大学校园里，中国银行济南历城支行为新生带来了一场别开生面的"支付第一课"。工作人员用通俗易懂的 语言结合生动案例，讲解数字人民币的概念、核心特点及安全使用要点。现场设置互动体验区，指导新生下载数 字人民币APP、开通个人钱包，并完整演示支付操作流程。"这是我第一次接触数字人民币，没想到这么方便快 捷！"一位刚完成支付体验的新生兴奋地表示。 与此同时，济南历下支行营业部的"校园卡办理+反诈宣传"专项行动也在火热进行。工作人员不仅为学生办理业 务，更通过真实案例揭露"校园贷""刷单返利"等骗局，引导广大学子安全使用网络、保护个人信息，增强反诈意 识，切实筑牢青少年金融安全"防护网"。 央广网济南11月4日消息（记者程立龙）数字人民币正悄然改变我们的支付习惯。作为山东省内数字人民币试点地 区分行，山东中行以"金融为民"为初心，聚焦场景化服务，开展数字人民币"进企业、进校园、进商圈、进社 区"专项宣传活动，助力数字金融知识走进百姓日常生活。 进企业：筑牢劳动者"安心支付"防线 中国银行济南起步区支行深入当地建筑公司，针对农民工群体集中、项目分布广的特点，以工资代发为切入点， ...

"数字人刘邦"跨越两千年"现身"第三届汉文化论坛 中新网徐州11月5日电 (唐娟 朱志庚)11月5日，第三届汉文化论坛在江苏徐州开幕。现场长约5分钟的开 场短片上演了一场跨越两千年的"相见"视听盛宴。通过 AI新技术赋能，汉高祖刘邦——不再是古籍中 静态的文字或后世想象中模糊的画像，而是以一个栩栩如生的数字人形象，穿越时空，昂然"出现"在众 人面前。 中新经纬版权所有，未经书面授权，任何单位及个人不得转载、摘编或以其它方式使用。 关注中新经纬微信公众号(微信搜索"中新经纬"或"jwview")，看更多精彩财经资讯。 第三届汉文化论坛现场，"数字人刘邦"与众人"相见"。 朱志庚 摄 "飞骑"铭陶骑马俑奔腾，陶跪坐甲胄俑枕戈待旦，陶绕襟衣女舞俑甩动长袖……来自徐州博物馆的西汉 陶俑组，也成为5分钟开场片叙述的焦点。(完) 来源：中国新闻网 编辑：熊思怡 广告等商务合作，请点击这里 本文为转载内容，授权事宜请联系原著作权人 这不仅是一次信息技术的集中展示，更是经过全国10多位权威史学、考古学、美术史专家的再三考证与 严格把关，利用人工智能，对这位汉文明塑造者进行的一次最接近真实历史的容貌复原。 关于刘邦的长相，史料 ...

Core Insights - The rapid development of artificial intelligence has led to the emergence of AI digital humans, creating a competitive and growing market [2] - Nanjing Silicon-based Intelligent Technology Group Co., Ltd. (Silicon Intelligence) has initiated its IPO process on the Hong Kong Stock Exchange, aiming to raise funds for enhancing R&D capabilities and marketing [2] - Silicon Intelligence has completed eight rounds of financing, with a valuation of 3.15 billion RMB after its D round in June 2023 [2][3] Company Positioning - As the largest provider of digital human intelligent agents in China, Silicon Intelligence holds a significant market share of 32.2% in the digital human intelligent agent solutions sector [3] - The company utilizes three core technologies: voice cloning, appearance cloning, and thought cloning, to create "Silicon-based Labor" [3] - Silicon Intelligence has delivered over 80,000 solutions across various industries, including telecommunications, finance, healthcare, and education [5] Financial Performance - The company has experienced continuous losses over the past three years, with net losses of 111 million RMB in 2022, 95.9 million RMB in 2023, and 112 million RMB in 2024 [8] - Revenue growth has slowed significantly, with a revenue of 32.6 million RMB in the first half of 2025, reflecting a year-on-year growth rate of only 11.15% [7][8] Market Dynamics - The global market for digital human intelligent agents is projected to grow from 1.3 billion RMB in 2021 to 4.1 billion RMB in 2024, with a compound annual growth rate (CAGR) of 44.7% [7] - Competition in the AI digital human sector is intensifying, with major players like Baidu, Tencent, and Alibaba entering the market, increasing pressure on Silicon Intelligence [9] - The company faces challenges in customer acquisition, with the average customer acquisition cost rising significantly, reaching 184,000 RMB in 2024 [10][11] Customer Dependency - Silicon Intelligence's revenue is highly concentrated, with the top five customers accounting for 87.5% of total revenue in the first half of 2025, and the largest customer contributing 64.4% [6] - The company has seen a decline in new customer acquisition, with the number of new customers dropping from 890 in 2023 to 461 in 2024 [10]

Core Insights - The article highlights the advancements made by Yunzhisheng in the field of AGI technology, particularly in four key areas: voice technology, medical large models, edge AI, and digital human technology, showcasing China's strong capabilities in artificial intelligence [1] Group 1: Medical Large Model - The "multi-modal medical vertical large model" integrates diverse medical data, achieving over 90% accuracy in liver focal lesion detection and significantly outperforming human averages in clinical exams, thus supporting the intelligent transformation of the healthcare industry [2][3] Group 2: Edge AI and Digital Human Technology - The "refined small model on chip" technology optimizes AI model performance on edge chips, reducing resource consumption by over 100 times while maintaining excellent interaction quality, addressing traditional challenges in edge model performance [6] - The "multi-modal emotional intelligent digital human" technology creates high-fidelity digital humans capable of natural behavior and emotional expression, applicable in various sectors such as customer service and education [6] Group 3: Comprehensive AI Industry Strategy - Yunzhisheng has established a four-layer technical architecture that connects technology, scenarios, and data, enhancing its core competitiveness in the AI industry and facilitating the development of specialized digital experts across various verticals [7] - The breakthroughs in four internationally leading technologies strengthen the company's foundational technology and ensure a complete link from research and development to implementation, aiming to accelerate the commercialization of cutting-edge innovations [7]

Core Insights - The development of digital RMB is transitioning from pilot exploration to systematic improvement, with the People's Bank of China (PBOC) emphasizing the need for steady development and optimization of the digital RMB management system [1][2] Group 1: Limitations of M0 Positioning - The non-interest-bearing nature of digital RMB reduces the willingness of individuals and enterprises to hold it, as it lacks the value appreciation feature that traditional bank accounts offer [2] - The disconnection between digital RMB and the banking account system complicates user experience, hindering its potential to enhance monetary policy transmission and financial regulation [2][3] - The single-functionality of digital RMB limits its usage scenarios, as it currently only supports payment functions, unlike third-party platforms that offer diverse services [2] Group 2: Need for M1 Transition - There is a growing consensus on the necessity to transition digital RMB from M0 to M1 to address its limitations, as M1 would allow it to be recognized as a liability of commercial banks, thus integrating it into their balance sheets [4][5] - Two proposed pathways for this transition include digitizing deposits to create a seamless connection between bank accounts and digital wallets, and introducing a new savings form that allows funds to be transferred to digital wallets while still being counted as bank liabilities [4][5] Group 3: Enhancing Bank Incentives - The current 100% reserve requirement for digital RMB limits banks' ability to profit from traditional lending spreads, reducing their motivation to promote digital RMB [6] - Experts suggest regulatory reforms to unify technical standards, lower development costs for banks, and allow digital RMB to be classified as a bank liability to enable lending and interest income [6][7] - Banks are encouraged to view digital RMB as an opportunity for customer acquisition and service development, creating high-frequency payment scenarios and leveraging transaction data for targeted marketing and risk management [7]