证券研究报告 传媒 行业定期报告 hyzqdatemark 2026 年 01 月 11 日 证券分析师 陈良栋 SAC：S1350524100003 chenliangdong@huayuanstock.com 陈逸凡 SAC：S1350525040003 chenyifan@huayuanstock.com 周政宇 SAC：S1350525050006 zhouzhengyu@huayuanstock.com 许孟婕 SAC：S1350525120004 xumengjie@huayuanstock.com 魏桢 weizhen@huayuanstock.com 板块表现： AI 应用仍是核心，重视 AI 营销 GEO 模式 投资评级： 看好（维持） ——传媒互联网行业周报（2026.1.5-2026.1.11） 投资要点： 电影/电视剧方向：《飞驰人生 3》正式定档 2026 年春节，或将随着春节档演绎。 我们建议关注相关电影出品/发行方以及院线公司：包括中国电影、猫眼娱乐、大麦 娱乐、上海电影、幸福蓝海、横店影视、金逸影视等。同时，电视剧行业方面，国 家广播电视总局印发实施《进一步丰富电视大屏内容 促 ...

针对外卖行业的无序"内卷"，监管部门再度出手。 国家市场监督管理总局（以下简称"市场监管总局"）发文称，国务院反垄断反不正当竞争委员会办公室 （以下简称"委员会办公室"）将对外卖平台服务行业市场竞争状况开展调查、评估。 北京商报综合报道 淘宝闪购方面表示，欢迎并将积极配合此次行业市场竞争状况调查、评估工作，严格落实合规主体责 任，持续与商户、生态合作伙伴等各方携手，提供更多元化、更高质量的服务，共同维护好公平有序的 市场环境，合力推动外卖服务行业进一步繁荣发展。 京东方面则表示，坚决支持并欢迎这一决定，高度赞同反内卷、维护市场公平竞争秩序、保障消费者与 经营者、骑手合法权益的措施。 本次并非相关监管部门首次出手整治外卖行业"内卷"。2025年，国内外卖行业上演补贴大战，在各平台 竞争升级的同时，相关监管部门也及时约谈相关方。2025年4月，京东上线"百亿补贴"；5月，饿了么淘 宝上线"零元免单"。随后，市场监管总局发文表示，针对当前外卖行业竞争中存在的突出问题，约谈京 东、美团、饿了么等平台企业。要求相关平台企业严格落实主体责任，主动履行社会责任，加强内部管 理，合法规范经营，公平有序竞争，共同营造良好市场环 ...

上市公司研究平台，专注价值发现、创造与传播 本文来自微信公众号： 锦缎 ，作者：张生 以下文章来源于锦缎 ，作者张生 锦缎 . 2025年12月中央经济工作会议中，锚定了未来经济工作的三大重点：扩内需、促创新（AI+）、 反"内卷"。这不仅是政策定调，更是清晰的监管行动指南：扩内需需防范通缩，拒绝低价消耗战；促 创新要求资源投向AI等未来增量；反内卷则直指存量市场的零和博弈。 然而风向之下，现实惯性仍在。1月8日，阿里巴巴面向机构投资者明确表态，淘宝闪购已确立其 2026年的核心战略：继续加大投入，目标达到市场绝对第一。 2024年末，我国网上外卖用户规模达5.92亿人，占网民总数的53.4%，渗透率持续提升并已覆盖过半 网民。在接近6亿用户的规模下，若排除年龄层与消费习惯的影响，基本意味着外卖市场已趋近饱 和。此时的市场竞争必将陷入存量博弈，而无法拉动市场整体增长。 图：网上外卖用户规模及使用率趋势，来源：中 商情报网 但监管的回应迅速而明确。 就在阿里此番表态的次日，箭在弦上的外卖市场新一轮内卷大戏，尚未正式打响便迎来转折：1月9 日下午，国务院反垄断反不正当竞争委员会办公室在答记者问时表示，将依据《中华 ...

Investment Rating - The report does not explicitly provide an investment rating for the AI industry Core Insights - The AI industry is experiencing significant advancements, with multiple companies releasing innovative products and models, indicating a robust growth trajectory in the sector [1][5][9][14][18] - The Chinese government is actively promoting AI integration in manufacturing, aiming to establish a leading position in AI technology and applications by 2027 [5] - Major companies like Lenovo and Tencent are launching new AI models and platforms, enhancing the capabilities and applications of AI in various fields [8][15][18] Summary by Sections 1. AI Industry Dynamics - The Ministry of Industry and Information Technology and other departments have issued guidelines for the "Artificial Intelligence + Manufacturing" initiative, targeting the development of core AI technologies and industry applications by 2027 [5] - In 2026, over 30 national standards related to data in AI will be introduced, focusing on areas like intelligent agents and embodied intelligence [6] - The Sichuan provincial government has launched a plan to develop a national digital economy innovation zone, aiming to enhance digital infrastructure and promote the growth of digital industries [7] 2. AI Application Insights - LuKe announced the global launch of the "Air Charging AI Smart Lock V7 Max" at CES 2026, featuring technology for continuous power supply [10] - Yushun released training videos for its humanoid robot Unitree H2, showcasing advanced capabilities [11] - DeepTing released the world's first dual-wheeled outdoor companion robot Rovar, designed for various outdoor scenarios [12] - Rokid introduced the "Style" AI glasses at CES 2026, featuring a dual-chip architecture and 4K video capabilities [13] 3. AI Large Model Insights - Alibaba Tongyi has released and open-sourced the Qwen3-VL-Embedding and Qwen3-VL-Reranker models, designed for multimodal information retrieval and understanding [14] - Tencent launched the HY-Motion1.0 model, a significant advancement in 3D character animation generation [15] - Xiaopeng unveiled its second-generation VLA model, marking a shift towards practical applications of physical AI technology [16][17] - Lenovo introduced the Lenovo Qira personal super intelligent agent, emphasizing cross-device functionality and personalized services [18] 4. Technology Frontiers - Alibaba Cloud launched a new multimodal interaction development kit, integrating various foundational models for diverse applications [19] - Tsinghua University published a paper on an AI-driven drug virtual screening platform, significantly improving screening speed and accuracy [20] - Chinese scientists made progress in cancer immunotherapy, developing a protein-targeting degradation technology [21] - A team from the University of Science and Technology of China achieved advancements in heterogeneous quantum communication networks [22]

2026 年 ， AI 正 在 加 速 走 向 物 理 世 界 。 大 洋 彼 岸 拉 斯 维 加 斯 刚 刚 结 束 的 2026 年 国 际 消 费 电 子 展 （CES）上，大会的核心议程转向了"物理AI"（Physical AI），英伟达创始人兼CEO黄仁勋在演讲中 提到，"物理AI的'ChatGPT时刻'已然到来。" 几乎同一时间，深圳举行的阿里云通义智能硬件展上，阿里云发布多模态交互开发套件，套件集成了大 语言、多模态、语音等三款基础大模型，并预置十多款Agent和MCP工具，可应用于AI眼镜、学习 机、陪伴玩具、智能机器人等硬件设备，国内许多热门AI智能硬件的背后都有千问。 AI硬件的登台时刻正在真正到来。来到2026年，伴随着能力的提升和场景的补足，AI硬件正从玩具性 的营销"噱头"蜕变为真正具备感知、理解与决策能力的"端侧智能体"，打造"AI落地的最后一公里"。 与此同时，大模型厂商竞逐物理AI领域的"落子"也早已开启，阿里千问正瞄准AI硬件的"安卓时刻"。 比如，AI正在帮人们听懂宠物语言。Pettichatx网易伏羲展出了Pettichat宠物翻译器，这是依托100 多万宠物声音样本训练 ...

本周观点 国务院对外卖平台开展调查、评估，OpenAI 押注 AI 医疗： 风险提示 后续政策不及预期风险；中美关系变化风险；内容上线及表现不及预期风险；宏观经济运行不及预期风险；AI 技 术迭代和应用不及预期风险；政策监管风险。 敬请参阅最后一页特别声明 1 咖啡茶饮：1）咖啡：高景气维持；行业具备β性红利，各品牌依旧积极开店，但淡季亦有所来临，警惕短期数 据波动风险。2）茶饮：略有承压；行业迎来淡季，尽管电商平台追求即时零售（含外卖）市占，补贴短期断崖 式退坡概率小，但国务院宣布对外卖平台开展调查、评估，补贴退坡依旧是趋势。 电商：持续承压。受国内消费环境影响，整体国内电商表现平淡且乏力。 流媒体平台：音乐流媒体平台为内需驱动的优质互联网资产，高性价比悦己消费，规模效应驱动盈利杠杆释放， 我们建议持续关注音乐订阅平台。 虚拟资产&资产交易平台：加密市场催化有限，币价仍波动。关注下周 1 月 15 日《Clarity 法案》审议结果。 汽车服务：中升集团公告在北京、乌鲁木齐、青岛等地开设 9 家全新 4S 店，持续关注汽车后市场板块。 020：互联网医疗：OpenAI 宣布推出"ChatGPT 健康"，建议 ...

Core Insights - The competition in large models has shifted from "Chat" to "Agent," focusing on executing complex tasks in real environments rather than just scoring on leaderboards. The industry anticipates 2026 as the year when commercial value will be realized, with a technological evolution towards verifiable reinforcement learning (RLVR) [2][4][5]. Group 1: Competition Landscape - The engineering challenges of the Chat era have largely been resolved, and future success will depend on the ability to complete complex, long-chain real tasks. The core value of AI is transitioning from "providing information" to "delivering productivity" [4]. - The bottleneck for Agents lies not in cognitive depth but in environmental feedback. Future training paradigms will shift from manual labeling to RLVR, enabling models to self-iterate in systems with clear right or wrong judgments [5][6]. - The industry consensus suggests that while China has a high chance of catching up in the old paradigm (engineering replication, local optimization, toC applications), its probability of leading in new paradigms (underlying architecture innovation, long-term memory) is likely below 20% due to significant differences in computational resource allocation [5][11]. Group 2: Strategic Opportunities - Opportunities for catching up lie in two variables: the global shift towards "intelligent efficiency" as scaling laws encounter diminishing returns, and the potential paradigm shift driven by academia around 2026 as computational conditions improve [5][19]. - The ultimate variable for success is not leaderboard scores but the tolerance for uncertainty. True advancement depends on the willingness to invest resources in uncertain but potentially transformative new paradigms rather than merely chasing scores in the old paradigm [5][10]. Group 3: Perspectives from Industry Leaders - Industry leaders express cautious optimism regarding China's potential to lead, with probabilities of success varying. For instance, Lin Junyang estimates a 20% chance of leading due to structural differences in computational resource allocation and usage [11][12]. - Tang Jie acknowledges the existing gap in enterprise AI lab research but bets on a paradigm shift occurring around 2026, driven by improved academic participation and the emergence of new algorithms and training paradigms [15][19]. - Yang Qiang believes that China may excel in toC applications first, drawing parallels to the internet history, while emphasizing the need for China to develop its own toB solutions to bridge existing gaps [20][24]. Group 4: Technological Innovations - The future of AI will require advancements in multi-modal capabilities, memory structures, and self-reflective abilities, which are essential for achieving higher levels of intelligence and functionality [68][70][73]. - The introduction of new optimization techniques, such as the MUON optimizer, aims to enhance token efficiency and long-context processing, which are critical for the performance of agent-based models [110][116]. - The development of linear attention mechanisms is expected to improve efficiency and performance in long-context tasks, addressing the limitations of traditional attention models [116]. Group 5: Future Directions - The industry is focused on distinguishing between scaling known paths through data and computational increases and exploring unknown paths to discover new paradigms [98][99]. - The potential for AI to participate in scientific research is anticipated to expand significantly, opening new possibilities for innovation and application [101].

Investment Rating - The report does not explicitly state an investment rating for the industry Core Insights - The competition in AI entry points is intensifying, with major companies increasing their investments. China's AI presence globally has significantly improved, with domestic large models continuously iterating. Despite GPT-5.2 and Gemini 3 Pro leading, Chinese models have effectively altered the North American dominance in the competitive landscape. In the global Top 10, three positions are held by Chinese models, and in the Top 15, there are six Chinese companies. By 2025, China's open-source AI model usage is expected to account for over 70% of the global market [2][11][19] - The demand for inference has surged, with the emergence of o1 class inference models unlocking approximately 10 times the potential of traditional models in terms of inference-time compute. The demand for computing power has shifted from being solely "training-driven" to a dual focus on "training + inference" [2][5][37] - The battle for entry points has evolved beyond mobile devices to OS-level intelligent agents and super apps. By December 24, 2025, ByteDance's AI application Doubao announced daily active users (DAU) exceeding 100 million, while Qianwen App reached over 30 million monthly active users within 23 days of public testing, becoming the fastest-growing AI application globally. Doubao bypasses traditional interfaces, creating an "AI operating system" that directly interacts with super apps like WeChat and Alipay, challenging the rules of the traditional app era [2][44][45] Summary by Sections AI Entry Point Competition - China's AI global presence has significantly improved, with domestic large models continuously iterating. In the global Top 10, three positions are held by Chinese models, and in the Top 15, there are six Chinese companies. By 2025, China's open-source AI model usage is expected to account for over 70% of the global market [2][11][19] - The competition for entry points has evolved beyond mobile devices to OS-level intelligent agents and super apps, with significant user engagement reported for new AI applications [2][44][45] Domestic Chip Breakthroughs - The smart computing center in China is expanding, with a projected compound annual growth rate (CAGR) of 57% from 2020 to 2028, reaching 2,781.9 EFLOPS by 2028. Domestic chip technology is steadily improving, with local cloud service providers accelerating the construction of heterogeneous environments [5][50] - Domestic general-purpose GPUs are upgrading from "usable" to "good," with performance metrics approaching those of leading international models. The production capacity of domestic chip manufacturers like SMIC is continuously increasing, providing solid support for domestic AI chip production [5][53][54] Supply and Demand Dynamics - The demand side is characterized by a surge in inference demand as AI applications become more prevalent, while the supply side sees continuous improvements in domestic GPU performance and accelerated adaptation by cloud service providers [5][59] - The AI server market is expected to see a shift towards inference servers becoming the mainstream, with a projected market size of approximately $39.3 billion in 2024, reflecting a year-on-year growth of 49.7% [5][64]

在科技公司忙着教输入法写诗、画画甚至预测未来时，用户只想准确地打字、表达自己的意思，这 个朴素的想法却成了一种奢望。 步入2026年，老牌霸主搜狗输入法迎来20岁生日。经济观察报获悉，腾讯内部正在酝酿着一场变 革，用AI把输入法重做一遍，这不仅是功能的优化，更是意在推动AI语音、AI打字、AI翻译等方面 全面升级。 然而，一个荒诞的反差正在发生：输入法试图越来越"聪明"地预测意图，却越来越"笨拙"地处理 用户的指令。 在各大社交平台和投诉渠道，有不少关于"输入法变笨"的讨论。用户吐槽，以前打首字母就能联 想出词组，现在全拼输入还可能出现完全不相关的词。例如，输入"nanyong（难用）"首先联想 出的是"男用"，"congyoudaozuo（从右到左）"第一个出现的是"葱油刀座"。词联想不准、语 音识别出错、广告弹窗过多……这些问题正在消耗用户对输入法的耐心。 词联想不准、语音识别出错、广告弹窗过多……这些问题正在 消耗用户对输入法的耐心。这并非大厂减少了对输入法应用的 投入。相反，在AI成为标配的时代，主流输入法们正卷入一 场"军备竞赛"…… 作者：钱玉娟 封图：本报资料库 这并非大厂减少了对输入法应用的投 ...

2026年新年伊始，国家新一轮"以旧换新"补贴政策全面落地，首批625亿元专项资金已拨付到位。从家电到数码，从汽车到智能家居，这场覆盖全国的消费 盛宴让普通人直呼"真香"。但如何精准领取补贴？哪些"坑"需要避开？记者走访多位消费者、商家及行业专家，为您奉上这份实用指南。 一、家电换新：一级能效成"硬门槛"，最高省9000元 "旧冰箱当了废品卖只能拿200元，用国补换新直接省了1500元！"北京朝阳区的李女士刚通过京东APP完成家电置换。她购买的65英寸一级能效电视，原价 8999元，叠加补贴后实付7499元，相当于打了83折。 根据政策，2026年家电补贴聚焦冰箱、洗衣机、电视、空调、电脑、热水器6类核心产品，仅限一级能效/水效机型，补贴比例为售价的15%，单件上限1500 元，每人每类产品限购1件，累计最高可省9000元。 操作指南： 打开京东APP搜索"国补333"，进入家电专区； 选择标注"国补适用"的一级能效产品； 完成旧机回收（可预约免费上门取件）； 支付时自动抵扣补贴，收货后核验完成。 商家反馈： "没想到智能眼镜也能享受补贴！"上海白领陈先生刚入手一款价值5999元的AR眼镜，通过淘宝APP领取 ...