英国《金融时报》报道截图 当地时间1月24日，英国《金融时报》发表题为《中国是否正在悄然赢得人工智能竞赛？》的文章，举例 说明中国的开源人工智能模型为何正受到诸多美国企业的青睐。 英国《金融时报》报道标题 文章表示，每个月，都会有数以亿计的用户涌入美国图片社交平台拼趣（Pinterest）。在一个名为"最天马 行空的东西"的页面里，充满了大量稀奇古怪的想法，激发着创意人士的灵感。而支撑这一切的技术未必 是美国制造的，拼趣正在尝试使用中国的人工智能模型，来优化其推荐引擎。自2025年1月深度求索 （DeepSeek）R-1模型发布以来，中国的人工智能技术在拼趣中的应用日益增多。文章援引拼趣首席执行 官比尔·雷迪的话称，"深度求索选择开源模式，引发了开源人工智能模型的浪潮。" 文章称，除了深度求索外，中国的开源模型还包括阿里巴巴旗下的千问（Qwen）和月之暗面 （Moonshot）公司旗下的Kimi，字节跳动公司也在研发类似技术。文章援引拼趣首席技术官马特·马德里加 尔的话表示，这些模型的优势在于，像拼趣这样的公司可以免费下载和定制这些模型，而美国的OpenAI等 企业提供的绝大多数人工智能模型并非如此。马德里加 ...

Core Insights - The 2026 Chongli Forum, held from January 23 to 25, focused on the theme "Standing Towards New" to discuss the new order in the intelligent era, featuring prominent figures from technology, industry, and capital sectors [1][3]. Group 1: Forum Highlights - The forum was guided by the Chongli District People's Government and co-hosted by various organizations, including Taiwu Ski Town and Zhi Chen Semiconductor [1]. - Keynote speakers included Yu Minhong, Zhou Hongyi, and Yang Guang, who shared insights on artificial intelligence (AI) development in their respective fields [3]. Group 2: Keynote Presentations - Yu Minhong emphasized the vast potential of AI in education, advocating for embracing change and addressing educational disparities, particularly in rural areas [3]. - Zhou Hongyi discussed the evolution of AI, predicting that by 2026, the scale of intelligent agents could reach hundreds of billions, highlighting the dual evolution of large models and intelligent agents [5]. - Yang Guang focused on the role of designers in the AI era, stressing the need for collaboration with AI and the opportunity for individuals to become boundary-breaking creators [7]. Group 3: Panel Discussions - A fireside chat featured discussions on the importance of balancing model capabilities with commercial applications, emphasizing long-term investment in foundational models [9]. - The summit dialogue explored the symbiotic relationship between humans and AI, with participants discussing the need for emotional engagement and maintaining human judgment in AI interactions [11]. Group 4: Forum Outcomes - The "Chongli Entrepreneurs Alliance" was officially launched during the forum, with representatives from various companies witnessing the establishment of this new collaborative entity [11].

Market Landscape - The global AI large model market is characterized by a clear "U.S. dominance and Chinese catch-up" pattern, with U.S. models accounting for 93% of global traffic as of August 2025. However, the launch of DeepSeek R1 in January 2025 led to a significant increase in Chinese model market share from 3% to 13%, with a 460% growth in traffic within two months [1][22]. U.S. Advantages - The U.S. maintains a strong competitive edge through a "technology-capital-ecosystem" triad. Companies like OpenAI and Google benefit from first-mover advantages, strong brand recognition, and user habits established since the generative AI revolution in late 2022. Despite projected losses of $50-80 billion for OpenAI in 2024, the venture capital ecosystem supports their growth strategies [2][22]. Chinese Strategy - China's competitive advantages lie in a "policy-cost-open-source" framework. Chinese model APIs are priced at 1/6 to 1/4 of their U.S. counterparts, with DeepSeek R1 costing $0.55 per million tokens compared to ChatGPT's $2. The Chinese government supports AI infrastructure as a public good, providing subsidies that enable companies to offer low-cost or free services [3][22]. Technological Race - Chinese models have made significant strides in multilingual support, with Qwen increasing from 26 to 119 languages. DeepSeek also supports 102 languages, effectively reducing the language barrier previously held by U.S. models. However, the U.S. still leads in foundational research and talent density, indicating that a qualitative leap from "catching up" to "surpassing" has not yet occurred [4][22]. Geopolitical Variables - Geopolitical factors are reshaping the AI market, with DeepSeek's growth showing a "South rising, North declining" trend. Market share increases of over 10% were noted in regions like the Middle East, Africa, and Latin America, where economic ties to China are strong. In contrast, U.S. allies showed less than 5% growth [5][22]. Future Variables - The future of the AI competition will hinge on three key variables: the ability to overcome performance bottlenecks, the sustainability of business models, and the construction of ecological lock-in effects. The current low switching costs present instability, and the ability to create real barriers through enterprise solutions will be crucial for maintaining market share [6][22].

在周末看一场机器人足球赛，逛商场时买一杯机器人做的咖啡，上下班时坐一趟无人驾驶出租车……在 北京，这座科创高地、智慧之城里，人工智能的影子，正在不断融入日常生活中，为普通人带来可感、 可知的变化，这些变化，也在改变着一座城市的发展模式。 北京被称为"人工智能第一城"，这座城市的今天，正迎来AI浪潮爆发式增长的关键期。1月25日，北京 市第十六届人民代表大会第四次会议开幕，大会审查了北京市国民经济和社会发展计划报告（以下简称 计划报告）。计划报告明确提出，全力建设人工智能创新高地。 "十五五"开局之年，北京要怎样建设"全球人工智能创新高地"？"人工智能第一城"的未来，又会是什么 模样？北京市经济社会发展研究院产业发展研究所助理研究员许志锰表示，2026年，北京将在技术自 主、产业生态等领域持续发力，并持续开展国际国内合作，共同推动技术和产业健康发展，这些发展最 终也会惠及更多普通人，"未来的AI，不再是少数领域的专属，而是会有更多新领域、新产品，让人们 可感、可知。"他说。 发展成绩单：北京AI实力有多强？ 新京报：能否介绍一下，北京市建设人工智能新高地，有哪些基础？ 许志锰：说起北京的人工智能，一组数据就能看 ...

每经记者｜孙宇婷 杜蔚 每经编辑｜张益铭 在"万物皆可对话"的时代，孩子们正在与一个不知疲倦、无所不答的"伙伴"——AI（人工智能）共处。然而，在这种看似积极、开放 的人机互动常态下，一系列隐蔽风险正悄然滋生。 从播放音乐到随时应答的"百科助手"，智能音箱已成为许多家庭的标配。北京岳女士5岁的儿子突然向"小爱同学"提问："有哪些奥特 曼被打死过？"设备随即平静描述出"被冰冻大卸八块""砍头"等细节，令岳女士紧急叫停。而在成都，4岁女孩自幼儿园了解到某个大 模型后，便沉迷于和大模型对话，一次聊天能持续近两小时。 随着ChatGPT引爆生成式AI浪潮，以豆包为代表的新一代AI产品，凭借语音交互等功能，正以前所未有的速度和深度，涌入学龄前儿 童的世界。孩子们向这位"全能伙伴"抛出无穷无尽的问题，其展现的"渊博"与"耐心"，让不少家长放下戒备，甚至视之为"数字家教"， 但矛盾随之而来。"我一开始觉得很开心可以解放自己，但是孩子现在有些成瘾了。"网络平台上，不少家长的相关发帖和留言，暴露 了"AI上瘾低龄化"的问题。而最早一批尝鲜"AI带娃"的家长，已有人开始卸载相关应用。 "用少了怕落后，用多了怕他不会思考。"一位家 ...

1月25日上午，北京市第十六届人民代表大会第四次会议开幕，市长殷勇作政府工作报告。报告提出，要加快建设全球数字经济标杆城市，全面实施"人工智 能+"行动。 在林剑华看来，北京的核心竞争力体现在三个方面：一是无与伦比的人才集聚效应，涵盖顶尖科学家与青年创新力量；二是全栈式技术能力布局，从国产 AI芯片算力、底层软件框架到大模型研发均形成完整链条；三是雄厚的产业基础，当前北京人工智能核心产业规模与企业数量均占全国约一半。 对于下一步如何支持北京人工智能产业发展，新京报贝壳财经记者采访了北京市发展改革委党组副书记、副主任林剑华。 林剑华告诉记者，北京拥有全国最完整的人才链、产业链和平台基础。政府将以算力基础设施、软件生态自主化、大模型技术演进与应用牵引等方向为关键 抓手，持续推进AI发展，"AI应用不仅是商业落地路径，应用的过程也能够形成高质量数据，反哺人工智能技术迭代。" d 北京市发展改革委党组副书记、副主任林剑华。受访者供图 人才集聚、全栈式技术能力和雄厚产业基础是北京核心竞争力 1月5日，北京市召开了被称为"AI新春第一会"的人工智能创新高地建设推进会。在会上，北京市以九大行动、十大揭榜挂帅任务、四大创新街 ...

Group 1 - The third National Artificial Intelligence Application Scenario Innovation Challenge (CICAS) concluded in Suzhou, with 93 teams winning awards, including 12 special prizes, 36 first prizes, and 45 second prizes [1] - Notable projects awarded in the special prize categories include an intelligent amphibious transport system and a multi-source panoramic visual perception system [1] - The event recognized 20 typical cases of national AI application scenarios, 10 excellent organizational awards, and 14 outstanding instructor awards [1] Group 2 - Three authoritative reports were released, including the "2025 National AI Application Scenario Innovation TOP10 Best Practice Cities Report," which highlights differentiated development paths of leading cities like Beijing, Shenzhen, and Shanghai [2] - The "National AI Application Scenario Innovation 100 Directory List" systematically outlines 100 high-value scenarios across five major fields, including smart manufacturing and healthcare [2] - The "National AI High-Value Application Scenario Demonstration Enterprises TOP50 List" identifies 50 benchmark enterprises through a multi-dimensional evaluation system [2] Group 3 - CICAS serves as a benchmark event for AI application scenario innovation and technology commercialization, focusing on key areas such as large models, generative AI, and embodied intelligence [2] - The competition aims to explore a development path driven by demand for research and development, application driven by scenarios, and ecosystem empowerment for industries [2] - The event facilitates deep integration of AI enterprises with the real economy across innovation, funding, talent, and industry chains, contributing to the development of the digital economy and new productive forces [2]

图片来源： The World Economic Forum 达沃斯的新隐喻 2026年世界经济论坛，全球精英的目光聚焦到了一个新概念：Neural Spine，神经脊柱。 这不仅是一个技术名词，更是一个精准的组织隐喻。企业正在寻找自己的智能核心，一个能感知、决策并协调的中枢系统。在瑞士达沃斯的雪山之巅， Lightspeed合伙人Ravi Mhatre、You.com创始人Richard Socher，以及来自中国的Kimi总裁张予彤围坐在一起。他们试图拆解一个核心命题：当AI不再是外 挂的工具，而是成为组织的骨架时，企业会变成什么样。 定义"AI原生"的两种视角 AI时代，企业应如何设计组织、衡量效率？这不再是一个理论问题，而是一个迫在眉睫的实践挑战。 Ravi Mhatre， Lightspeed合伙人 ，也是Anthropic和Reflection AI的投资人，他 首先抛出了一个极具颠覆性的想象力框架，传统公司与AI原生公司的根本区 别在于思维起点："传统公司往往聚焦于现有技术和既定工作流，思考如何用AI优化存量；而真正的AI原生公司，思考的起点是' 如果我们拥有无限的智 能，我们应该创造什么？ ' ...

Core Insights - The emergence of new professions such as AI trainers, AI product managers, and AI ethics reviewers is reshaping the job landscape, requiring new skills from workers [1][2] - AI is creating new employment opportunities and transforming the employment ecosystem towards higher value-added roles characterized by human-machine collaboration [2][3] Group 1: New Job Opportunities - AI is driving the creation of new roles, including data annotators and AI content creators, leading to an increased demand for hybrid and application-oriented talent [2] - According to PwC's 2025 Global AI Employment Outlook, the number of jobs involving AI applications is increasing, with an average salary premium of 56% for workers with AI skills, double that of the previous year [2] - 58.com reported nearly 50 types of "human-machine collaboration" jobs and 40 new intelligent services emerging on its platform [2] Group 2: Impact on Traditional Industries - AI is becoming a crucial infrastructure for intelligent societies, enhancing traditional industries and creating new business models that improve employment quality [3] - In the robotics sector, embodied intelligence is advancing humanoid robots towards greater autonomy and human-machine collaboration, creating numerous new jobs across various applications [4] Group 3: Entrepreneurship and New Business Models - The rise of "One-Person Companies" (OPC) is lowering the barriers to entrepreneurship, allowing individuals to leverage AI tools for content production, product operations, and service delivery [6] - The trend of "single-person + AI" entrepreneurship is gaining traction, with cities like Suzhou promoting OPC as a preferred startup model [6] Group 4: Skills and Education - Experts emphasize that human imagination, judgment, aesthetic ability, critical thinking, and emotional interaction will remain irreplaceable advantages in the context of human-machine collaboration [7] - Future talent development should focus on interdisciplinary skills and comprehensive capabilities, alongside the establishment of a new training system for AI professions [7] - The concept of "slash" careers may become a primary lifestyle, encouraging individuals to avoid reliance on single skills or professions and to enhance their overall capabilities [9]

资本与商业研究 来源丨 市值榜（ ID：shizhibang2021） 作者丨相青 编辑丨 赵元 图源丨Midjourney OpenAI 正在告别技术理想主义，回归互联网的流量变现逻辑。 以下文章来源于市值榜 ，作者市值榜团队 市值榜 . 近期，ChatGPT将开始在面向美国用户的答案旁边添加广告，这些广告将在未来几周内上线，并会显示在答案的上方或下方，而不是答案内部。 据OpenAI内部预测，预计2027年，仅广告这一项新增业务的收入，就将突破100亿美元大关。到2030年，从非付费用户群体（通过广告投放及交易分 成）获取的收入目标，将高达1100亿美元。 一个新的广告巨头正在诞生。 历史政治反复证明，所有入口级产品，最终都会走向广告。搜索如此，社交如此，短视频亦如此。现在，轮到了AI。 AI广告终于来了 当一个AI产品成为入口级，变现路径只有四条，API调用、会员订阅、企业定制化解决方案以及广告，对于OpenAI 来说，前三条路已经跑通，但是远远无 法支撑其财务压力。 OpenAI预计到 2025 年收入将达到 130 亿美元，每周活跃用户达 8 亿。但与此同时，算力、模型训练和推理带来的成本压力仍在持 ...