编辑 | 虞尔湖 出品 | 潮起网「于见专栏」 在当今数字化与智能化飞速发展的时代，云计算已成为推动各行业创新与变革的关键力量。百度智能云自2015年正式对外开放运营以来，凭借其独特的技术 优势、卓越的市场表现以及广泛的产业赋能，逐渐在智能云领域崭露头角，成为行业内的领航者。 构建坚实的智能云技术基石 据了解，百度智能云作为中国AI云的先行者，在技术创新方面成果斐然，为其在智能云市场的领先地位奠定了坚实基础。 在芯片领域，百度自主研发的昆仑AI芯片，为智能云业务提供了可靠的动力保障，有效提升了计算效率和性能。 在网络方面，百度智能云拥有100T级别的带宽，近千个边缘节点，并且是国内最早大规模应用自研的万兆、25G、100G网络设备的企业之一，确保了数据 传输的高效与稳定。 其数据中心覆盖北京、保定、苏州、南京、广州、阳泉、西安、武汉、香港、新加坡等地区，其中阳泉数据中心更是亚洲单体最大数据中心，为海量数据的 存储和处理提供了有力支撑。 2025年8月28日，在百度云智大会上推出的升级版百舸AI计算平台5.0和千帆企业级AI开发平台4.0，更是技术创新的重要体现。 百舸AI计算平台5.0在网络通信上更快、延迟更 ...

近日，2025百度云智大会在北京成功举办。百度智能云全新升级百度百舸AI计算平台5.0和千帆企业级AI开发 平台4.0，让企业以更低成本、更高效率部署和开发AI产品。 会上，百度集团执行副总裁、百度智能云事业群总裁沈抖分享了百度智能云新一代AI云基础设施最新进展和 落地实践。 其中百度百舸AI计算平台为具身智能 "大脑" 迭代提供支持，已全面适配主流开源具身VLA模型，将视觉语言 模型强化学习训练速度提超1倍、单轮迭代时间缩超50%。 该平台已支持包括北京、上海、浙江、广东创新中心在内的具身智能"国家队"，并已服务产业链超20家重点企 业，加速技术成果落地；同时，提供高效云端开发工具，实现研发两倍提效。 作为本届大会的重要环节，"AI+机器人 " 专场论坛吸引了全网超过数十万网友的在线观看。 ▍ 具身智能基础模型多技术路线快速迭代 落地场景应用可期 现场，北京大学计算机学院助理教授、智源具身多模态大模型中心负责人仉尚航详细介绍了具身智能多模态大 模型演进趋势。 百度智能云副总裁、泛科技业务部总经理张玮 仉尚航认为，构建具身快慢系统是具身智能从 "单一任务/本体" 迈向 "通用泛化" 的关键路径。技术路线上， ...

市场分析人士指出，百度在AI云基础设施方面的持续创新，以及与多家企业建立的深度合作关系，都反映出公 司在快速增长的AI市场中的强劲发展势头。 新京报贝壳财经讯 8月29日，百度（BIDU）美股上涨超5%，表现显著优于大盘。 百度最新财报数据显示，2025年第二季度，受AI驱动，涵盖智能云在内的AI新业务收入增长强劲，首次超过100 亿元，同比增长34%。IDC数据显示，百度智能云已连续六年位居中国AI公有云市场第一，这也增强了投资者的 信心。 在近日举办的2025云智大会上，百度智能云作为全球AI云首倡者，对其"云智一体"战略下的核心产品进行了重磅 迭代，推出旨在帮助企业更低成本、更高效率开发AI应用的百舸AI计算平台5.0和千帆企业级AI开发平台4.0。会 上，百度智能云还公布了最新AI落地进展，65%+央企，95%+主流车企以及80%+具身智能企业落地大模型选择 百度智能云。 编辑 徐倩 校对 赵琳 ...

Core Insights - The focus of the cloud computing market has shifted from "resource scale" to "intelligent capability" in the AI-native era, driven by significant advancements in AI models and infrastructure [1][4][19] - Baidu's "Artificial Intelligence +" initiative aims to deeply integrate AI technology across various industries, enhancing quality, efficiency, and innovation [1][2] - Baidu's intelligent cloud strategy emphasizes the need for new infrastructure to support the intelligent economy, positioning AI cloud as a new profit center rather than a cost center [4][7][19] Cloud Infrastructure Developments - Baidu has upgraded its AI infrastructure with the launch of the Baijie AI Computing Platform 5.0 and Qianfan Enterprise AI Development Platform 4.0, enabling lower-cost and more efficient AI product deployment [4][12] - The new platforms focus on enhancing computational efficiency, reducing latency, and improving model training and inference capabilities [10][12] - Baidu's AI cloud services have achieved a market share of 24.6%, maintaining its position as the leading provider in China's AI public cloud service market [18][19] AI Model and Agent Development - The core elements of AI cloud are identified as computing power, models, data, and engineering capabilities, which are essential for supporting the rapid growth of AI applications [8][9] - Baidu's Qianfan platform has been upgraded to support agent development, providing a comprehensive service platform for developers with over 150 models available [14][15] - The introduction of the RFT toolchain significantly reduces the data requirements for model fine-tuning, making it easier for enterprises to enhance model performance [15] Industry Transformation and Future Outlook - The AI-native era is expected to create a "super cycle" in the industry, where the value creation process is fundamentally restructured, leading to new opportunities for businesses [6][19][21] - Baidu's strategic shift from traditional cloud services to AI cloud services is seen as a "change of lane" competition, positioning the company as a leader in the AI cloud sector [19][21] - The integration of AI capabilities into cloud services is viewed as a critical factor for enterprises to drive value creation and operational efficiency [21]

如果 把时间拉 回到 三 年前，很少有人预料到，云计算市场的竞争焦点会从 "资源规模"彻底转 向"智能能力"。 从 2022年11月30日OpenAI发布ChatGPT算起，到 现在 短短 33个月，整个世界围绕大模型发生 了巨大变化：算力投入巨大，成就了英伟达3年10倍的市值增长，达到了4.4万亿美元；模型快速迭 代，催生了4000个大模型，在榜单上各领风骚；应用爆发式增长，诞生了将近500家AI独角兽，吸 引了近5000亿美元的投资。 在 AI重构产业的超级周期里，云服务厂商的下一站究竟在哪 儿 ？ 日前，国务院印发《关于深入实施 "人工智能 + "行动的意见》。百度副总裁石清华表示，"人工智 能 + "行动将加速 AI 技术向千行百业深度渗透，推动制造业、服务业、金融、交通等重点领域实现 质量变革、效率变革和动力变革，全面激发新产业、新业态、新模式的创新活力。百度将持续加大在 AI 技术创新上的投入，依托"云智一体"优势，打造更易用、更高效、更安全的 AI 基础设施与平台 服务。 8月28日的2025百度云智大会上，百度集团执行副总裁、百度智能云事业群总裁沈抖给出了云厂商 在AI原生时代的新定位 —— ...

Core Insights - The demand for AI cloud infrastructure is rapidly increasing, driven by various user groups from individual users to enterprises [1][2] - Baidu Smart Cloud holds a 24.6% market share in China's AI public cloud market, maintaining its leading position for six consecutive years [1] - The launch of upgraded platforms, Baidu's "百舸AI计算平台5.0" and "千帆企业级AI开发平台4.0," aims to help enterprises deploy and develop AI products more efficiently and cost-effectively [1][2] Market Trends - The Chinese AI public cloud service market is projected to reach 19.59 billion RMB in 2024, reflecting a year-on-year growth of 55.3% [1] - The shift of AI cloud infrastructure from a cost center to a new profit center is a significant industry trend [1] Product Developments - The "百舸AI计算平台5.0" upgrade introduces the Kunlun Chip Super Node, enhancing computing power for AI applications [2] - The "千帆企业级AI开发平台4.0" offers a comprehensive AI development infrastructure, including a model library and agent orchestration [2][3] - New self-developed industry-specific models have been introduced to address the limitations of general models in performance and cost-effectiveness [3] Ecosystem Expansion - Baidu Smart Cloud has developed a series of ready-to-use agents, enhancing operational efficiency for enterprises [5] - The introduction of the "吴彦祖数字英语教练" digital employee showcases the application of AI in various business functions [5] - Collaborations with emerging companies in the AI sector demonstrate Baidu Smart Cloud's commitment to fostering growth in the intelligent economy [5]

Core Insights - The core viewpoint emphasizes the necessity of new infrastructure, specifically AI cloud, to support the intelligent economy era, as articulated by Baidu's executive vice president [1] - Baidu's intelligent cloud has launched upgraded platforms, indicating a significant growth trajectory in AI-related revenue, projected to exceed 10 billion yuan by Q2 2025, marking a 34% year-on-year increase [1][7] Group 1: AI Cloud Infrastructure - The transition from cost-saving to value creation in enterprise infrastructure is highlighted, with AI cloud becoming a profit center rather than a cost center [2] - Four core elements of true AI cloud are identified: computing power, models, data, and engineering capabilities, which together form a unified and evolving AI cloud infrastructure [3] Group 2: Upgraded Platforms - Baidu has launched the upgraded Baidu AI Computing Platform 5.0, which features faster communication, lower latency, and enhanced model training efficiency [4] - The platform supports the largest open-source model parameters reaching 1 trillion, demonstrating significant computational capabilities [4] Group 3: Collaboration and Development - Companies like LingSheng Technology have benefited from Baidu's cloud technology, significantly reducing testing costs and accelerating the transition from lab to large-scale application [5] - The 千帆 4.0 platform aims to lower the development threshold for Agents, with over 460,000 enterprise users and 1.3 million Agents developed [6] Group 4: Market Position and Growth - The Chinese AI public cloud service market is projected to reach 19.59 billion yuan in 2024, with Baidu and Alibaba sharing the top market position at 24.6% market share [6] - The natural language processing public cloud service market is expected to grow by 51.1% in 2024, with Baidu, Huawei, and Alibaba leading the market [6]

Group 1 - The core viewpoint emphasizes the necessity of new infrastructure, specifically AI-first cloud solutions, to support the intelligent economy era [1] - Baidu Smart Cloud has upgraded its Baichuan AI Computing Platform to version 5.0 and Qianfan Enterprise-level AI Development Platform to version 4.0, enabling businesses to deploy and develop AI products at lower costs and higher efficiency [3] - The Baichuan AI Computing Platform 5.0 has achieved comprehensive capability enhancements in four key areas: network, computing power, inference systems, and integrated training and inference systems, addressing AI computing efficiency bottlenecks [3] Group 2 - The Qianfan platform has been upgraded to version 4.0, focusing on creating a one-stop enterprise service platform centered around Agents, providing developers with necessary models, Agent orchestration, data, and enterprise-level capabilities [3] - Baidu Smart Cloud has developed ready-to-use Agents on the Baichuan + Qianfan intelligent infrastructure, including the "Wu Yanzu Digital English Coach," created in collaboration with Yashi Education, utilizing Baidu's self-developed end-to-end voice and semantic large model [3] - The Wu Yanzu digital persona will officially serve as the first batch of recommended officers for Baidu Smart Cloud's digital employees starting immediately [3]

转自：北京日报客户端 从个人调用AI聊天，到独立开发者使用AI打造"一人公司"，再到企业借力AI重构传统业务，各种群体 对于AI大模型的需求，共同催生了AI云需求的大爆发。8月28日在京举行的2025百度云智大会上，百度 智能云宣布昆仑芯超节点正式启用，并升级了百舸AI计算平台5.0和千帆企业级AI开发平台4.0，让企业 能够以更低成本、更高效率部署和开发AI产品。 IDC发布的《中国AI公有云服务市场份额2024》报告指出，2024年中国AI公有云服务市场规模达195.9 亿元人民币，相比2023年增长55.3%。算力抬高智能天花板，极致的计算效率仍然是模型不断突破智能 上限的核心驱动力，如何平衡AI计算的投入成本和产出效益，成为行业的核心关切。 据悉，百舸AI计算平台5.0版在网络、算力、推理系统以及训推一体系统等四大方向上实现了能力提 升，以打破AI计算效率瓶颈——网络上，通信更快，延迟更低，提升模型训推效率；算力上，昆仑芯 超节点上线公有云服务，超级算力正式可用；推理系统上，通过"解耦""自适应"和"智能调度" 三大核心 策略，提升吞吐，降低延迟；训推一体上，发布百舸强化学习框架，极致压榨算力资源，提 ...

新浪科技讯 8月28日下午消息，今日举办的2025百度云智大会上，百度集团执行副总裁、百度智能云事 业群总裁沈抖指出："智能经济时代，企业对AI基础设施（简称：AI Infra）的要求，已经从'降本增 效'转向'直接创造价值'，所有计算产生的智能都会被封装进Agents，参与价值创造和交付，企业的AI 云，不再是成本中心，而会成为新型的利润中心。" 沈抖指出，互联网时代，核心是信息流通，网站和App封装的是信息，交付的也是信息，信息的流通创 造了巨大的价值，对于大多数企业来说，一个官网、几个媒体账号就够用了。但智能时代不一样，它的 核心是Agent智能体，封装的是智能，交付的是结果，所以企业可以只有一个官网、几个账号去展示信 息，但一定会有大量的Agents去接管从生产到经营的各种环节，直接创造价值。 "去年我们说，有多少场景，就有多少Agents，现在看来还是有些保守，现在的一个岗位、一个场景， 以后就可能是几个、几十个Agents。"在沈抖看来，"当价值的创造方式被重构，产业链也会进化，AI就 会进入真正的'超级周期'，而今天，我们正站在这个超级周期的起点，加速进入智能经济时代！" 沈抖指出："智能经济时 ...