每一拍考验的不仅是选手的技术细节，更在极致地考验着耐性、韧性和全面性，以及背后长期的战略定 力。 " 大模型来临后的云计算下半场格局，其实就正如竞技体育中的多 拍相持，看似混乱、动荡，实则考证选手的方方面面。 " 作者丨周蕾 编辑丨包永刚 在网球或羽毛球等持拍运动中，有一种至关重要的战局，称为 " 多拍相持 " 。 双方选手在场上来回拉锯，场面看似波澜不惊，实则暗流涌动。选手们通过持续的对抗来试探与调整，捕 捉对手的弱点，同时耐心等待最适合发力一击的时机。待时机成熟，他们便会立刻挥拍杀向对方漏洞，由 守转攻占据主动，赢下这一分。 这正是百度智能云过去十年的真实写照。 在中国云计算这片高手云集的赛场上，当早期玩家凭借规模与资源加速奔跑时，百度智能云选择了另一条 路：以 AI 云为锚点，耐心蛰伏、长期布局，将核心力量投向全栈产品技术的构建 —— 从昆仑芯片夯实算 力底座，到百舸计算平台打破 AI 计算效率瓶颈，再到千帆 AI 平台全面支持 Agent 落地。 十年间，云计算市场从 " 资源驱动 " 迈向 " 智能驱动 " ，技术窗口终于开启。大模型的浪潮席卷而来， 成为那个打破均势的关键变量。而百度智能云，也等来 ...

Core Viewpoint - The article emphasizes the transition of cloud computing from cost reduction and efficiency enhancement to directly creating value through AI integration, highlighting the importance of AI capabilities in driving business growth and innovation [3][5][10]. Group 1: AI Integration and Business Impact - A report from MIT indicates that 95% of companies attempting to use AI have not seen commercial returns, but the remaining 5% have generated significant value, with some startups achieving revenues of $20 million within a year by focusing on specific business pain points [3]. - Baidu's intelligent cloud has supported 65% of central enterprises and 80% of systemically important banks, showcasing its extensive reach and impact in various industries [5]. - Baidu's AI capabilities are being transformed into revenue-generating tools for enterprises, with AI new business revenue expected to exceed 10 billion yuan by Q2 2025 [14]. Group 2: Cloud Computing Evolution - The cloud computing landscape is shifting from resource leasing to supporting AI capabilities for training, deployment, and invocation, marking a new phase in cloud competition [17]. - Baidu's cloud strategy has evolved from "cloud intelligence integration" to "intelligent priority," focusing on redefining how enterprises utilize cloud services starting from AI [7][10]. - The competition among cloud providers is now centered on engineering capabilities, scene adaptation, and system integration rather than just model capabilities [8]. Group 3: Technological Advancements and Offerings - Baidu has upgraded its AI computing platform to version 5.0 and its Qianfan platform to version 4.0, enhancing multi-model high-concurrency scheduling and integrating various capabilities for enterprise services [10]. - The company has developed a digital employee and compliance analysis tools, aiming to standardize AI applications across industries [10]. - Baidu's self-developed Kunlun AI chips and training platforms provide a competitive edge in terms of cost efficiency and scheduling capabilities [8][9]. Group 4: Market Position and Future Outlook - Baidu holds a leading position in China's AI public cloud service market with nearly 25% market share and 15% share in the large model platform market [14]. - The company aims to convert complex AI capabilities into practical business solutions, addressing the dual challenges of finding new growth curves and ensuring secure development space in a competitive global landscape [18][19]. - Baidu's strategy includes comprehensive self-research across infrastructure, model frameworks, and engineering platforms to reduce reliance on external ecosystems [19].

Core Insights - The integration of AI and cloud computing is reshaping the technology and business landscape, marking a new starting point for the industry [2][4] - In 2024, China witnessed the emergence of 50 trillion-level parameter models, leading to a 74.1% year-on-year growth in intelligent computing power, significantly outpacing general computing growth [1] - Baidu's intelligent cloud has become a leader in the AI field, leveraging its comprehensive AI technology stack to drive innovation and transformation in cloud services [4][6] Industry Trends - The cloud computing sector is undergoing unprecedented changes due to the incorporation of AI technologies, which are redefining service architectures and value propositions [1][4] - The competitive landscape in cloud computing is shifting from cost-based competition to a focus on innovative solutions that address specific business challenges [4][6] Company Developments - Baidu has made significant advancements in AI, launching the first deep learning framework "PaddlePaddle" in 2016 and the AI chip "Kunlun" in 2018, establishing a robust foundation for its AI capabilities [4][5] - Baidu's intelligent cloud has achieved a 24.6% market share in China's AI public cloud market, maintaining its position as the leader for six consecutive years [6][7] - The "Cloud Intelligence Integration" strategy of Baidu's intelligent cloud is set to evolve further, aiming to enhance intelligence capabilities and expand its market penetration [6][7] Product Innovations - The Baidu "Baihe" AI computing platform has been upgraded to version 5.0, focusing on enhancing inference capabilities and optimizing resource utilization [6][7] - The "Qianfan" platform has transitioned to version 4.0, providing a one-stop enterprise service platform centered around AI agents, aimed at accelerating application deployment [7] Financial Performance - Baidu's intelligent cloud revenue grew by 27% year-on-year, contributing to the company's AI new business surpassing 10 billion yuan for the first time, with a 34% year-on-year increase [7]

AI与云计算，正在以前所未有的紧密姿态绑定在一起。 2024年，中国新增千亿级别参数大模型50个，推动智能算力规模同比增长74.1%，是通用算力同期增幅的三 倍以上，智算中心利用率持续突破70%[1]。 云计算是AI规模化落地的基石，AI则彻底改变了云服务的架构与价值逻辑，催生了"AI云"这一新范式，使 得云计算产业突破长期桎梏，跑出了增长的第二曲线。 正如 百度集团执行副总裁、百度智能云事业群总裁 沈抖说，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"。 随着产业进入应用规模化落地期，从"知识增强"走向"执行增强"，对AI云的需求也从底层的AI算力基础设施 走向了更全面的一站式端到端解决方案，给云厂商提出了新的挑战。 AI与云计算的共生关系重构技术与商业的边界，正在画出全新的起跑线，云计算固化已久的竞争格局也开 始松动。 AI 重塑云计算赛道 2020年5月，微软用28.5万颗CPU和1万颗GPU，在Azure云平台为OpenAI构建了第一台AI超级计算机，为两 年后的"ChatGPT时刻"按下了倒计时。 奥特曼（左）和纳德拉（右） 奥特曼将这次合作称为"科技界最佳基友情"，而对当 ...

Core Viewpoint - The rapid evolution of large models, particularly since the release of ChatGPT in November 2022, has led to significant market transformations, with Nvidia's market value increasing tenfold to over $4.4 trillion, and the emergence of 4,000 large models and nearly 500 AI unicorns attracting around $500 billion in investments [2] AI Cloud Market Dynamics - The AI cloud market in China is projected to reach 19.59 billion yuan in 2024, with the top three players—Baidu, Alibaba, and Huawei—holding a combined market share of 62%, an increase of 8% from 2023 [5] - The shift from CPU server competition to GPU-based intelligent computing clusters marks a significant transition in the cloud services industry, emphasizing the delivery of intelligent outcomes rather than merely renting computing resources [3][6] Challenges for Smaller Players - Smaller cloud service providers face significant barriers, including high costs for GPU procurement, which can be 10-30% higher than those for leading firms, and the complexity of developing AI cloud capabilities that require a full-stack approach [6][7] - Over 30% of regional cloud service providers are expected to exit the market by 2024 due to unsustainable operational costs and low utilization rates of computing resources [7] Importance of Ecosystem and Industry Experience - Enterprises increasingly seek cloud service providers that can address specific industry pain points, with 65% of central enterprises and 80% of systemically important banks choosing Baidu Intelligent Cloud for its industry experience [8] - Baidu's approach focuses on creating value for enterprises by helping them reduce costs and improve efficiency, as demonstrated by successful collaborations in various sectors [8][19] Competitive Landscape and Strategic Positioning - Baidu Intelligent Cloud leads the AI public cloud market with a 24.6% share in 2024, emphasizing an AI-driven cloud model that redefines traditional cloud services [13][20] - The company has established a strong competitive edge through its full-stack capabilities and strategic focus on industry-specific solutions, allowing it to secure significant contracts and partnerships [13][19] Future Trends and Innovations - The evolution of AI applications towards more autonomous and intelligent agents is expected to create a market exceeding 3 trillion yuan by 2027, highlighting the urgency for players to innovate and lead in technology [15] - Baidu's advancements in AI infrastructure, such as the "Kunlun" super node and self-developed computing clusters, position it as a key player in the domestic AI industry, reducing reliance on external technologies [16][20]

全球云计算，经历过一场又一场范式转移。 关键抉择：预判，云计算的拐点 时代的更迭、技术的演进，并非只是"性能更强"、"成本更低"的线性提升，它改变的是企业使用云计算的方式——也改变了云计算本身的定义。 从2006年AWS对外提供云服务至大模型出现前，云计算的市场逻辑清晰而粗放：谁的资源更多、价格更低，谁就能赢得客户。在那个阶段，企业上云的 主要诉求是成本优化与弹性扩容，云厂商的竞争也主要围绕服务器规模、带宽能力和价格策略展开。 这一时期，国内各个厂商快速崛起，依托各自的基础资源、资金实力和渠道优势，占据了公有云市场的大部分份额。这场竞争本质上是一场卖水的游戏 ——客户关心的是水是否够多够便宜，而不是水背后是否能生成价值。 拐点很快到来。随着OpenAI、Anthropic等AI创企的爆发，全球云厂商逐渐形成共识：客户的需求已不再仅限于算力托管，而是需要智能能力。金融行业 客户要求大模型能实时识别风险、生成报告；制造业希望用视觉大模型提升质检准确率，降低损耗；游戏厂商希望AI成为陪伴式的交互角色，而不仅仅 是后台支撑。 AI开发与推理需要更高的算力、更稳定的延迟，以及端到端的优化，这让云厂商必须从资源提供者转型 ...

当云计算的竞争从"卖资源"转向"卖智能、 卖价值"，百度为什么能够站上舞台中央？ 全球云计算，经历过一场又一场范式转移。 每一次技术演进，背后都是一轮产业结构的重新洗牌：在PC互联网时代，云是企业搭建IT基础设施的外包方案，本质是服务器租赁；到了移动互联网时 代，云被用来支撑海量用户的高并发需求，强调的是稳定性与弹性扩容；在如今大模型驱动下，云的角色进一步演化，它不仅是计算资源的提供者，还要承 担企业增长、发展的智能能力。 这对云厂商提出了更高的要求：要从拼资源、拼价格，转向拼智能、拼行业理解能力和交付能力。 现在已经能够看到，大模型的爆发式发展带来了新一轮洗牌：OpenAI借助微软Azure构建起世界领先的模型生态，Anthropic借力AWS快速扩张，连甲骨文 这样的传统玩家也凭借AI实现了业务的重新增长。 在中国，转变同样剧烈。百度智能云正在借着这次机会超车，从追赶者，变成AI云赛道的领跑者之一。问题也愈发清晰：当云计算的竞争从"卖资源"转 向"卖智能、卖价值"，百度为什么能够站上舞台中央？ 关键抉择： 预判，云计算的拐点 时代的更迭、技术的演进，并非只是"性能更强"、"成本更低"的线性提升，它改变的是 ...