更多资料加入知识星球：水木调研纪要 关注公众号：水木纪要 March 24, 2026 06:52 AM GMT Hardware Technology | Asia Pacific Datacenter Market Insights – Part 1: Overall Servers Total server shipments grew 10% y/y in 4Q25, as cloud remains the main driver, but enterprise demand was also recovering from troughs. We continue to expect AI and non-AI servers to see strong growth throughout 2026. Global server shipments totaled 4.4mn units in 4Q25 (+4% q/q, +10% y/y): The sequential increase continued to be driven by cloud, particularly for gen ...

3月26日，联想集团正式发布面向企业市场的"龙虾湖"智能体解决方案。该方案整合万全异构智算平 台、问天WA5480 G3服务器及配套存储网络设备，为企业提供私有化部署的高阶龙虾湖算力集群，主 打高智商、高安全性与高性价比、一体化交付与私有化部署等诸多特点。 据官方数据显示，该方案可实现每百万高质量Tokens成本不到1元，仅为同档云服务解决方案的三分之 一，同时将响应延迟控制在12毫秒以内。方案最多支持1000个智能体并发运行，4K上下文长度下日输 出可达百亿Tokens。并提供分场景的稳定峰值能力，可广泛适配集团IT、行政、业务中台等高价值部 门。 从个人"养虾"到企业"用虾"的跨越 "龙虾"引发的智能体热潮正从消费端向企业端蔓延。然而，个人版智能体在迁移至企业场景时面临多重 障碍：数据外泄风险、推理幻觉、上下文记忆缺失、高并发性能不足以及运营成本居高不下。联想集团 此次推出的"龙虾湖"方案，正是针对这些痛点设计的系统性解决方案。 该方案采用私有化部署架构，企业可完全掌控数据流通与访问权限，从而确保企业数据主权自主。通过 知识库的统一存储和管理，实现个人与企业知识的同步沉淀，加速组织知识资产构建。 安全性是 ...

Core Viewpoint - The consumer electronics industry is facing significant pressure from rising memory prices, leading to price increases across various brands, including Xiaomi, Transsion, Lenovo, OPPO, and vivo, as they attempt to offset the impact on profit margins [3][4][5]. Group 1: Memory Price Surge - Memory prices have skyrocketed, with Xiaomi's president stating that in Q1 2026, memory prices increased by 400% year-on-year, rising from $30 to over $120 [4][5]. - The surge in memory prices has led to a structural impact on the Bill of Materials (BOM) costs for smartphones, with low-end devices particularly affected, as memory now constitutes 43% of total material costs [10][11]. - Counterpoint's report indicates that DRAM prices increased by over 50% and NAND prices by over 90% in Q1 2026, significantly affecting the cost structure of low-end smartphones [10][11]. Group 2: Impact on Companies - Transsion, known as the "king of African phones," has been particularly affected by rising storage costs, leading to a decline in overall gross margins [4][5]. - Apple is also feeling the pressure from memory price increases, with CEO Tim Cook highlighting supply constraints and cost pressures during earnings calls [6][7]. - Lenovo has responded to the rising memory prices by signing long-term supply contracts to ensure adequate supply, but has also announced price increases for some of its computer products [8][9]. Group 3: Market Dynamics - The current market is characterized by a shift from a buyer's market to a seller's market, with companies now negotiating memory prices on a quarterly basis rather than annually [7][9]. - The demand for memory is being driven by the AI sector, which is squeezing supply for consumer electronics, particularly in the low-end market [10][12]. - Companies are facing a dilemma where they must raise prices to maintain margins, but this could lead to decreased consumer demand for their products [10][12]. Group 4: Future Outlook - Analysts predict that memory prices will remain high until at least the end of 2027, with no immediate relief in sight due to ongoing demand from AI applications [13][14]. - The structural changes in the memory market are expected to persist, with companies needing to adapt their product lines and pricing strategies to cope with the new reality [11][15]. - The potential for a market correction, such as an AI bubble burst, could alter the dynamics, but for now, the focus remains on navigating the high-cost environment [16].

三是AI数据标注，通过将碎片化信息转化为可用且一致的数据资产，相关组织可以利用数据支持内容 生产，增强球迷互动，并发掘新的收入机会。 作为2026年FIFA世界杯官方技术合作伙伴，联想集团将为足球历史上规模最大的赛事提供关键技术支 撑。为了应对104场比赛，数十亿的全球观众规模，联想集团于2026年FIFA世界杯计划部署的AI技术包 括：增强型数字转播可视化方案、裁判视角AI视频增强系统、智能指挥中心，以及面向海量数据的生 成式AI分析平台，用于支撑跨场馆及全球媒体网络的数据管理与分析。 三大AI体育方案的推出，展现出联想集团在"AI+体育"领域的长期投入——该集团为多类客户开发并部 署了定制化端到端解决方案，积累了成熟的实践经验。目前，联想集团涵盖设备、基础设施、解决方案 和服务的全栈AI能力，已在全球最具挑战性的现场体育赛事环境中，实现落地应用。 在世界一级方程式锦标赛(F1)中，联想集团为全球180多个国家和地区、超过8.2亿观众提供赛事直播技 术支持，亚秒级数据处理能力、长期稳定运行能力、全球同步能力，将成为联想集团案的"试金石"。 当地时间3月19日，为期四天的英伟达GTC 2026大会正在进行时。 ...

3月18日，联想集团发布天禧 AI Claw，并启动内测。据介绍，天禧AI Claw具备零成本部署、零门槛使 用、全天候跨端、安全可托付等核心优势。 当前AI智能体普遍面临部署繁琐、配置复杂、执行不稳定等问题，普通用户常被环境搭建、参数调 试、权限管理等技术环节挡在门外。联想基于这一行业痛点推出天禧 AI Claw，避免了传统AI工具复杂 配置的问题，让普通用户打开系统即可直接使用。系统内置经过安全验证的海量技能，覆盖办公、学 习、娱乐等日常场景，用户可直接调用，不用额外寻找与配置插件。 安全方面，天禧AI Claw依托独立云主机运行，配合严格的数据安全隔离机制，确保执行环境清晰可 控、权限边界明确、操作过程可追溯、风险可管理。既保证AI高效执行任务，又杜绝越权操作，让用 户放心交付核心任务，建立深度信任。 硬件配套方面，联想同步发布小新Pad Pro 13平板，首发搭载天禧AI PadClaw，专注AI学习场景，内置 笔记整理Skills。 近期 OpenClaw 引爆"养龙虾"热潮，AI 智能体从极客圈走向全民话题，算力、云服务、终端硬件全线躁 动，资本市场将 AI 从"会说"推向会做、可落地的价值重估窗 ...

Core Insights - Lenovo Group has launched Tianxi AI Claw, which offers zero-cost deployment and zero-threshold usage, transforming from a traditional tool into an executor capable of breaking down steps, running processes, allocating resources, and making judgments [1] Group 1: Product Features - Tianxi AI Claw eliminates the cumbersome deployment process of traditional AI agents, allowing users to start using the system immediately without installation or configuration [1] - The system is pre-installed with a vast array of high-frequency skills covering office, learning, and entertainment scenarios, all of which have undergone strict screening and security verification [1] - Users can directly invoke functions based on their needs without searching for or configuring third-party tools [1] Group 2: Technical Architecture - Tianxi AI Claw employs a hybrid deployment model that integrates local computing power with cloud resources, ensuring tasks initiated in the cloud can continue running even if the device is turned off [1] - The design prioritizes security and trust, utilizing independent cloud hosts and strict data security isolation mechanisms to ensure a clear and controllable execution environment [1] - The system features defined permission boundaries, traceable operation processes, and manageable risks [1]

Core Insights - The GTC 2026 conference highlighted the emergence of physical AI, with NVIDIA's CEO Jensen Huang stating that the "ChatGPT moment for physical AI has arrived" as machines begin to understand, reason, and act in the real world [1] Group 1: Product Launch and Features - Lenovo Group launched the Auto AI Box, a vehicle AI computing platform designed to integrate into existing automotive systems without requiring a complete redesign of the vehicle's electronic architecture [1] - The Auto AI Box is built on the NVIDIA DRIVE AGX Thor platform and utilizes Lenovo's expertise in automotive-grade hardware, featuring the agentic AI operating system FusionOS 4.0 to run multimodal large models locally within vehicles [1] - The platform supports natural language interaction and can operate models with parameter scales up to 13 billion (13B), providing robust and scalable computing power for in-vehicle AI applications [1] Group 2: Market Implications and Strategic Positioning - The introduction of the Auto AI Box allows automotive companies to quickly implement multimodal interaction, edge reasoning, and in-vehicle intelligent services, thereby shortening the development to deployment cycle [2] - Lenovo's move aligns with the clear signal from GTC 2026 that physical AI is transitioning from a supplementary function to a foundational capability in automotive infrastructure [2] - Lenovo is focusing on the transformation of vehicles into intelligent terminals rather than traditional vehicle manufacturing, aiming to secure a position in the next generation of smart mobility infrastructure with the Auto AI Box [2]

Core Insights - The company reported steady revenue growth with all business segments achieving double-digit growth, with FY3Q2026 revenue at $22.2 billion, up 18.1% year-on-year and 8.6% quarter-on-quarter, while net profit was $546 million, down 21.2% year-on-year but up 60.3% quarter-on-quarter [1] Business Performance - The company's smart devices, infrastructure solutions, and service solutions segments generated revenues of $15.755 billion, $5.176 billion, and $2.652 billion respectively, with year-on-year growth rates of 14.3%, 31.4%, and 17.5% [1] - The company maintained its leadership in the personal computer market with a global market share of 25.3% in Q4 2025, an increase of 1.0 percentage points year-on-year, and AI PC revenue saw high double-digit year-on-year growth [1] - The mobile phone business benefited from strong performance of the Motorola Edge series, achieving record sales and activations [1] AI and Infrastructure Solutions - AI server revenue experienced high double-digit year-on-year growth, supported by the deployment of the GB300 NVL72 rack-level solution, with a project reserve of $15.5 billion [2] - The company’s ISG (Infrastructure Solutions Group) reported an operating loss of $11 million in FY3Q2026, but with expectations of continued improvement and potential revenue balance in FY25/26 [2] Financial Metrics - The company’s operating expense ratio was 11.6%, down 0.7 percentage points quarter-on-quarter, with specific expense ratios for sales, management, R&D, and financial expenses [2] - Gross margin was 15.1%, down 0.3 percentage points quarter-on-quarter, primarily due to rising raw material costs and lower profit margins in the ISG business [2] - Net margin improved to 2.5%, up 0.8 percentage points quarter-on-quarter [2]

Core Insights - The future of data centers is shifting from mere storage facilities to "token factories" that enhance efficiency through AI and token production [1] - Lenovo has launched the Hybrid AI Advantage solution to accelerate AI implementation and reduce time to first token (TTFT), providing measurable business outcomes across various sectors [1][2] Group 1: AI Implementation and Solutions - Lenovo's new AI platform focuses on real-time enterprise-level inference, utilizing ThinkSystem and ThinkEdge servers to support multiple scenarios such as retail, manufacturing, healthcare, sports, and smart cities [1] - The solution promises a return on investment in less than six months, with token costs potentially reduced by up to 8 times compared to similar cloud IaaS offerings [2] Group 2: Product Offerings and Infrastructure - Lenovo has introduced two hybrid AI platforms: one for horizontal enterprise AI and multimodal inference, and another for training, fine-tuning, and large-scale inference [2] - The company collaborates with Nutanix, Cloudian, and Veeam to provide a comprehensive AI infrastructure that addresses data protection and operational concerns for enterprises [2] Group 3: Market Trends and Customer Adoption - The rise of agentic AI is driving exponential growth in inference workloads, making cost control and single token performance critical for businesses [2] - Customers like Iren and Cary Town are modernizing operations and expanding data-driven services using Lenovo's infrastructure [2]

Core Viewpoint - Lenovo Group and NVIDIA jointly launched the next-generation Hybrid AI Advantage solution at the GTC 2026 conference, aimed at accelerating AI implementation and reducing time-to-first-token (TTFT) while delivering quantifiable business results across personal, enterprise, and cloud environments [1] Group 1: Product Launch and Features - The new Hybrid AI solution extends from endpoints to data centers and up to gigawatt-level AI cloud deployments, enabling real-time decision-making, operational efficiency, and intelligent automation globally [1] - Lenovo became the global launch partner for NVIDIA's Vera Rubin NVL72 system, which boasts a throughput increase of up to 10 times compared to the previous generation, with the cost per token reduced to one-tenth of the prior version [1] Group 2: Strategic Importance - Lenovo's Chairman and CEO, Yang Yuanqing, emphasized the critical importance of cost control and single-token performance as agentic AI drives exponential growth in inference workloads [1] - By integrating NVIDIA AI Enterprise software with Lenovo's full-stack hybrid AI platform and services, the company aims to help clients scale AI applications more efficiently, at lower token costs, and with faster deployment speeds [1]