Core Insights - Celestica, Inc. is well-positioned to benefit from the growing AI proliferation in the data center and edge infrastructure market, with the AI hardware market expected to grow at a 22.4% CAGR, reaching $210.5 billion by 2034 from $34.05 billion in 2025 [1] - The company is expanding its offerings through innovation and strategic collaborations to capitalize on emerging market trends [2] - Celestica's extensive 800G Family of Networking Switches is anticipated to be a major growth driver, optimized for AI/ML and Big Data Analytics [3] Company Developments - The DS4100 switch, designed with Broadcom's TH4-12.8T switch chipset, is gaining traction in the data-center networking market [2] - The introduction of the ES1500 switch is tailored for edge AI deployments and IoT expansion in various sectors [3] - Celestica's shares have surged 167.5% over the past year, outperforming the industry's growth of 96.4% [7][8] Competitive Landscape - Competitors include Jabil, Inc. and Flex Ltd., both of which are also benefiting from demand in AI-related sectors [4] - Jabil announced a $500 million multi-year investment initiative to expand its manufacturing capabilities in the AI data center infrastructure market [5] - Flex is focusing on industrial AI and has improved its data center solutions through the acquisition of JetCool Technologies [6] Valuation Metrics - Celestica trades at a forward P/E ratio of 27.89, which is above the industry average, potentially raising valuation concerns [8][9] - The Zacks Consensus Estimate for Celestica's earnings for 2025 and 2026 has remained unchanged over the past 60 days [11]

Core Viewpoint - The article focuses on the disassembly and analysis of the headlight driver module of the Xiaomi SU7, highlighting its advanced technology and components that differentiate it from traditional fuel vehicles [5][16]. Summary by Sections Overview of the Headlight Driver Module - The headlight driver module is described as the "brain" of the lighting system, essential for high-precision light control, dynamic lighting, communication, and energy management in automotive applications [5]. Disassembly Insights - The disassembly process of the SU7 headlight driver module was challenging due to the extensive use of adhesive between the bottom shell and the metal front cover, requiring significant effort to extract the driver board [7]. Component Analysis - The core controller is an Infineon automotive-grade MCU (CYT2B75CADQ0AZEGS) based on the Arm Cortex-M4 architecture, responsible for lighting logic operations, PWM waveform generation, fault diagnosis, and communication protocol stack execution [11]. - Two CAN bus controllers from NXP are included for communication redundancy, with one supporting CAN FD for high bandwidth lighting command transmission [11]. - The module features a linear voltage regulator from MPS (MP2019) that can handle 12V vehicle transient fluctuations and minimizes standby power consumption [12]. - An ON Semiconductor micro-stepping motor driver (NCV70517) is utilized for adaptive front lighting system (AFS) functionality, typically found in mid to high-end traditional fuel vehicles [13]. - A voltage follower from ROHM (BD42530) ensures dynamic voltage tracking for the light group, maintaining brightness control accuracy [14]. - The main LED driver system is powered by Texas Instruments chips, with a dual-channel constant voltage LED driver (TPS92682Q) and three synchronous buck LED drivers (TPS92520Q) for auxiliary lighting [14][15]. Conclusion on Component Selection - The analysis reveals that the Xiaomi SU7 headlight driver module employs a mix of high-performance international components and some local suppliers, indicating a strategy for supply chain diversification and reliability [16].

镁信健康首席商务官郎立良受邀出席,并发表了题为《从数据到智能体:生成式AI如何赋能保险服务链 路》的主题分享,为与会专家学者带来前沿的行业洞察与实践经验。 近日,保险与精算四十人论坛青年论坛2025年年会在南开大学八里台校区省身楼举行。本次大会以"数智 赋能与保险新质生产力"为主题,汇聚了保险、科技、医疗等领域的专家学者与企业精英。与会专家学者 聚焦"保险科技与新质生产力创新""生成式AI和大数据在保险经营中的应用""数智赋能背景下保险行业 生态建设面临的机遇与挑战""保险业数智化转型机遇与挑战"等议题展开深入研讨,凝聚各方智慧,为保险 业数智化转型与创新发展提供前瞻思路和实践路径。 镁信健康早在多年前就敏锐地捕捉到了这一趋势,并积极投入研发,致力于将AI技术与保险服务深度融 合。凭借在医药险全场景的深厚积淀,镁信健康成功构建了以mind42.ai为核心的AI模型体系,为用户提供 一站式医疗健康服务和保险支付等服务,推动行业服务迈向智能化、高效化的新阶段。 郎立良还特别强调了数据治理在AI应用中的重要性。他指出,数据是AI模型的"燃料",只有将海量的数据 进行有效的治理,转化为可用、有用的信息,才能确保AI模 ...

点击下方 卡片 ，关注" 自动驾驶之心 "公众号 戳我-> 领取 自动驾驶近15个 方向 学习 路线 中科院 & 理想 ICCV'25 World4Drive 中科院自动化所（CASIA）、理想汽车（Li Auto）等机构提出无感知标注的端到端自动驾驶框架 World4Drive，通过融合视觉基础模型的空间语义先验与多模态驾驶意图建模，在nuScenes和NavSim基准上 实现L2误差降低18.1%（0.61m→0.50m）、碰撞率下降46.7%（0.30%→0.16%）、训练收敛速度提升3.75 倍。 主要贡献： 算法框架： 实验结果： 论文标题：World4Drive: End-to-End Autonomous Driving via Intention-aware Physical Latent World Model 论文链接：https://arxiv.org/abs/2507.00603 项目主页：https://github.com/ucaszyp/World4Drive 1. 提出意图感知的潜在世界模型，通过模拟不同驾驶意图下物理世界的演化过程，实现多模态轨迹的生成 与评估，贴近人类驾驶员 ...

Key Takeaways Micron's Embedded Business revenues rose 20% sequentially in Q3 to $1.2 billion on strong AI-driven demand. AI use in factory automation and autos boosted sales, aided by tight supply and low channel inventory. Micron reorganized its business units in June to focus more deeply on AI-led growth opportunities.Micron Technology’s (MU) Embedded Business Unit saw strong growth in the third quarter of fiscal 2025. The business unit’s revenues were up 20% sequentially to $1.2 billion. The company a ...

完成IPO前最后一轮融资后的估值为116.775亿元人民币。 本文为IPO早知道原创 作者｜ Stone Jin 微信公众号｜ipozaozhidao 据 IPO早知道消息， 上海镁信健康科技集团股份有限公司 （以下简称 " 镁信健康 "）于2025年6月30日正式向港交所递交招股说明书，拟主板挂牌上 市，高盛、中金公司和汇丰担任联席保荐人。 成立于 2017年的 镁信健康致力于解决患者、保司与药企面临的筹资与支付挑战，助力推动中国医疗支付体系变革。根据弗若斯特沙利文的资料，镁信 健康 现已成为 中国最大的医药多元支付平台 。 截至 2024年12月31日，镁信健康 基于自己的 解决方案累计为患者节省自付费用约 67亿元。 截至目前， 镁信健康 已 开发了两大行业解决方案： 其一、 智药解决方案 ： 为药企提供了药品全生命 周期 的商业化方案，帮助药企整合多元支付方 。 根据弗若斯特沙利文的资料，以 2024年GPV计算，镁信健康在中国创新药械解决方案提供商之中排名第一。 其二、 智保解决方案整合自有 AI技 术、更优质的医疗健康资源，为保司提供端到端方案，推动健康险创新。根据弗若斯特沙利文的资料，以2024 ...

4分14秒！在2025张江具身智能开发者大会暨国际人形机器人技能大赛的激烈角逐中，这个数字不仅定格了 同济大学具身双臂操作机器人团队（以下简称"同济具身智造团队"）在具身工业场景链盒上下料赛项的夺冠时 刻，更以绝对优势创造了该项目的 "张江纪录"。 这不仅是速度的胜利，更是具身智能技术向真实工业场景深度落地的一次突破性进展。 团队聚焦于制造业中 最基础也最关键的环节——链盒上下料。这个看似简单、却在工厂车间日复一日重复千万次的操作，对机器人 的精度、协同性及环境适应能力提出了极致挑战。同济团队如何突破重重技术难关，让机器人在这一核心任务 中展现出 可替代和辅助 人工的效率和可靠性？ 机器人大讲堂深入采访了团队 的参赛 指导教师及核心成员， 为您深度解析这场技术攻坚的制胜关键。 ▍ 毫厘之间的较量：技术重新定义工业精度 "连续完成 6个乱序零件的精准抓取和放置，每个零件的定位精度都控制在0.1毫米以内。"同济大学自主智能 无人系统科学中心2025级博士生顾羲和介绍道。评委们特别肯定了该系统在狭小空间作业时的稳定性和效 率，认为其充分展现了具身智能在工业场景中的实用价值，证明了机器人替代人工完成高重复性、环境恶劣 ...

Group 1 - Shanghai Youning Biological Technology Co., Ltd. won a bid for reagent procurement from Shandong First Medical University, with a total transaction amount of 7403.70 RMB [1] - The procurement includes multiple items, all supplied by CST, a manufacturer based in the United States [2] - Each item has a unit price of 2467.90 RMB, with a delivery requirement of within 7 days after contract signing [1][2] Group 2 - The items procured include K63-linkage Specific Polyubiquitin, Phospho-NF-κB p65, and NLRP3 Rabbit mAbs, each with a specification of 100ul [2] - The procurement emphasizes standard service for warranty and after-sales support [1][2] - Payment is contingent upon successful delivery and acceptance of the goods [1]

规划建面超过32万的宝山南大"巨无霸"地块被中建八局以82亿元收入囊中。 6月19日，上海开启第五批次土拍，此次土拍将延续两天，19日当天先行出让5宗涉宅地块，总计揽金 191.57亿元。 本场土拍共吸引了14家房地产企业参拍，其中包含6家央企、7家国企和一家民企。 在此次土拍开始前，宝山区南大多用途组合用地因为其体量巨大，又是宝山南大板块TOD产城融合的 地标，备受市场关注。根据地块出让信息，该组合地块总出让面积接近10万平方米，建筑面积达到32万 平方米，起拍价高达81亿元。 而从房企报名情况来看，杨浦区东外滩地块、闵行区颛桥地块热度更高，受到更多开发商的争夺。 最终土拍结果显示，当日最热门地块是杨浦区东外滩地块，出让面积为1.47万平方米，规划建面3.66万 平方米，容积率为2.5。该地块吸引了上海城建、中海、保利发展、越秀、招商蛇口和财投联合体、华 润置地和中能建联合体共6个竞买人参与竞价，经过95轮竞价，最终由保利发展以总价35.00亿元竞得， 成交楼面价为95530元/平方米，溢价率31%。 超过9.5万/平方米的楼面价，再次刷新杨浦区单价地王纪录。2020年7月，来自北京的国企金隅集团进 军上海 ...

Group 1 - The Guangdong Provincial Market Supervision Administration released an announcement regarding 18 products that failed quality supervision checks and have not been corrected within the stipulated time [1][2] - Three companies from Jiangmen were specifically named for producing four non-compliant products, which include commercial gas burners and lighting fixtures [1] - The companies identified are: - Kaiping Dingli Kitchenware Co., Ltd. with the SZCT30-A commercial gas frying stove failing on thermal load accuracy - Kaiping Baihe Town Lianxi Kitchenware Factory with the STCR35-C2 gas frying stove failing on grounding measures - Guangdong Minjiang Aquatic Industry Co., Ltd. with MJ-LED45AW4.5W and MJ-LED60AW6W LED aquatic lights failing on structure (including light biological hazards), dust, solid foreign matter, and waterproofing [1][2] Group 2 - The products from the three Jiangmen companies were deemed non-compliant after supervision checks, and despite being ordered to rectify the issues, they remained non-compliant [2] - The announcement states that the local market supervision department will oversee a comprehensive inspection and correction of these products, with a re-inspection scheduled 60 to 90 days after the announcement [2] - Producers are prohibited from manufacturing or selling the same products until they are certified as compliant by the market supervision department, and non-compliance after re-inspection may lead to business suspension or revocation of business licenses [2]