本文来源:https://www.eepw.com.cn/article/202609/483668.htm
若有版权问题,请来信告知:13951431913#139.com 或在右方提交评论。
9月22日,2026英特尔工艺创新与产业生态大会在苏州国际博览中心开幕。英特尔在大会上发布具身智能开发套件,并首次以官方口径披露18A-P制程关键数据。大会为期两天,设11场主题演讲、80余场工艺课堂、15000平方米展区,展区汇集1400件创新演示。
英特尔公司首席执行官陈立武
英特尔CEO陈立武以视频形式致辞。他在致辞中重申英特尔的五张牌:CPU、GPU、ASIC、先进封装、晶圆制造,加上覆盖云、边、端的开放生态。
英特尔中国区董事长王稚聪
英特尔中国区董事长王稚聪在现场的表述更直接:“把产业界的‘各有所长’聚成‘共创所成’。”
英特尔首席营销与传播官Annie Shea Weckesser
英特尔首席营销与传播官Annie将战略拆为三条:开放计算平台覆盖PC、边缘和数据中心;围绕特定工作负载外观定制化芯片;借助制程与先进封装达成更复杂的平台集成。
英特尔副总裁、端侧计算和物理AI事业部中国区总经理高嵩
端侧是当天硬新闻最集中的板块。英特尔公司副总裁兼端侧计算和物理AI事业部中国区总经理高嵩宣布,18A-P制程将达成每瓦表现提高9%、同等表现功耗降低18%。基础版18A已于同日进入量产,18A-P进入风险试产。
搭载第三代酷睿Ultra处理器的具身智能开发套件同步发布,联合科通工业、神州数码推出,实时控制平均抖动降至1.6微秒。高嵩称,该套件让开发者“开箱即用,快速跑通感知—规划—执行全链路”。
第三代酷睿Ultra及酷睿系列处理器自发布以来已获超400项客户外观。新一代锐炫Pro B70/B65面向端侧与边缘AI计算。软件层面,英特尔推出Web Engine,用于提高智能体复杂网络任务成功率,同时降低内存占用。应用侧,英特尔AI PC智能体伙伴入驻WorkBuddy,集成智效办公、智享生活、智创内容、智优电脑四个板块;OS层面,英特尔联合阿里云千问预览Qwen Book智能体原生桌面操作平台的深度适配。
情形落地覆盖办公、创作、游戏、家居出行和垂直行业。办公情形中,智能体可自主完成任务拆解、工具调用、知识读取和结果迭代,在关键节点与人确认。创作外观领域,零刻AI NAS搭载第三代酷睿Ultra、飞牛OS与Marvis AI思路;小云雀利用本地算力完成AI视频超分;锐炫Pro B70/B65 32GB显存配合MiniMax H3模型达成本地AI视频与漫剧制作。垂直行业中,MAXHUB、北科瑞声与亿道推出医疗智能体思路,问诊语音可自动生成病历初稿;石基的酒店智能体AI Box将智能体嵌入运营工作流,数据留在企业内部;奇矩智能的工业智能体AI Box称可提高产线稼动率,换线耗时缩短,产线提前3小时开工。
英特尔数据中心与人工智能集团副总裁、中国区总经理陈葆立
数据中心维度由英特尔数据中心与人工智能集团副总裁、中国区总经理陈葆立补全。他提出“以CPU为中枢建设面向智能体AI时代的智算中心”,达成CPU、GPU、存储、网络全面协同。阿里云已推出超过117种基于英特尔至强6处理器的云实例类型,其中超过50种应用于智能体AI情形。百度搭子、Marvis、千问办公、TRAE、Workbuddy与英特尔在模型路由、推理框架优化和本地技能集成上合作。节卡机器人将实时运动控制、感知、自主规划与执行整合到英特尔边缘计算架构中。
圆桌对话
另外,在圆桌环节英特尔市面营销集团副总裁、中国区总经理郭威与联想集团副总裁李伟昌、中国移动咪咕公司副总经理许杰、京东科技副总裁刘洋同台,讨论智能体规模化落地的障碍与路径。
展区设“算力中枢·AI工厂”、企业AI、行业AI、物理AI、游戏家庭个人AI、“芯动之源”及AI开发者专区,覆盖制造、能源、交通、医疗、教育、零售等情形,同期举办70余场互动分享。
面对端侧智能体机遇,笔者看到了英特尔坚持开放生态,携手广泛的OEM/ODM合作伙伴、模型厂商、ISV及平台集成商,共同构建全栈应对思路,推动端侧智能体在本地及混合情形下的规模化落地,共同拓展产业空间。
四十余年来,英特尔始终与中国产业伙伴携手推动计算工艺创新。面向智能体AI时代的新机遇,笔者相信英特尔将依托开放平台长处,促进工艺、产业与生态协同创新,把行业经验、客户需要与工程能力更紧密地连接起来,为产业创新注入新动能。
Source: https://www.eepw.com.cn/article/202609/483668.htm If there are any copyright issues, please contact me at: 13951431913#139.com or submit a comment on the right.
On September 22, the 2026 Intel Process Innovation and Industry Ecosystem Conference opened at the Suzhou International Expo Center. At the conference, Intel released the Embodied AI Development Kit and for the first time officially disclosed key data on the 18A-P process. The two-day event features 11 keynote speeches, over 80 process workshops, and a 15,000-square-meter exhibition area showcasing 1,400 innovative demonstrations.
Pat Gelsinger, CEO of Intel
Intel CEO Pat Gelsinger delivered a video speech in which he reiterated Intel’s five major advantages: CPU, GPU, ASIC, advanced packaging, and wafer manufacturing, all supported by a comprehensive open ecosystem covering cloud, edge, and endpoint.
Wang Zhicong, Chairman of Intel China
Wang Zhicong, Chairman of Intel China, expressed this more directly: “Transforming the industry’s ‘individual strengths’ into ‘collective achievements’.”
Annie Shea Weckesser, Intel’s Chief Marketing and Communications Officer
Annie Shea Weckesser, Intel’s Chief Marketing and Communications Officer, broke down the strategy into three parts: an open computing platform covering PCs, edge devices, and data centers; customization of chips for specific workload appearances; and the achievement of more complex platform integration through process and advanced packaging technologies.
Gao Song, Vice President of Intel and General Manager of the Client Computing and Physical AI Business Unit in China
The edge computing segment was the focus of the most significant announcements. Gao Song, Vice President of Intel and General Manager of the Client Computing and Physical AI Business Unit in China, reported that the 18A-P process will improve performance per watt by 9% and reduce power consumption by 18% while maintaining equivalent performance. The base version of 18A entered mass production on the same day, with 18A-P entering risk production.
The company also introduced the Embodied AI Development Kit, built on the third-generation Core Ultra processor. Collaborating with Kotong Industry and Founder Technology, this kit achieves an average latency of 1.6 microseconds for real-time control. Gao stated that the kit allows developers to “quickly run the full chain of perception, planning, and execution right out of the box.”
Since its release, the third-generation Core Ultra and Core series processors have secured over 400 customer designs. The new generation of Arc Pro B70/B65 is aimed at edge and endpoint AI computing. On the software side, Intel has launched a Web Engine to improve the success rate of complex network tasks for intelligent agents while reducing memory usage. Regarding applications, Intel AI PC partners have integrated features into WorkBuddy, covering efficient office work, smart living, creative content, and optimized computing. At the OS level, Intel has worked with Alibaba Cloud to preview the deep integration of the Qwen Book intelligent agent native desktop operating platform.
The real-world applications span office, creative, gaming, home, travel, and vertical industries. In the office setting, intelligent agents can autonomously handle task breakdown, tool invocation, knowledge retrieval, and result iteration, checking with humans only at critical junctures. In the creative field, the ZKNV AI NAS utilizes the third-generation Core Ultra and Marvis AI; Qianshui relies on local computing power for AI video upsampling; while the Arc Pro B70/B65 32GB VRAM enables local AI video and animation production with the MiniMax H3 model. In vertical markets, MAXHUB, North Technology Voice, and Edao have created medical intelligent agents that can automatically generate preliminary medical records from voice data. Shiji’s hotel intelligent agent AI Box embeds intelligence into operational workflows, keeping data internal, and Qiju Technology’s industrial intelligent agent AI Box claims to improve production line uptime and reduce changeover times, allowing the line to start production 3 hours earlier.
Chen Baoli, Vice President of Intel Data Center and AI Group and General Manager for China
Chen Baoli, Vice President of Intel Data Center and AI Group and General Manager for China, addressed the data center aspect. He proposed building an “intelligent computing center for the AI era with a CPU as the hub,” ensuring full synergy between CPUs, GPUs, storage, and networks. Alibaba Cloud has deployed over 117 types of cloud instances based on Intel Xeon 6 processors, with more than 50 of these used in intelligent agent scenarios. Companies like Baidu, Marvis, Qwen Office, TRAE, and Workbuddy are collaborating with Intel on model routing, inference frame optimization, and local skill integration. Jeka Robotics has integrated real-time motion control, perception, autonomous planning, and execution into the Intel edge computing architecture.
Roundtable Discussion
Additionally, a roundtable discussion was held where Intel Market Marketing Group Vice President and China General Manager Guo Wei, Lenovo Group Vice President Li Weichang, China Mobile Migu Company Deputy General Manager Xu Jie, and JD Technology Vice President Liu Yang discussed the obstacles and paths to the widespread adoption of intelligent agents.
The exhibition area is divided into sections such as “Computing Hub • AI Factory”, Enterprise AI, Industry AI, Physical AI, Gaming, Home, Personal AI, “Chip Source”, and the AI Developer Zone. It covers scenarios in manufacturing, energy, transportation, healthcare, education, and retail, with over 70 interactive sharing sessions held in parallel.
Facing the opportunities presented by edge-based intelligent agents, I can see that Intel is adhering to a philosophy of an open ecosystem to collaborate with a wide range of OEM/ODM partners, model vendors, ISVs, and platform integrators. Together, they are building a comprehensive solution to drive the large-scale deployment of edge-based intelligent agents in both local and hybrid scenarios, thereby expanding the space for the industry.
Over the past forty years, Intel has been partnering with Chinese industrial counterparts to drive computing technology innovation. Looking toward the new opportunities of the intelligent agent AI era, I believe that Intel will leverage its open platform advantages to foster collaborative innovation across processes, industries, and ecosystems. By more closely connecting industry expertise, customer needs, and engineering capabilities, Intel will inject new energy into industrial innovation.







