Arm Technology’s Anmou (Arm China) shines at WAIC | An AI-forward sharing session—Arm is everywhere, making AI within reach

ME AI news, July 19, 2026, Shanghai, China -- Arm Technology (China) Co., Ltd. (hereinafter referred to as "Arm China"), a leading domestic chip IP design and service provider, recently announced its debut at WAIC. Working with the Arm ecosystem, it presented a complete AI deployment blueprint covering three major business exhibition areas—Edge AI, Physical AI, and Cloud AI—systematically showcasing the full AI layout from compute platforms to development boards and terminal applications. It explains how pervasive Arm technology drives large-scale, real-world adoption of AI across industries on a massive scale. The Future of AI is Built on Arm! At a concurrent AI foresight conference, multiple technical experts from Arm China delivered keynote speeches, providing an all-around display of Arm China’s forward-looking insights and innovative technologies for the AI era. Keeping pace with AI development trends, Arm China has built a "Star-300" AIoT prototype platform, a "Zhouyi" X3-Pro NPU end-to-end efficient computing foundation for the Agentic AI era, "Linglong" AI VPU end-to-end technologies, and has taken the lead in initiating the "Open-source AIOS Alliance," laying a solid technical foundation for innovation in the AI era.

【Highlights of the AI foresight conference: CPU, NPU, VPU, and AIOS work together to build a solid technical foundation for AI across all scenarios】

Building the "Star-300" AIoT prototype platform: expanding AI compute power from traditional MCUs
The IoT ecosystem is rooted in the Arm architecture. Over the years, Arm has followed market demands and technological developments, and has already built a rich set of IP solutions for the embedded computing field, mainly including the Arm® Cortex®-M CPU processor series, Arm Helium™ vector extension technology, and the AI accelerator Arm Ethos™ NPU series. Leveraging Arm’s mature ecosystem, Arm China has built an AIoT prototype platform for embedded AI scenarios based on "Star" CPU + Arm Helium vector extensions + Arm Ethos NPU.

Zhu Xiaoming, Director of CPU Products at Arm China, said that the first-generation platform selected STAR-MC2 (i.e., Cortex-M52) and Ethos-U55, with the codename "Star-300." It fully supports mainstream small models, extending AI compute power beyond traditional MCUs and adding an AI engine to embedded devices.

Next-generation "Zhouyi" NPU roadmap: a unified, layered, and scenario-driven computing architecture that takes AI agents into the real world
The Agentic AI era has arrived. AI model capabilities—especially those of small and medium models—continue to improve; inference paradigms are being reshaped; inference workloads are highly differentiated; and AI inference needs on edge and end-side scenarios vary widely. These changes impose entirely new requirements on the AI computing foundation. Arm China has built an all-scenario, high-efficiency computing foundation for Agentic AI—"Zhouyi" X3-Pro.

Dr. Shu Hao, head of the NPU product line and chief architect at Arm China, said, "Zhouyi" X3-Pro NPU adopts a unified, layered, scenario-driven computing architecture. Its design philosophy is 'one-time adaptation, multi-core reuse; flexible configuration for different scenarios.' In multi-industry and multi-product scenarios, it can reuse the work and experience for model adaptation, tuning, and deployment, and efficiently meet differentiated requirements across various application scenarios. This will significantly improve R&D efficiency, accelerate innovation adoption, and enable AI agents to truly enter every edge-side device.

"Wudang" unveiled! AI VPU technology helps humans see the world so machines can "understand" the vision
In the AI era, AI itself is becoming the biggest consumer of video. Video processing needs are upgrading toward higher resolutions, better picture quality, and lower latency. This requires strong VPU capabilities to efficiently encode/decode massive volumes of video. Arm China’s next-generation "Linglong" VPU "Wudang" was revealed for the first time. The product fully inherits the CAE technology of the previous generation. While maintaining leading encoding quality, its PPA has improved by over 40% compared with the prior generation, and the encoder specifications are also more comprehensive.

Dr. Huang Xin, R&D Director of VPU at Arm China, said that the "Linglong" VPU AI encoding technology continues to evolve. Arm China has built end-to-end AI video encoding technology for the AI side. Based on a new AI VPU training framework, it realizes "end-to-end global optimization," moving from enhancing "subjective visual quality" to "empowering AI across all possibilities." "Linglong" VPU is committed to becoming an AI vision engine—helping humans see the world and enabling AI to understand the vision.

Taking the lead in initiating the "Open-source AIOS Alliance" to explore the next generation of AI product infrastructure
Currently, new-generation AI technologies represented by AI agents and embodied intelligence continue to break through. Various new computing carriers keep emerging, and operating systems are accelerating their evolution from traditional tool-based forms toward agentic forms. AIOS intelligent systems specially built natively for these new AI products and embodied intelligence will become a key support for unlocking edge-side AI potential and accelerating widespread, accessible AI applications. Arm China officially announced its initiative to establish the "Open-source AIOS Alliance." It has already attracted more than 20 ecosystem partners to join. Together, they will explore the next generation of AI product infrastructure and build an open-source AIOS system foundation.

Liang Quan, Senior Vice President for Market and Ecosystem at Arm China, said that Arm China will fully leverage Arm’s rich global ecosystem resources, the advantages of end-to-end cloud-edge-edge-stack technologies, and model capabilities to provide solid underlying technical support for innovation and evolution in next-generation AI products, and promote large-scale deployment of AIOS across diversified forms of emerging smart terminals with ecosystem partners.

【Three major business exhibition zones showcase highlights: Edge AI, Physical AI, Cloud AI—Arm technology empowers AI across all scenarios】

Arm technology is making AI within reach. At Arm China’s booth, three major business exhibition areas were set up: Edge AI, Physical AI, and Cloud AI. From smart wearables and smart home to AI PCs, smartphones, smart cars, robots, and data centers, the exhibition concentrated on showcasing diversified AI applications enabled by Arm technology across industries.

Edge AI exhibition zone
Arm provides Edge AI with a full-stack solution—from IP, through compute subsystems, to software. For example, the Arm Lumex compute subsystem and Arm Kleidi™ software applications are used in smartphones; the Arm terminal compute subsystem is deployed in AI PCs; Cortex-M CPU + Helium + Ethos NPU are adapted for embedded devices; and Armv9 technology applies to smart homes.

In the exhibition zone, visitors can see "how the chip is formed," as well as chip IP and end-application showcases covering smart wearables, consumer electronics, intelligent office environments, smart homes, smart healthcare, smart industry, and other fields. "One sand, one world; one chip enables thousands of industries" clearly illustrates how Arm technology empowers edge-side AI.

On-site, three major solutions were重点展示:

• TTS models from the Qwen-Audio family based on Arm technology,
• an AI MCU product matrix, and
• the JishuShell one-click deployment tool for edge-side agents.

TTS models of the Qwen-Audio family: based on Arm SME2, the Qwen-Audio family’s TTS models can run efficiently on a single Arm CPU, bringing low-latency, high-quality AI voice experiences to Arm devices.

AI MCU专区: focused display of a solution matrix based on Arm Cortex-M CPU + Helium + Ethos NPU, covering diverse applications such as in-vehicle monitoring, traffic supervision, AI companion toys, visual tracking, gesture recognition, and industrial predictive maintenance. The Arm China "Star-300" AIoT prototype platform is also demonstrated on an FPGA platform, including applications such as object detection, speech recognition, keyword spotting, and super-resolution.

JishuShell joint exhibition zone: on-site demonstrations feature development boards showing and running the one-click deployment tool for end-side agents, along with hands-on practice demonstrations. JishuShell leverages the developer ecosystem to quickly enter the Agentic AI field, and works with Arm ecosystem chip and solution vendors to explore AI applications and interactions for next-generation edge-side devices. In the future, it will further integrate AIOS capabilities.

Physical AI exhibition zone
Arm provides Physical AI with full-stack support including IP, compute subsystems, interconnect technologies, tools, and integrated software. Among them, the Arm Zena™ compute subsystem and the AE (Automotive Enhanced) technology are both applicable to intelligent vehicles and robotics fields.

The Physical AI exhibition zone is one of the most popular areas on-site, concentrating on showcasing Arm technology achievements in two major domains: robots and intelligent vehicles.

More than ten exhibits form a "robot group," including humanoid robots, emotion-interaction robots, robotic arms for latte art, robotic dogs, and robotic panda-like machines, which all appeared together. Arm technology runs through every aspect of robots—central brains, interaction computing, battery management, local control, sensors, and actuators—to form a complete system support.

Multiple intelligent automotive chips and intelligent interconnect computing platforms appeared simultaneously. Arm technology covers five domains: powertrain, chassis, body, cockpit, and intelligent driving, helping the automotive industry achieve full-domain intelligent upgrades.

Cloud AI exhibition zone
With leading "performance per watt" advantages, the Arm Neoverse® computing platform is becoming a key cornerstone for cloud computing and AI infrastructure, providing system deployment solutions that balance performance, efficiency, and security for scenarios such as cloud computing and data centers. Currently, among the compute power shipped to top-tier hyperscale cloud service providers, nearly 50% is based on the Arm architecture.

This Cloud AI exhibition zone focuses on demonstrating DPU chips based on Arm technology, smart NICs, SuperNIC cards, and server racks in multiple types such as 2U, 4U, and 6U, covering the complete chain from network acceleration to full-system deployment.

Not just hardcore technology—there is also AI temperature. The booth also sets up rich interactive experiences: robotic arms grind coffee at a coffee bar and perform precise latte art; brain-computer interface challenges with a fun "mind-popcorn" concept bring novel interaction experiences.

These playful interactions take AI from the booth to everyday life, allowing the public to personally feel the innovation charm of AI. Looking ahead, under the guidance of the "AI Arm China" strategic direction, Arm China will use Edge AI, Physical AI, and Cloud AI as the main lines to advance the "All in AI" product strategy, build a solid compute-power foundation, and work with ecosystem partners to empower the AI era.

(Source: Ifnar)