Edge AI: Reconstructing the underlying logic of intelligent terminals.
Edge AI (Edge AI) achieves localized data processing by embedding the computing, reasoning, and decision-making capabilities of artificial intelligence directly into terminal devices (such as smartphones, smart cars, industrial robots, etc.). Compared to traditional architectures that rely on cloud services, its core advantages are:
Low latency response: Local processing makes it possible for scenarios with high real-time requirements (such as autonomous driving and industrial control);
Privacy and security: Sensitive data does not need to be uploaded to the cloud, reducing the risk of leakage;
Offline availability: Basic functions can still be maintained in weak or no network environments.
Currently, Edge AI has formed a full-link ecosystem of “hardware-algorithm-scenario,” covering three major levels:
Upstream: AI chips, sensors, communication modules, and operating systems;
Midstream: Device manufacturers with model optimization and system integration capabilities;
Downstream: Solution providers focusing on vertical industries.
Shipments of terminals such as AI PCs and AI smartphones are surging (by 2025, domestic AI PC shipments will exceed 100 million units, with AI smartphone penetration exceeding 40%).
Eight Core Targets: Exemplars of Technical Barriers and Scene Deepening!
The following A-share companies have become direct beneficiaries of the Edge AI wave due to their technological advantages and ecological layout:
1. Rockchip (603893): Heterogeneous computing architecture
Technical highlights: The flagship chip RK3588 adopts an 8nm process, integrating a self-developed NPU with 6TOPS, supporting multi-modal perception;
Application scenarios: Automotive intelligent cockpits, machine vision, industrial robots;
Performance: In the first half of 2025, revenue reached 2.046 billion yuan (+64%), with a net profit of 531 million yuan (+191%), and the proportion of mid-to-high-end products increased.
2. Allwinner Technology (300458): Audio and video SoC
Technical highlights: Leading domestic audio and video main control chip, layout in home and automotive intelligence;
Performance highlights: In 2024, revenue reached 2.288 billion yuan (+36.76%), with a net profit of 167 million yuan (+626%), with significant breakthroughs in the smart automotive electronics field.
3. Hengxuan Technology (688608): Intelligent audio SoC
Technical highlights: RISC-V architecture chip market share exceeds 30%, new generation wearable chips in mass production;
Performance: In the first half of 2025, revenue reached 1.938 billion yuan (+26.58%), with a net profit of 305 million yuan (+106%).
4. Amlogic (688099): Multimedia SoC
Technical highlights: Global market share of smart set-top box chips exceeds 60%, T series high-end chips support 8K decoding;
Client coverage: Major companies such as Xiaomi, Amazon, and Google.
5. Zhaoxin (688332): Low-power Bluetooth chips
Technical highlights: Released the BT897x series supporting Bluetooth 6.0 and LE Audio, entering the supply chain of Xiaomi and Honor;
Market positioning: Leading in the subfield of TWS headphones and smart wearable devices.
6. Goke Microelectronics (688049): High-end audio SoC
Technical highlights: Edge AI processor chip in mass production, entering the supply chain of Huawei and Harman;
Performance: In the first half of 2025, revenue reached 449 million yuan (+60%), with a net profit of 91 million yuan (+123%).
7. Fuhuan Microelectronics (300613): Security monitoring
Technical highlights: Leading domestic security chip, significant advantages in ISP technology;
Cooperative ecosystem: Deeply bound with Hikvision and Dahua, covering all scenarios of smart security.
8. StarChip (300807): Video security chips
Technical highlights: Self-developed AI ISP technology recognition accuracy of 98%, supporting local deployment of multi-modal large models;
Mass production of smart vehicle chips for BYD and Geely models, with ADAS orders increasing by 240% year-on-year in 2025;
Winning bids for smart city projects exceeding 500 million yuan, with a TTM price-to-earnings ratio of only 35 times (industry average 60 times).
Edge AI drives SoC chips towards the evolution of “three bodies”: computing power upgrade (NPU optimization), algorithm adaptation (heterogeneous computing), and ecosystem construction (chip + algorithm + scenario).
Recommendations:
Automotive intelligence (Rockchip, StarChip);
Video security (Fuhuan Microelectronics, StarChip);
Smart wearables (Hengxuan Technology, Goke Microelectronics).
Daily sorting of popular tracks and high-potential companies, click to follow, send 126 to get exclusive in-depth analysis!
This article does not constitute investment advice. The market has risks, and decisions should be made cautiously.