On November 12, Zhuhai Amicro Technology Co., Ltd. officially listed on the Main Board of the Hong Kong Stock Exchange. The stock closed up 38.18% on its first day, with its market capitalization reaching HKD 5.9 billion.
The unique positioning of Amicro Technology lies in its industrial role; it is neither another robot OEM nor a pure-play chip design company. According to Frost & Sullivan, Amicro Technology is China's third-largest intelligent mobile robot solution provider and the largest single-modality robot solution supplier. A more accurate description might be that it operates in the business of the robot's "cerebellum," packaging foundational perception and execution capabilities—such as localization, mapping, and motion control—into standardized chip-level modules sold to downstream OEMs.
Why "Algorithm-on-Chip"?
The core of Amicro Technology's technical narrative lies in a counter-intuitive strategic choice. Most robot algorithm companies follow the path of "software licensing" or "cloud-side services," which offer low marginal costs and high gross margins but weak bargaining power and difficulty in deeply integrating with customer hardware. Yiwei chose a heavier path by hardening algorithms into chips.
In 2016, it mass-produced inertial navigation solutions; in 2019, it implemented the Alpha SLAM single-modality solution; and in 2024, it achieved mass production of the Ultra SLAM multi-modal solution. Behind this timeline is a complete "perception-decision-control" closed loop: the self-developed SLAM SoC handles mapping and localization, the AI SoC performs edge-side visual computing, self-developed sensors and algorithms achieve parameter-level coordination to eliminate losses from matching third-party components, and motor and PMIC (Power Management IC) solutions ensure motion accuracy.
The industrial significance of this architecture lies in lowering the barrier for downstream players. For manufacturers of robotic vacuums and commercial service robots, developing a reliable SLAM system in-house requires maintaining an algorithm team of dozens of people over a period of years. Yiwei packages this capability into a "turnkey" solution, allowing customers to focus solely on whole-machine integration and scenario tuning. Eight of the world's top ten intelligent cleaning companies adopt its solutions, with cumulative shipments reaching tens of millions of units, demonstrating that this logic has already proven commercially viable.
In terms of computing metrics, its AM970 chip integrates approximately 5 TOPS of edge-side AI computing power, exceeding the industry average of 3.5 TOPS. It has also disclosed plans to mass-produce programmable NPUs supporting Transformer architectures. This points to a clear industrial trend: edge-side computing in robotics is replicating the computing arms race seen in mobile devices, where the energy efficiency advantage of dedicated chips cannot be easily replaced by general-purpose GPUs.
Product Structure Upgrade Behind Gross Margin Leap
In Yiwei's financial data, what merits the most attention from industry observers is not revenue growth speed, but the jump in gross margin. The comprehensive gross margin was only 6.1% in 2023 but rose to 27.7% in 2025. In the hardware design industry, a nearly five-fold recovery in gross margin over three years is extremely rare.
Breaking it down, the driving force is not merely cost control but an upward migration in product structure. Single-modality solutions address basic localization and navigation, characterized by relatively limited technical barriers, intense competition, and thin margins. Multi-modal solutions, however, bundle more complex sensor fusion, higher-computing-power chips, and a more complete toolchain. In 2025, the gross margin for this category reached 44.0%. Revenue from multi-modal solutions hit CNY 181 million in 2025, accounting for 44.5% of total revenue and becoming the core growth engine.
Downstream robot manufacturers are upgrading from "just moving" to "moving accurately, understanding vision, and adapting to complex scenarios." As embodied intelligence moves from demo videos to actual deployment, the demand for multi-modal perception is rigid. Yiwei occupies the critical layer of chip-algorithm coupling that is unavoidable in this upgrade process.
Hengqin's Chess Piece and a Footnote to Domestic Substitution
Another layer of value for Amicro Technology lies not in the financial statements, but in regional industrial layout.
Established in Hengqin in 2014, it is the first national-level "Specialized, Refined, Distinctive, and Innovative" Little Giant enterprise cultivated in the cooperation zone and the second locally incubated listed company in Hengqin. Its development path is summarized as "R&D in Macau — Transformation in Hengqin — Manufacturing in Zhuhai — Global Application." Joint laboratories at the University of Macau undertake cutting-edge algorithm research, Hengqin hosts industrialization implementation, Zhuhai completes manufacturing, and subsequently connects with global capital via the Hong Kong capital market. This cross-jurisdictional collaborative model holds sample significance among Greater Bay Area hard-tech enterprises.
From the perspective of domestic substitution, the gap filled by Yiwei is specific: the world's first dedicated SLAM SoC for intelligent mobile robots. Previously, this category was long served by general-purpose MCUs combined with software solutions or relied on general-purpose vision chips from overseas vendors. The significance of dedicated chips lies not in "existence" but in "usability." Robot scenarios impose stringent requirements on power consumption, real-time performance, and sensor synchronization, which general-purpose chips struggle to optimize deeply at the system level. Yiwei's self-developed sensors and algorithms achieving parameter-level coordination follow the vertical integration logic validated by Apple and Tesla.
Approximately 70% of the proceeds from this IPO will be invested in deepening technology for software, hardware, and robot-specific AI SoCs, clearly targeting embodied intelligence and advanced perception. Considering that the Chinese intelligent mobile robot solution market is expected to grow from CNY 9.7 billion in 2025 to CNY 38.1 billion in 2030, with a compound annual growth rate of 31.4%, the beta of the sector itself is substantial. The question remains whether Yiwei can convert its first-mover advantage in chip definition into sustained ecological niche dominance amidst the squeeze between top-tier customers' in-house R&D and latecomers' catch-up efforts.
In Conclusion
A company making "cerebellum" chips for robots has proven that the heavy path of "algorithm-on-chip" is feasible within China's hardware ecosystem, while also exposing the structural vulnerability of solution providers under the shadow of giants.
As growth in the robotic vacuum market slows, humanoid robots pose entirely new requirements for chip architectures, and top-tier customers begin seriously considering Plan B for in-house SoC development, can Yiwei evolve from being the "king of vacuum cleaner chips" to becoming the underlying computing platform for the era of embodied intelligence? The answer to this question is not in the prospectus, but in the volume orders over the next two to three years.
All content in this article is derived from public information and does not constitute any investment advice. If you have any questions, please contact us (WeChat ID: global360iot01).