Juchip Technology: Low Power + Computational Capability of Edge SoC

D20250702-Dongwu Securities-Juchip Technology-688049-Smart Audio Empowering AIoT Integrated Computing Growth.pdf20250626-Hua Chuang Securities-Juchip Technology-688049-Deep Research Report First Release CIM Product Positioning Edge AI Golden Track New Product Volume Performance Continues to Grow.pdfRockchip: High-Performance SoCEspressif: IoT Wi-Fi MCUK

(The following K chapter content is AI-generated and can be skipped)

1. Key Points of Juchip Technology’s Business Logic:

1. Actions Intelligence Strategy: Juchip Technology’s edge AI strategy for battery-powered IoT devices aims to provide 0.1-1 TOPS general AI computing power at a power consumption of 10mW-100mW, challenging the 10-100 TOPS/W energy efficiency ratio limit.

2. Low Power AIoT Chip Designer: The company’s market positioning is as China’s leading low power AIoT chip designer, focusing on wireless audio, smart wearables, and smart interaction, adopting a Fabless light asset model.

3. Compute-In-Memory (CIM): A technology path that breaks through the Von Neumann architecture’s “memory wall” bottleneck by performing computations directly within the memory unit, which is the core principle for achieving ultra-low power AI computing power.

4. MMSCIM Architecture (Mixed-Mode SRAM CIM): Juchip Technology’s core technological innovation, realizing digital computing functions using customized analog circuits within SRAM media, combining high energy efficiency with mass production consistency.

5. Tri-Core Heterogeneous AI SoC Architecture: The underlying design of Juchip’s chip integrates CPU + DSP + NPU (based on MMSCIM), where the NPU single core provides 100 GOPS computing power, achieving an energy efficiency ratio of 6.4 TOPS/W @ INT8.

6. Energy Efficiency Ratio (TOPS/W): The core performance metric for edge AI chips, Juchip’s first-generation products achieve 6.4 TOPS/W, with the third generation (2026) targeting 15.6 TOPS/W, far exceeding the theoretical limit of 10 TOPS/W for traditional architectures.

7. Commercialization of Technology: Transforming the MMSCIM architecture into mass-producible chips and successfully integrating into the supply chains of international top brands such as Harman, Sony, Xiaomi, Bose, BYD.

8. Top Brand Penetration: Market strategy results show that the revenue share from the top five customers exceeds 56%, creating a technology benchmark effect that supports product premium pricing and channel expansion.

9. Solution Provider Sales Model: The main channel strategy is to cover downstream solution providers through distributors, with the top five customers accounting for about 80% of revenue, achieving rapid volume growth.

10. Smart Wireless Audio SoC: The company’s foundational product, based on a tri-core heterogeneous architecture AI audio platform, achieving a gross margin of 70.82%, contributing the majority of revenue and profit.

11. Edge AI Processor Chip: The second growth curve product, targeting new scenarios such as smartwatches, AI glasses, and wireless home theaters, with an annual growth rate exceeding 100%.

12. Second Growth Curve: Expanding from traditional audio chips to edge AI processors, opening up a hundred billion-level incremental market and reducing dependence on a single business.

13. Fabless Light Asset Model: Focusing on chip design and R&D without engaging in manufacturing, achieving production through foundry partners, realizing high operational leverage.

14. Scale Effect: Under the Fabless model, revenue growth far exceeds expense growth, with net profit growth in Q1-Q3 2025 being 2.1 times that of revenue growth, and net profit margin increasing to 20%+.

15. Patent Moat: As of March 2025, holding 337 patents, covering MMSCIM core technologies and processes, creating dual barriers of legal and technical protection.

2. Business Logic Tracking Verification

Vision: Juchip Technology is upgrading from selling chips to selling standardized capabilities of AI computing power per unit power consumption, continuously reducing the cost of this capability through the MMSCIM architecture, ultimately becoming a computing power infrastructure provider for battery-powered edge AI devices.

Preliminary Verification: The combination of the Fabless light asset model and scale effect has resulted in Q3 2025 net profit growth (+113.8%) being 2.1 times the revenue growth (+54.7%), and this profit elasticity supports a R&D expense ratio consistently above 25%, forming a positive cycle of technology and business.

(The following K chapter content is AI-generated and can be skipped)I

The essence of SoC is to integrate the functions of a complete computer motherboard into a single chip to meet the diverse needs of edge scenarios. Therefore, selling SoC chips is essentially delivering scenario-based application solutions.

The core of its business logic lies in the triangular trade-off of Performance, Power, and Area, achieving architecture-level precise optimization for different application scenarios. Juchip Technology’s core technology anchor is the extreme low power computing capability.

Juchip Technology’s foundation lies in audio signal processing. This field is undergoing a paradigm shift from traditional signal processing to “computational audio”— reshaping the sound signal processing process with dense computing power, a logic that parallels the evolution of the photography industry towards “computational photography” (see the acceleration of computational photography: STMicroelectronics, Lantech Optics, ArcSoft Technology). Technological iteration is reconstructing the industry landscape: traditional manufacturers in Europe, America, and Japan are accelerating decline, while Chinese companies are rising in brand voice and supply chain status.

Juchip Technology’s second growth curve is currently positioned in smartwatches and AI glasses and other wearable devices. The company effectively breaks the “computing power bottleneck in low power scenarios” with its Compute-In-Memory (CIM) architecture, thereby establishing an unreplicable competitive advantage in the wearable field.

C

1. The audio market has seen a slight growth in total scale, considering that the introduction of computational audio has significantly increased the value per unit, so the overall market is still growing, and with Juchip Technology’s enhanced market competitiveness, its share will also increase.

In the wearable market, Juchip Technology’s low power design architecture has a certain technological lead, and we are optimistic about its future.

2. The edge SoC chip market, previously dominated by European, American, and Japanese brands, is now led by mainland and Taiwanese brands, all competing fiercely. However, Juchip Technology’s technology, products, and customers are relatively high-end, so the competitive landscape is slightly better.

3. Revenue growth of 54.74%, net profit growth of 113.85%. Consensus expectations for 2026 are 39.66%, and for 2027, 35.15%. Currently, the PE ratio is 47 times, which is moderate.

4. There are many catalytic opportunities related to semiconductors, AI, and wearable devices.

Rockchip, Juchip Technology, and Espressif are currently the three most promising edge SoC suppliers.

Rockchip: High-Performance SoC

Espressif: IoT Wi-Fi MCU

Mark:688049$Close == 51.00 @ 2025/11/25