A research report from Orient Securities Company Limited (DFZQ) highlights that the AI computing power supply chain is evolving towards higher sophistication and complexity. This trend is rapidly increasing the product value and manufacturing intricacy of core components such as optical modules, high-end PCBs, IC substrates, server power supplies, and connectors. Rising product complexity is not only expected to increase the quantity of inspection equipment needed but also to drive upgrades in the equipment itself. This dynamic is likely to boost the value per production line and enhance customer loyalty. The essential inspection requirements driven by AI hardware upgrades, combined with the trend of domestic substitution, position domestic inspection equipment and key component manufacturers to benefit.
AI Hardware Value Drives Shift to Full Inspection, Elevating Quality Control Importance
The AI computing power supply chain's ongoing evolution towards higher-end, more complex products is significantly raising the value and process complexity of its core components. For instance, the expansion of AI cluster scale is accelerating the upgrade of optical modules from 400G to 800G and 1.6T. New technology pathways like silicon photonics and CPO impose stricter demands on photoelectric coupling precision, signal consistency, thermal stability, and long-term reliability. In the case of AI server PCBs, trends include high frequency and speed, multiple layers, low loss, and high-density interconnects, resulting in substantially higher per-board value compared to traditional servers. For such high-value products, the scrap cost, rework expense, and customer compensation risk associated with a single defect are significantly higher, making traditional sampling inspection methods inadequate for high-reliability delivery requirements. The report suggests that the increasing value of AI hardware is transforming the inspection process from a peripheral support function into a comprehensive quality control measure throughout the entire production workflow. Inspection frequency and coverage are expected to rise continuously, with the trend towards full inspection becoming a key driver for demand growth in high-end inspection equipment.
Rising Product Complexity Spurs High-End Inspection Equipment, Increasing Per-Line Value
As AI computing hardware upgrades towards higher speeds, density, and integration, inspection equipment must also advance to higher-end specifications. The evolution of optical modules from pluggable forms to silicon photonics and CPO demands higher precision, speed, and consistency in inspection for packaging, coupling, mounting, and testing phases. Traditional 2D visual inspection struggles to cover issues like internal structure, solder joint voids, packaging defects, and optical coupling deviations. High-end PCBs and IC substrates require more stringent standards for line width/spacing, layer-to-layer alignment, impedance consistency, hole copper quality, warpage control, and micro-defect identification. Taking back-drilling inspection for AI server PCBs as an example, traditional electrical testing can only detect existing shorts or opens but cannot locate potential risks like stub length. 2D X-ray is limited by overlapping copper layers, making precise quantification of stub boundaries difficult. Destructive cross-section sampling is inefficient and unsuitable for full inspection requirements in mass production of high-value AI server PCBs. Consequently, inspection equipment is evolving from traditional 2D visual inspection and offline sampling towards 3D CT, AI-based intelligent analysis, SPC process statistics, MES system integration, and automatic NG/OK sorting. The report posits that increasing product complexity will not only increase the number of inspection devices but also drive equipment upgrades, thereby raising the value per production line and strengthening customer stickiness.
Ample Room for Domestic Substitution in High-End Inspection, Local Players to Benefit
High-end industrial inspection equipment and core components are still predominantly supplied by overseas manufacturers. In segments like X-Ray, industrial cameras, and precision measurement, foreign brands such as Omron, Keyence, Cognex, Teledyne Dalsa, and Basler maintain strong first-mover advantages. However, the rapid capacity expansion in the AI computing power supply chain, coupled with customer demands for cost reduction, efficiency improvement, and supply chain security, presents opportunities for domestic high-end inspection equipment. For example, Uni-Trend Technology has made continuous breakthroughs in industrial X-ray inspection, upgrading its products from traditional X-Ray to AI-integrated 3D CT, online full inspection, and intelligent analysis, potentially achieving localization breakthroughs in applications like high-end PCBs. Aeco Photonics focuses on core machine vision components like industrial cameras and image capture cards, aiming for domestic breakthroughs in visual inspection scenarios for PCBs, semiconductors, and 3C electronics. The report concludes that the essential inspection needs arising from AI hardware upgrades, combined with the domestic substitution trend, are likely to benefit domestic inspection equipment and key component manufacturers.
Key Investment Risks
The primary risks include intensifying industry competition and the potential for downstream demand to fall short of expectations.