With the demand for AI computing power exponentially increasing, the cost of a single AI server rack has repeatedly reached astronomical levels. Due to the extremely high power and highly precise system design, "quality control" has transformed from a silent supporting role in the manufacturing process to an absolute key player that directly affects the performance, yield, and final delivery capability of AI servers.
At this year's COMPUTEX 2026 forum, ZEISS brought "quality issues" to the big stage for the first time. It not only showcased its cutting-edge testing technologies across multiple business units, but also proposed a "Chip-to-Rack" full value chain quality solution, which comprehensively covers everything from the bottom layer of AI chips, HBM high-bandwidth memory, ABF carrier boards, and PCBs, all the way to the outermost liquid cooling heat dissipation modules and co-packaged optical (CPO) components.
Breaking the limitations of traditional financial planning: Quality control is shifting from "post-event analysis" to "online prevention".
In the past, quality inspection in the electronics manufacturing industry relied heavily on post-production failure analysis (FA). However, in the era of AI servers, the cost of components is extremely high, and if defects are only discovered after assembly, the cost of rework is simply unbearable for companies.
The core breakthrough that Zeiss showcased this time lies in significantly advancing the quality inspection defense line, extending it to inline and at-line inspections at the manufacturing process end.
Non-destructive testing cutting-edge technology:By integrating advanced technologies such as non-destructive 3D X-ray microscopy with computed tomography (CT), electron microscopy, structured light, and three-dimensional measurement (CMM), Zeiss can perform high-resolution internal defect analysis without damaging expensive components.
The specific benefits are astonishing:According to data released by Zeiss, in certain high-value bottleneck processes (such as highly complex liquid-cooled modules, high-speed interconnects, and system assembly), the implementation of these advanced non-destructive testing technologies can significantly improve yield by up to 20% to 30%. Furthermore, the implementation of automated real-time non-destructive testing on PCBs can reduce quality-related costs by up to 80%.
Deepening Local R&D: A Dual-Axis Layout of "Taiwan to Global"
Richard Gärtner, a member of the Executive Committee on Industrial Quality and Research at Zeiss headquarters, emphasized that Zeiss can provide a "one-stop" quality solution covering lithography, mask repair, advanced packaging, and system assembly. The practical application of this vast technological capability is based in Taiwan, which plays a central role in the global semiconductor supply chain.
Zeiss Taiwan General Manager Tsai Hui pointed out that in order to deepen the "Taiwan to Global" strategy, Zeiss not only established the Hsinchu Science Park Innovation Center, but also officially opened the Taichung Zeiss Quality Excellence Center this year. This center is entirely led by a local Taiwanese engineering team, directly connecting Taiwan's semiconductor, high-tech manufacturing, and research ecosystem, providing ODM giants with localized support that meets their core needs of "rapid iteration" and "cost control".
Facing the challenges of "fiber-to-the-home" and high-voltage direct current
Faced with two major trends that will inevitably emerge in next-generation AI server architectures—high-voltage direct current (HVDC) and "optics in, copper out" (co-packaged optics)—traditional testing methods are already facing physical limits.
In response, Yen Tzu-teng, Global Head of Electronics Business at ZEISS Industrial Measurement Solutions, stated that ZEISS's next-generation testing technology is ready to provide corresponding quality solutions for power modules (such as SST and Busway) and extremely precise optical communication components (such as high-speed optical transceivers and fiber optic array units FAU), helping the supply chain to balance performance, reliability, and mass production efficiency during the painful period of adopting new technologies.
