Abstract: The selection of future industries is determined by technology push and demand pull, and the emergence of computing-in-memory (CIM) technology perfectly exemplifies this proposition. In recent years, the sophisticated concept of “large-scale models” has gradually entered households, with people growing dependent on the efficient and comprehensive thinking capabilities of machines. However, artificial intelligence large-scale models rely heavily on the high computing power of chips, which imposes requirements for chips to better support parallel processing and high-efficiency data flow. In recent years, numerous projects and policies have been launched both domestically and internationally to support the research of this technology. The academic and industrial communities have revisited the technical concept proposed half a century ago; they have conducted extensive research across multiple dimensions, including architecture, process technology, and integration, to explore next-generation chip technologies in the post-Moore era. The market scale of CIM continues to expand, seemingly approaching the eve of an explosive growth phase. Currently, technology implementation remains the core challenge for the integration of computing and memory. Drawing on the authors’ experience in both academia and industry, this article discusses the key scientific issue in CIM: process integration technology. It is hoped that this work will benefit peers at all career stages and inspire more in-depth discussions on the topic.