Textbook in PDF format In this comprehensive reference work for researchers, engineers, and students, Kawahara provides a one-stop exploration of next-generation computing at the LSI circuit level, with a focus on the integration of AI, advanced LSI design, Ising machines, and memory innovations. While current GPUs have high parallel processing capabilities suitable for computations on large datasets, their power consumption is approaching its limit and requires further development. Additionally, edge computing is becoming increasingly important alongside cloud computing. Amid these significant technological trends, this book provides readers with important insights into next-generation computing, namely (1) neural network (artificial intelligence) LSIs and their low power and high performance, (2) hardware design technology for combinatorial optimization problems and Ising machines, and (3) semiconductor memory and data-centric computing. Kawahara first describes the basics of LSI design and neural networks before then describing their large-scale integration, power efficiency and performance enhancements. He then also explains hardware design techniques for Ising machines, offers case studies of fully coupled Ising machine LSI. Last, he discusses the basics of semiconductor memory, near/in-memory AI computing, and then examines the future prospects. Readers will be able to apply this knowledge to the design and manufacture of such devices to overcome the limitations of current hardware and computational methods, driving future advancements in artificial intelligence and optimization. The content and structure described in this book are as follows. We begin with the fundamentals. Chapter 2 describes the components of AI LSI. Next, in Chapter 3, the basic electronic circuit blocks and their realization as digital or analog circuits are explained. In Chapter 4, based on the elements in Chapters 2 and 3, the basic configuration of NNs and their realization will be explained. It will show the basic LSI circuitry of NN expressivity and inference functions, the configuration of NN learning function LSIization, and the structure of NNs. Chapter 5 will focus on key technologies, including cloud computing, while focusing on edge AI. Specifically, integrated circuits, low-precision NNs, sparse NNs, hardware optimization, FPGAs, spiking NNs and neuro-morphic computing, embedding learning functions, and quantum computing in edge AI. The above provides a general overview of AI chips, explaining the basic structure of AI LSIs, electronic circuit elements, NN structures, and important technologies for edge AI and cloud computing, so that readers can gain a basic understanding of AI chips. In the following Chapters 6 and 7, we focus on an Ising machine that solves combinatorial optimization problems as a concrete example of a circuit configuration similar to that of an AI chip. The processing performed by this Ising machine is the same as that of an AI chip for inference processing. Chapter 6 begins with an overview of Ising models and describes the various Ising machine developments underway. Specific topics include an explanation of the Ising model, the relationship between ferromagnetic materials and the Ising model, applications of the Ising model to combinatorial optimization problems, quantum Ising machines, Ising machines using CMOS integrated circuits, and components of Ising LSIs. In Chapter 7, the fully coupled 13Ising machine LSI developed in the author’s laboratory is presented as a case study. The specific contents of the chapter include the configuration of the fully coupled Ising LSI, the configuration of the large-scale Ising machine LSI, and the scalable configuration. This is followed by a discussion of memory technologies and their application methods that are essential for improving the performance of these AI chips. Chapter 8 begins the description with the hierarchical structure of memory in computing, followed by an overview of major memory technologies in Chapter 9. Specifically, the hierarchical structure of memory in signal processing and major semiconductor memories (SRAM, DRAM, flash memory, MRAM, FeRAM, PCRAM, and RRAM) is described. Chapter 10 addresses near-memory computing and in-memory computing, which are the foundation for utilizing memory technology in AI chips at the edge. The final chapter, Chapter 11, will show the evolution and future development of these techniques. Future prospects, including advances in brain-based information processing and integration with other technologies, will also be discussed. This is a valuable reference for students, engineers and researchers alike in this field. As it begins with the basics, it enables all readers to follow the direction of next-generation computing and its important technical content without the need for prior knowledge or reference to other books. Preface Introduction: AI Computing Overview of Artificial Intelligence Hardware LSI (AI Chips) and Its Components Basics of Large‑Scale Integrated Circuits for AI Basic Structure of AI Chips and Various Neural Networks. Low‑Power, High‑Performance AI Chips Structure and Related Computing Ising Machines and Combinatorial Optimization Problem Fully Coupled Ising Machines: A Case Study Semiconductor Memory and Computing Main Semiconductor Memory Features AI Computing with Semiconductor Memory (In‑Memory Computing) Future Prospects