Industry Deep Report: GPU Research Framework

The processor chip industry has undergone two major shifts: from specialized to general-purpose and back to specialized, driven first by the von Neumann architecture and the 1971 X86 CPU, and later by the slowdown of Moore's Law and the emergence of heterogeneous computing exemplified by GPUs.

In the heterogeneous era, chips must integrate multiple modules—such as GPUs, CPUs, NPUs, and others—to meet diverse application needs; automotive chips, for example, integrate at least ten different processing units.

GPU Investment Logic Framework

1.1 GPU: a critical demand in the era of specialized computing.

1.2 GPU investment map: rapid development under oligopolistic dominance.

1.3 GPU industry chain depth: three major routes forming a spindle-shaped structure.

1.4 GPU industry chain: advanced process digital chip ecosystem.

Detailed GPU Overview

2.1 GPU composition: microarchitecture and APIs.

2.2 "XPU" competition: GPU versus CPU, FPGA, ASIC.

2.3 Historical perspective: analysis of GPU microarchitecture, process technology, and API trends.

2.4 GPU supply‑demand analysis: five driving forces and two production models.

Global GPU Landscape and Industry Leaders

3.1 Market size and structure: high growth under oligopoly.

3.2 The three GPU giants: NVIDIA, AMD, Intel.

3.3 GPU IP leaders: ARM, Imagination.

3.4 Mobile GPUs: Qualcomm Adreno, Apple A‑series.

Domestic GPU Development Path

4.1 Chinese GPU reliance on imports.

4.2 Leading domestic GPU companies: Jingjia Micro, Xin Yuan.

4.3 Other domestic GPU manufacturers.

The global GPU industry is dominated by overseas giants across design, materials, EDA/IP, and manufacturing, with domestic capabilities lagging, especially in advanced processes and packaging. The first high‑performance, low‑power Chinese GPU (JM5400) highlights the strong dependency of GPUs on CPUs and the high technical difficulty of GPU development, which together slow domestic progress.