On March 12, Morgan Stanley released a report on China’s artificial intelligence (AI) GPUs, noting that China has made substantial progress in breaking through bottlenecks in equipment and wafer foundry capabilities over the past 12 months. With policy support, China’s wafer foundry capacity and chip supply are expected to meet core domestic demand around 2028.
Morgan Stanley believes that China’s localization strategy—expanding scale in chips, fabs, and equipment to offset disadvantages in advanced process technology—has been continuously effective. In an optimistic scenario, China’s domestic AI GPUs will expand into training workloads and may be adopted by overseas customers. In a downside scenario, fading differentiation will lead to commoditization and industry consolidation. The firm forecasts that China’s AI chip market size will reach $67 billion by 2030, representing a compound annual growth rate (CAGR) of 23% from 2024 to 2030, and China’s AI chip self-sufficiency rate will hit 76% by 2030.
The report pointed out that while policy support can accelerate early-stage development, long-term value depends on commercial competitiveness. Chinese AI GPU suppliers must demonstrate convincing economic efficiency to sustain growth beyond 2028.
According to Morgan Stanley’s analysis, China’s AI data centers enjoy competitive total cost of ownership (TCO) thanks to lower prices, cheaper power costs, and increasingly mature infrastructure. In inference workloads, cost per token is more critical than peak performance, further strengthening the competitive edge of Chinese solutions.
(Reprinted from https://news.eccn.com/）