According to Omdia's semiconductor industry observations, AI application models are being widely deployed across various vertical industries in China, signaling the arrival of the edge AI era. Numerous large language models (LLMs) are actively deploying vertical application models across different sectors. Digital terminals with edge inference capabilities are set for rapid growth, becoming a major driver for the expansion of China's semiconductor industry—particularly in the mature process technology segment. The latest data from the fourth quarter of 2025 indicates that China's semiconductor market is projected to grow by 31.26% in 2026, reaching a market size of $546.5 billion. The "Semiconductor Application Market Forecast Tool (AMFT) - China Region (2Q25)" had previously forecasted growth of 16.17% for the Chinese semiconductor market in 2025 and 13.63% in 2026. However, the updated "Semiconductor Application Market Forecast Tool (AMFT) - China Region (4Q25)" report, reflecting the Q4 2025 update, has revised these projections upward to 21.63% growth for 2025 and a significant 31.26% growth for 2026.

The forecast for the memory market size was substantially revised upward in the Q4 2025 projections. Compared to the Q2 version, the forecast for the 2026 memory market increased by 62.8% in the new version, while projections for 2027 rose by 53%, for 2028 by 36%, and for 2029 by 25.8%. Driven by the global AI infrastructure boom beginning in Q4 2025, AI-related memory consumption from both China and the US is experiencing exponential growth. Large-scale data center deployments have led to a surge in demand for high-performance memory chips like HBM. With supply constrained and demand increasing against a backdrop of supply-demand imbalance, memory prices are undergoing significant volatility. According to Omdia's forecast, this tight supply situation is expected to persist until 2027-2028. As the momentum of AI development becomes increasingly clear, the imbalance in memory chip supply and demand is worsening, with global shortages continuing to spread. China's self-sufficiency rate for high-end memory chips (HBM, high-end DRAM, NAND) remains low, with approximately 90% of supply reliant on Samsung and SK Hynix, while Micron continues to face political restrictions. Consequently, China's market bargaining power is relatively weak, keeping the unit prices of memory chips high.

The "Computing" and "Wireless Communications" categories are significantly impacted by memory products, leading to substantial growth in their market sizes. The "Semiconductor Application Market Forecast Tool (AMFT) - China Region (4Q25)" revised the growth trends for mainstream applications. Compared to the "Computing and Data Storage" category in the Q2 2025 version of the AMFT, the Q4 2025 version shows a more pronounced growth trend. Data for both 2026 and 2027 were revised upwards by more than 20% compared to the previous version. The primary drivers are the increased usage and unit prices of high-end memory chips propelled by data center demand, alongside the adoption of AI-related logic chips at 5nm and more advanced process nodes. The "Wireless Communications" category is another application area showing significant growth. However, this revenue growth is primarily due to the supply-demand imbalance causing a significant increase in the Average Selling Price (ASP) of memory chips (like LPDDR, 3D NAND) used in electronic devices, rather than an increase in market demand or shipment volumes for wireless terminals (such as smartphones, tablets). Market feedback from Q4 2025 indicates that continuously rising memory ASPs and suppliers' practices of signing non-long-term agreements have prompted several OEMs to lower their shipment volume expectations for 2026.

AI is set to bring new semiconductor applications to China. Due to the ban on NVIDIA's AI chip sales to China, the market share of local Chinese AI chip suppliers is expected to further expand in 2026. Edge AI will present significant growth opportunities for China's inference AI chips in 2026. The penetration rate of AI in terminal devices is projected to continue increasing throughout 2026, making wireless device connectivity increasingly important to meet low-latency network requirements. For devices with end-to-end AI processing capabilities, adopting a Mixture of Experts (MoE) architecture is key to accelerating the development process, particularly in compressing and miniaturizing large models. Omdia states that it has seen strong evidence of deep synergy between computing power and the AI ecosystem, demonstrating China's diversified breakthroughs in the field of AI infrastructure. In 2026, with the continued synergy between technology and the ecosystem, China's intelligent computing capabilities will play an increasingly important role.