Workflow
高端存储芯片(HBM
icon
Search documents
中国半导体,预计增长31.26%
半导体芯闻· 2026-01-27 10:19
Core Insights - The article highlights the rapid deployment of AI application models across various verticals in China, marking the arrival of the edge AI era. The semiconductor market in China is projected to grow significantly, with a forecasted increase of 31.26% in 2026, reaching a market size of $546.5 billion [3]. Semiconductor Market Growth - The semiconductor market in China is expected to grow by 16.17% in 2025 and 13.63% in 2026, with an updated forecast predicting growth of 21.63% in 2025 and 31.26% in 2026 [3]. - The storage market forecast has been significantly revised upwards, with predictions for 2026 storage market growth increased by 62.8%, 53% for 2027, 36% for 2028, and 25.8% for 2029 [5]. AI and Storage Demand - The demand for high-performance storage chips (HBM) is surging due to the global AI infrastructure boom, leading to a supply-demand imbalance that is expected to persist until 2027-2028 [5]. - China's self-sufficiency in high-end storage chips remains low, with approximately 90% of supply dependent on Samsung and SK Hynix, resulting in weak market bargaining power and high storage chip prices [5]. Application Trends - The "Computing and Data Storage" category shows a more pronounced growth trend, with data for 2026 and 2027 revised upwards by over 20% compared to previous forecasts, driven by increased usage and prices of high-end storage chips [8]. - The "Wireless Communication" category also shows significant growth, primarily due to supply-demand imbalances leading to increased average selling prices (ASP) of storage chips, rather than an increase in market demand for wireless terminals [8]. Local AI Chip Market - Due to the ban on NVIDIA's AI chips in China, local AI chip suppliers are expected to gain market share, with edge AI presenting significant growth opportunities for inference AI chips in 2026 [9]. - The penetration of AI in terminal devices is anticipated to increase, necessitating wireless device connectivity to meet low-latency network demands, with hybrid expert model architectures being key for accelerating development [9].