Semiconductor Equipment: AI Training and Inference Demand Soars, Computing Power and Memory Chip Volume Drive Semiconductor Equipment Demand [Soochow Machinery]

English summary

### Semiconductor Equipment: AI Training and Inference Demand Soars, Computing Power and Memory Chip Volume Drive Semiconductor Equipment Demand [Soochow Machinery] Time: 2026-08-01 00:35:47 Content: Semiconductor equipment: AI training and inference demand grows rapidly, and the ramp-up of computing power and memory chips drives demand for semiconductor equipment [Soochow Machinery]. The performance of global large models continues to improve, driving continuous increases in training computing investment. As of the end of 2025, there were 87 publicly disclosed large models with over 100 billion parameters worldwide. In 2025, 124 new open-source large models were released, of which more than 20 had over 70 billion parameters. Competition among many models at home and abroad, such as Claude, Gemini, ChatGPT, Deepseek, Zhipu GLM, and Kimi, is intensifying. In 2026, the combined AI capital expenditure of the four major tech giants is expected to be raised again to $725 billion, up about 77% year over year, far exceeding $410 billion in 2025, with unprecedented scale.