China’s AI inference data reportedly surpasses training data
A report from the China Electronics Information Industry Development Institute, as cited by IT之家, 36氪 and AIbase, says China’s AI inference data volume has exceeded training data for the first time. The report says daily token calls exceeded 175 trillion as of June 2026, more than 1,000 times the early-2024 level, while leading-model iteration cycles shortened from months to weeks or days. IT之家 also reports that more than 500 excellence-level smart factories have been built and domestic computing power and large models have reached over 1,000 application scenarios across 20 industries.
The report presents inference demand, model deployment and manufacturing adoption as evidence that China’s AI activity is expanding beyond model training toward scaled industrial use.
报告:我国人工智能推理数据量首超训练数据量-36氪
登录
搜索
#China Electronics Information Industry Development Institute