China’s daily token consumption has surged more than 300-fold in the past 18 months, reflecting the rapid growth of the country’s AI applications, a senior official said Thursday. At a State Council press conference, Liu Liehong, head of the National Data Bureau, said daily token usage rose from 100 billion at the start of 2024 to more than 30 trillion by the end of June. Liu also highlighted progress in building high-quality datasets, noting that China had created more than 35,000 datasets totaling over 400 petabytes as of June. AI model training has spurred demand for data trading, with cumulative transactions reaching nearly 4 billion yuan ($552 million). The Beijing Data Exchange now accounts for nearly 80% of such trades, up from 10% last year. Chinese-language data has played a vital role in model development, Liu said, with most domestic AI models now trained on datasets that are more than 60% Chinese—and in some cases, up to 80%. [Xinhuanet, in Chinese]
Related
