China’s AI landscape has experienced explosive growth, with daily token consumption skyrocketing over 300-fold in just 18 months. Liu Liehong, head of the National Data Bureau, reported that usage surged from 100 billion tokens at the start of 2024 to over 30 trillion by June. This rapid increase reflects advancements in AI applications and significant progress in constructing high-quality datasets, totaling over 400 petabytes across more than 35,000 datasets. Demand for data trading has also surged, with cumulative transactions reaching nearly 4 billion yuan ($552 million), and the Beijing Data Exchange now handling nearly 80% of these trades, up from just 10% last year. Liu emphasized the importance of Chinese-language data for model training, revealing that most domestic AI models utilize datasets composed of over 60% Chinese data, with some reaching 80%. This trend underscores the pivotal role of localized data in enhancing AI development in China.
Source link
