In the AI model race, computing power gets the headlines, but data - the other critical resource - is quietly becoming the decisive factor. According to The New York Times, Beijing plans to turn China into a data powerhouse by 2028, aiming to close the gap with the US in acquiring high-quality AI training data and gain greater influence in AI chatbot development.
China's data strategy has several distinctive features. First, scale: China has the world's largest internet user base, generating massive daily data volumes. But quantity doesn't equal quality - high-quality labeled datasets in Chinese remain scarce, and this is the core problem the strategy aims to solve.
Second, systematic construction. Unlike the US market-driven approach, China's data strategy is government-led, from data labeling industry support to corpus construction and data trading markets. Universities and research institutions are participating in building large-scale Chinese datasets across medical, legal, and financial domains.
Third, data sovereignty awareness. China is increasingly focused on controlling its data resources independently, reducing reliance on overseas data sources. This "data localization" trend is reshaping the global AI data supply chain.
China has one advantage the US lacks: public data accessibility. Government agencies control vast amounts of high-quality data - medical records, traffic flows, weather data, education data - which can be used for AI training under privacy protections. The US faces stricter legal restrictions and privacy controversies on public data access.
For the AI industry, competition for data resources may outlast the computing power race. Chips can be purchased, but high-quality training data takes time and institutional building to accumulate. China has the scale advantage but lags in data quality and labeling precision. The outcome of this "data war" will directly shape the global AI model landscape over the next five years.