Key Takeaways
China's artificial intelligence ambitions are rapidly translating into industrial might, with the core sector's value now exceeding 1.2 trillion yuan as it enters a new phase of application-driven growth. A landmark report released on April 29 reveals that for the first time, the volume of data used for AI inference has surpassed data used for training, a key indicator of the industry's maturation from development to widespread commercial deployment.
The "National Data Resources Survey Report (2025)," presented at the Digital China Construction Summit, frames the shift as a pivotal moment. It marks the industry's entry into "a new stage of training and inference parity, driven by applications," suggesting a strategic focus on turning foundational models into practical tools across the economy.
The scale of this expansion is staggering. According to the report, China's daily AI token calls grew from over one trillion at the start of 2025 to 100 trillion by year-end, with a cumulative total of approximately 21,100 trillion for the year. This surge was supported by a national intelligent computing capacity of 1.59 million PFlops and over 110,000 high-quality datasets, providing the foundational resources for this boom.
For investors, the report quantifies a seismic shift that puts domestic technology champions in the spotlight while posing a long-term challenge to international AI leaders. The findings underscore a powerful, state-supported ecosystem designed to integrate AI into everything from healthcare to manufacturing, backed by new government venture funds aimed at securing leadership in strategic technologies.
The transition to an inference-led market is the report's most significant revelation. While the training phase involves building and teaching large models—a capital-intensive process dominated by a few major players—the inference phase is about deploying those models to perform tasks in the real world. The fact that inference data volume reached 101.34 exabytes, overtaking training data, shows that China's AI industry is now overwhelmingly focused on user-facing applications and services. This creates a virtuous cycle where practical use generates new data and revenue, driving further iterative development.
This application-centric ecosystem is being built on a robust national infrastructure. The report notes that over 80% of new computing power is being deployed in China's eight major hub regions, part of a national integrated network. This state-directed build-out is complemented by significant investment in the digital economy, with fixed-asset investment in information services growing 28.4% year-over-year in 2025.
The government is also steering capital toward the sector. A newly launched national venture capital fund is explicitly targeting key areas like AI and integrated circuits. This policy support, combined with the immense scale of data generation, creates a formidable competitive advantage. The report highlights AI's accelerating fusion with the broader economy, with the digital economy's core industries now accounting for over 10.5% of GDP.
This national strategy for AI mirrors China's approach to other critical sectors. Just as Beijing has moved to secure its global dominance in strategic minerals like rare earths and tungsten, it is now clearly aiming for self-sufficiency and leadership in the foundational resources of the 21st-century economy: data and compute power. While the report focuses on domestic progress, the implications are global, signaling the rise of a powerful, integrated AI ecosystem where giants like Alibaba, Tencent, and Baidu will compete directly with Silicon Valley.
