China's Data Industry Becomes AI Infrastructure, Hits 6.78T Yuan

China's data industry reached 6.78 trillion yuan (approximately $1 trillion) in 2025, growing 15.7 percent year on year, according to figures released at the 2026 China International Big Data Industry Expo in Guiyang, Guizhou province. The numbers confirm what has been increasingly evident: China's data sector is no longer a standalone industry but is fast becoming the infrastructure layer powering artificial intelligence.

The figures were disclosed at the expo's opening on August 28 and elaborated in the China Data Industry Development Report (2026), released by the National Data Development Research Institute at the event's main forum.

AI now drives nearly three-quarters of the industry

Of the 6.78 trillion yuan total, AI-related data products and services — including AI application software and token services — generated 4.99 trillion yuan in output value, accounting for 73.6 percent of the entire data industry. That single statistic captures the scale of the shift: data is no longer primarily about storage, analytics, or governance. It is about feeding AI.

"AI development is heavily driven by data," said Liu Liehong, head of China's National Data Bureau, at the expo. "Wherever AI advances, we push forward the construction and application of high-quality datasets."

The industry's trajectory over three years tells a consistent story. Scale rose from 5.06 trillion yuan in 2023 to 5.86 trillion in 2024, then to 6.78 trillion in 2025 — accelerating rather than plateauing. The number of data enterprises reached 482,000, and granted patents related to AI agents exceeded 3,400.

126,000 high-quality datasets built nationwide

The most striking growth is in high-quality datasets — collections of processed data ready for training and fine-tuning AI models. As of August 2026, China had built over 126,000 such datasets with a combined volume exceeding 1,815 petabytes. That represents an increase of more than 89 percent from the end of the first quarter alone.

The National Dataset Management Service System, launched in April 2026, has already published more than 1,700 datasets, steadily expanding the supply of data available to developers and enterprises.

This buildout is not abstract. Guizhou, host of the expo and a key hub in China's "East Data, West Computing" project, housed 50 key data centers by late July 2026 with a total computing capacity of 176.57 EFLOPS. Nearly 98.4 percent of that capacity was dedicated to AI-related computing. The national strategy channels heavy data processing from China's economically dense eastern coast to western regions rich in renewable energy — making the AI infrastructure buildout simultaneously a green-energy milestone.

A marketplace, not just a policy push

The expo itself reflected the maturing of the data-AI ecosystem. Themed "Token: A New Path to Value of Data Elements," it attracted 372 domestic and international companies and over 16,000 registered guests. For the first time, the event hosted a "data marketplace" where 95 data enterprises presented 253 data products, enabling direct matchmaking between data suppliers and consumers.

The report identified four forward-looking trends: token services will accelerate, creating new business models; AI's demand for data will continue to grow; AI-native data technologies and synthetic data will become increasingly important sources; and the data industry will maintain its rapid growth trajectory.

From policy first-mover to infrastructure provider

China's path to this point has been shaped by deliberate policy. In 2019, it became the world's first country to formally recognize data as a factor of production. The National Data Bureau was established in 2023 to coordinate data governance and development. The cumulative result is an industry that, by 2025, is not merely large but structurally oriented toward AI — with the majority of its output value tied directly to AI products and services.

The significance extends beyond headline numbers. When 73.6 percent of a 6.78-trillion-yuan industry is AI-related, and when 98.4 percent of new computing capacity in a national hub is allocated to AI, the data industry has effectively become AI infrastructure. The datasets, the token services, the computing capacity, and the marketplace mechanisms are all being built and scaled with AI as the primary consumer.

For enterprises building AI applications in China, the implication is concrete: the bottleneck is shifting from access to data to the quality and specificity of data. The 89-percent quarterly growth in high-quality datasets suggests that the supply side is responding aggressively. Whether that supply can keep pace with the demands of next-generation models remains the open question — but the infrastructure to try is now in place.