China’s AI Industry Leaps Forward, Backed by Strengthened Computing Power and Data Support

China’s artificial intelligence (AI) industry is experiencing an all-round leap in capability. Leading internet companies have intensively released new-generation main models with trillions of parameters, while emerging AI enterprises are also rapidly iterating their core architectures. From fierce competition around application portals to the strong synergy formed by traffic acquisition and ecological integration, the large model matrix is showing robust innovation vitality, driving AI to transform from a single tool to a high-frequency service portal.

Industry analysis indicates that with the improvement of model capabilities, market demand for underlying computing power and high-quality data is growing exponentially. A recent research report predicts that the compound annual growth rate of China’s large model Token consumption will reach as high as 330% in the next five years. As a key indicator reflecting the usage intensity and user stickiness of large models, the soaring Token calls mean AI is accelerating its penetration into practical application scenarios across various industries.

Meanwhile, the massive call demand has put unprecedented pressure on the carrying capacity of computing power and data infrastructure. Relevant analysis shows that the continuous evolution of multi-modal large models and the rapid iteration of large language model architectures are significantly pushing up basic computing power consumption. As AI transitions from a single tool to a high-frequency service portal, the demand for inference-side computing power is rising at a faster-than-expected rate.

Faced with the blowout demand for computing power, systematic layout at the national level is accelerating. In recent years, several policy documents have been issued successively, including the "Implementation Opinions on Further Implementing the ‘East Data West Computing’ Project and Accelerating the Construction of a National Integrated Computing Network" and the "Three-Year Action Plan for ‘Data Elements ×’ (2024—2026)". These top-level designs have clarified clear directions for the underlying support of the AI industry, from coordinating the construction of national computing power infrastructure to promoting the high-level application of data elements.

To respond to national strategic arrangements, various regions have intensively introduced supporting measures to fully support the construction of the AI industry base. Recently, Sichuan Province issued an implementation opinion proposing to consolidate AI infrastructure, seize the commanding height of data information services, promote the construction of AI datasets and corpora, expand provincial "computing power vouchers", and explore the formulation of "scenario vouchers" and "model vouchers" policies.

Across the country, regions are competing to make efforts: Beijing has intensively issued implementation plans for computing power infrastructure construction, increasing computing power supply and providing computing power subsidies to innovative enterprises; Shanghai, Shenzhen and other places are accelerating the construction of urban-level intelligent computing power overall scheduling platforms to improve the efficiency of computing power resource utilization; central and western provinces such as Guizhou and Ningxia are leveraging their climate and energy advantages to build green intelligent computing bases. Guizhou, for example, has upgraded its computing power voucher policy to version 3.0, expanding incentives to cover model services and corpus procurement with a maximum subsidy ratio of 30% of the valid contract amount.

In addition to the overall construction at the hardware level, the cultivation of high-quality datasets is also a key focus of local policy layout. Establishing special funds to support enterprises in building trusted data spaces, national-level data annotation bases and public service platforms has become a common consensus among many regions to seize the commanding height of data information services.

The accelerated construction of computing power infrastructure has also provided a strategic opportunity for the leapfrog development of the domestic computing power industry chain. After multiple rounds of technological polishing and application feedback, the performance and ecological construction of domestic computing power chips have crossed the key inflection point from "usable" to "easy to use". Relying on comprehensive advantages such as localized service response, cost control and adaptation to national strategic orientation, domestic computing power equipment is accelerating its integration into the core industrial chain.

With the strong support of policies, infrastructure and industrial innovation, China’s AI industry is expected to maintain its upward momentum, continuously enhance its core competitiveness, and inject new vitality into high-quality economic development.