Bernstein: Computing power is running wild, and China's AI cost advantage detonates long-term opportunities for grid energy storage

The Zhitong Finance App learned that the research report released by Bernstein in September 2026 focuses on the linkage between China's AI industry and power systems. The series aims to explore the next generation of key industries and leading enterprises in the industry in China. The report points out that China relies on power costs, energy infrastructure, and policy collaboration to form a structural competitive advantage for AI development. The large-scale expansion of AI computing power will explode data center electricity demand, and the power grid and energy storage industry chain will usher in long-term investment opportunities.

Bernstein mentioned that artificial intelligence in China has structural cost advantages. China's leading model. Under the premise that smart ratings are competitive, pricing is often only a fraction of the US similar model, forming a very attractive “get smart per dollar” product advantage.

This advantage comes from a more efficient model architecture, lower infrastructure costs, and cheap electricity, which enables domestic service providers to provide AI inference services at a significantly lower cost. As the focus of AI development shifts from training to large-scale reasoning, “intelligence obtained per unit cost” will become a core competitive dimension, and low-cost computing power will become an important moat for domestic models to break through.

The bank expects that in the next ten years, the electricity consumption of hyperscale Internet companies will become one of the fastest growing sources of electricity demand. With the spread of artificial intelligence and the continued expansion of data center infrastructure, the bank estimates that data center electricity demand will grow from 202 terawatt-hours in 2025 to 1,476 terawatt-hours in 2035, with a compound annual growth rate of 24%. Currently, China's AI computing power is only about 15% of that of the US, but domestic AI chips and computing infrastructure are rapidly expanding production. If we want to match the US in terms of computing power, we need to expand data centers on a large scale, which will bring huge incremental demand for electricity.

Cheap, sufficient, and scalable electricity provides support for China's artificial intelligence development goals. China's current total power generation capacity is more than double that of the US; in 2025 alone, it will add more than 500 gigawatts of installed capacity. At the same time, China's industrial electricity prices are at the lowest level in the world; AI core hubs such as Inner Mongolia, Ningxia, and Gansu have the lowest electricity costs in the world. China also has both low-cost renewable energy and cost-competitive nuclear power to provide cost-effective and highly reliable power guarantees for AI businesses.

Data centers will increasingly be powered by renewable energy. Relying on “East Digital and Western Computing”, green data center related initiatives, and the emerging “Computing and Telecommunication Collaboration” framework, the policy level is promoting the expansion of computing power and the collaborative development of power infrastructure. For national-level computing power hubs to build new data centers, green electricity must account for more than 80%; leading operators have also set goals to achieve 100% use of renewable energy by 2030.

The implementation of computing power is inseparable from the power grid and energy storage facilities. In 2026-2030, the domestic power grid plans to invest more than 5 trillion yuan to vigorously build UHV lines and open up a channel for transporting clean energy from the western region to the computing power load center. In terms of energy storage, China accounts for 80% of the world's lithium battery production capacity. Domestic energy storage capacity is expected to increase by 95% to 300 GWh year-on-year in 2026. New energy consumption is compounded by AI computing power supply reliability requirements, and the energy storage industry is experiencing high growth.

The bank believes that the most valuable investment direction is to focus on power grid infrastructure and energy storage. Large-scale investment in UHV transmission and power grids favors equipment suppliers; the increase in the share of renewable energy is compounded by rising demand for AI electricity, further boosting demand for battery energy storage. Among the targets covered by the report, the Ningde Era fully benefited from its leading position in grid-level energy storage systems; Sunshine Power lays out a full circuit integrating renewable energy, power conversion equipment, energy storage and power grids.