Bernstein's September 2026 report examines the powerful connection between China's AI sector and its electrical grid, exploring opportunities across the country's next-generation industries and market leaders. The report reveals that China's structural advantages in electricity costs, energy infrastructure, and coordinated policies are creating a formidable competitive position in AI development. As AI computing capacity expands at a massive scale, the surge in data center power consumption is set to unlock substantial long-term investment prospects within the grid and energy storage supply chains.

The report highlights that Chinese AI companies possess a significant structural cost advantage. Leading domestic large language models, while achieving competitive intelligence scores, are often priced at just a fraction of their US counterparts. This creates a compelling 'intelligence per dollar' product advantage. The edge stems from more efficient model architectures, lower infrastructure expenses, and cheaper power availability, enabling Chinese service providers to deliver AI inference at substantially lower costs. As AI development pivots from training to large-scale inference, the metric of 'intelligence gained per unit cost' becomes the central battleground, with low-cost computing emerging as a key defense for domestic model developers.

Bernstein projects that over the next decade, electricity consumption by hyperscale internet companies will become one of the fastest-growing sources of power demand. With AI adoption spreading, data center infrastructure expanding continuously, and China's ambitious goal of closing the computing power gap with the US, the firm calculates data center electricity demand will climb from 202 terawatt-hours in 2025 to 1,476 terawatt-hours by 2035, representing a compound annual growth rate of 24%.

Currently, China's AI computing capacity stands at only about 15% of the US level, but domestic AI chips and computing infrastructure are rapidly scaling up. Closing this gap in computing power requires enormous data center construction, which in turn generates massive incremental electricity needs. Cheap, abundant, and scalable power supplies the foundation for China's AI development objectives. The country's total power generation already exceeds double that of the US, with over 500 gigawatts of new capacity added in 2025 alone. Meanwhile, industrial electricity prices in China rank among the lowest globally; in key AI hubs like Inner Mongolia, Ningxia, and Gansu, data center power costs are the cheapest in the world. China also combines low-cost renewable energy with commercially competitive nuclear power, delivering high-value, reliable electricity for AI operations.

Data centers will increasingly rely on renewable energy sources. Through initiatives like 'East Data, West Computing' and green data center policies, along with the emerging 'computing-electricity coordination' framework, policy efforts are driving synchronized development of computing expansion and power infrastructure. New data centers in national computing hubs must source more than 80% of their electricity from green sources, and leading operators have set targets to achieve 100% renewable energy usage by 2030.

Computing power deployment cannot succeed without grid and storage support. From 2026 to 2030, China plans over 5 trillion yuan in grid investment, with major construction of ultra-high-voltage transmission lines to channel western clean energy to computing load centers. In energy storage, China controls 80% of global lithium battery production capacity. Domestic storage installations are expected to grow 95% year-on-year to 300 GWh in 2026, driven by renewable energy absorption demands and AI computing's need for reliable power supply, placing the storage sector on a high-growth trajectory.

The firm identifies grid infrastructure and energy storage as the most promising investment directions. Ultra-high-voltage transmission and large-scale grid investments favor equipment suppliers, while rising renewable energy share combined with growing AI power demand further boosts battery storage requirements. Among the covered companies, Contemporary Amperex Technology Co Ltd (SZSE: 300750), known as CATL, stands to benefit significantly from its leading position in grid-scale energy storage systems, while Sungrow Power Supply Co Ltd (SZSE: 300274) leverages its full-spectrum presence across renewables, power conversion equipment, storage, and grid integration.