Chinese Investors Ditch Overheated AI Chips for Power & Metals—The Next Big Play?

Smart money flees AI chip mania as valuations hit boiling point

Power grids and industrial metals emerge as the new alpha hunt

Shanghai traders pivot faster than a meme coin rug pull—liquidity chases the next narrative

Bonus jab: Just wait until Wall Street 'discovers' this trend in 6 months and sells it back to Asia as an 'innovative thematic ETF'

TLDRs;

• Investors are moving from expensive AI chip stocks into China’s power, metals, and infrastructure sectors tied to AI growth.
• Energy stocks outperformed as analysts predict one-third of China’s AI spending will shift to facilities and power.
• Rising electricity demand and major data-center investments are pushing utilities and storage companies into the spotlight.
• The Eastern Data, Western Computing initiative drives infrastructure growth but lacks clear long-term scheduling transparency.

Chinese investors are rebalancing away from high-flying AI semiconductor stocks and piling into companies powering the country’s enormous AI computing boom.

The MOVE comes amid growing worries that valuations for chipmakers have surged too quickly, making energy, utilities, and metals appear more attractive for long-term exposure.

This investor rotation is becoming increasingly visible across China’s major indices. The country’s energy stock index climbed 10% in October, positioning it to outperform the CSI 300 for a second consecutive month. Companies tied to electricity generation, grid expansion, and materials needed for data-center-grade infrastructure are now receiving heightened interest from both institutional funds and retail traders.

AI Infrastructure Becomes the New Growth Engine

Bank of America analysts note that China is uniquely positioned to scale AI infrastructure due to its abundant power generation capacity and comparatively lower electricity costs. With AI models requiring massive compute and energy inputs, investors are now pricing in a decade-long buildout of national digital infrastructure.

Analysts estimate that by 2030, nearly one-third of all AI spending in China will go toward facilities, metals, and power systems, marking a significant pivot away from chip-centric spending. This broader ecosystem includes data centers, cooling systems, grid reinforcement, renewable power deployments, and the materials needed to build them.

UBS has also raised its forecast for China’s electricity demand, projecting 8% annual growth between 2028 and 2030. That demand will be driven not only by consumer and industrial usage but also by the accelerating rollout of AI data centers across the country.

Key Winners Emerge Amid Infrastructure Surge

Several companies have already benefited from this momentum. Shares of CSI Solar, TBEA, and Aluminum Corp. of China have risen sharply in recent months, reflecting confidence in sectors that supply photovoltaic components, grid equipment, and raw metals essential to high-density computing facilities.

Meanwhile, demand is increasing for energy storage suppliers and optical fiber producers, two industries directly linked to the performance and reliability of AI infrastructure.

As China moves to modernize and distribute computing capacity across its eastern and western regions, these supporting technologies are expected to see sustained procurement cycles.

Eastern Data, Western Computing Shapes Investment Path

Much of China’s AI infrastructure push hinges on the multi-year national program known as Eastern Data, Western Computing, which aims to distribute data-center workloads more efficiently across the country.

By mid-2024, China had invested over $6.1 billion into eight national computing hubs, with total investment surpassing $28 billion. These hubs collectively housed 1.95 million server racks, operating at about 63% utilization, suggesting significant room for scaling.

Guizhou province, one of China’s key data-center zones, reported 92.6 exaflops of computing power as of late 2025, with 97% devoted to intelligent computing tasks. Latency between the country’s east and west hubs is meeting the 20-millisecond target, positioning the network for high-performance AI workloads. Newer data centers are also achieving PUE ratios as low as 1.04, reflecting cutting-edge efficiency.

However, the rollout lacks a transparent timeline. Project lists for 2024–2025, grid expansion details, and long-term capacity plans remain opaque, introducing measured uncertainty even as infrastructure spending accelerates.