Forget chips, real bottleneck in AI is power

The Gansu-to-Shandong ±800 kV ultra-high voltage direct current transmission line originates from the Qingyang converter station in Gansu province. [Photo provided to chinadaily.com.cn]

As artificial intelligence continues to expand, the real contest may no longer be about who builds the smartest models, but who can power them sustainably.

For years, the global conversation around artificial intelligence has centered on chips, algorithms and computing power. But the AI boom is exposing a more fundamental reality: AI ultimately runs on electricity, not just on code.

Without abundant, reliable and affordable power, computing capacity is merely a technical promise.

The explosive growth of AI and data centers is revealing the first hard physical constraint of the digital age — not capital, not software, but the capacity of energy systems.

This shift is quietly redefining national competitiveness. The International Energy Agency has warned that global electricity demand is entering a new phase of rapid expansion, increasingly driven by AI, cloud computing and data centers rather than traditional industrial growth.

Similarly, the World Economic Forum notes that access to power is becoming a decisive factor in the location of digital infrastructure.

In other words, the AI race is turning from a competition over algorithms and semiconductors into a contest over electricity systems.

This transition is producing sharply different outcomes across governance models. In parts of the United States, Big Tech's AI expansion is colliding with local power systems.

Massive data centers are coming up near communities that were never designed to absorb such concentrated electricity demand, leading to increased costs of upgrading grids and supporting infrastructure.

These costs often spill over to households and small businesses through electricity pricing mechanisms, resulting in higher energy bills and local resistance in several regions.

The problem extends beyond isolated disputes, reflecting a structural tension within market-led energy systems when faced with dense digital loads.

The gains from computing expansion are often concentrated within technology firms, while the infrastructure burden is distributed across society.

What begins as an AI story can quickly become a question of affordability, equity and public legitimacy.

China is attempting a different response. The 2026 Government Work Report emphasizes improving national coordination of computing power, and building hyper-scale intelligent computing clusters.

Computing infrastructure is no longer treated as a stand-alone technology sector, but is planned alongside power generation, grid capacity, transmission networks and long-term energy security.

This is important because electricity is increasingly becoming a strategic productive force rather than just an industrial input. In 2025, China's total electricity consumption surpassed 10 trillion kilowatt-hours, making it the first major economy to cross that threshold.

Meanwhile, power demand from internet data services is expanding dramatically, while electricity consumption in high-tech manufacturing continues to outpace traditional industries. These trends reflect a broader transformation.

Electricity is no longer only powering factories and households, but is also becoming the hidden foundation beneath digital expansion, industrial upgrading and AI deployment.

Seen through this lens, China's 5 trillion yuan ($739 billion) investment in grid infrastructure during the 15th Five-Year Plan (2026-30) is more than conventional stimulus spending. It amounts to a systemic redesign for the computing age.

Ultra-high-voltage transmission projects are increasing the ability to move electricity across regions, allowing energy resources to be allocated on a much larger scale.

That, in turn, is gradually reshaping the geography of computing itself.

As transmission improves, data infrastructure can migrate toward energy-rich regions rather than remaining confined to traditional urban technology hubs.

At the same time, the expansion of local distribution grids, microgrids and energy storage systems aims to absorb growing digital demand without ignoring residential consumption or smaller businesses.

China's broader ambition — to meet most new electricity demand with renewable energy before 2030 — reflects its efforts to expand computing power while keeping carbon and system risks under control.

The underlying challenge is not unique to China. It is global. Electricity is rapidly emerging as a new frontier of geopolitical competition. Energy infrastructure is increasingly vulnerable amid geopolitical tensions and regional conflicts, and energy pricing is frequently distorted by political shocks.

The line separating energy security from digital security is becoming harder to define.

In such an environment, power systems are no longer simply development tools. They have become components of national strategic capacity.

That reality has important implications for the future of AI competition. The decisive question may no longer be who possesses the most advanced chips or the largest models.

It may be who can build an electricity system capable of supporting digital growth without triggering social backlash, grid instability or unsustainable carbon costs.

The AI age is revealing a deeper truth often overlooked in technology discourse: technological breakthroughs require a physical foundation that is not only powerful, but also resilient, equitable and sustainable.

China's emphasis on integrating computing and electricity is ultimately an attempt to answer a larger governance question: how to pursue technological transformation without deepening social strains, and how to sustain rapid innovation without sacrificing systemic stability. It is a test of state capacity in the age of artificial intelligence.

The author is an associate research fellow at the Chongyang Institute for Financial Studies, Renmin University of China.

The views don't necessarily reflect those of China Daily.

If you have a specific expertise, or would like to share your thought about our stories, then send us your writings at opinion@chinadaily.com.cn, and comment@chinadaily.com.cn.