Strong power capacity fuels AI boom

As artificial intelligence evolves into a core driver of global industrial competition, China's strong power supply system and its forward-looking computing-electricity synergy strategy will give it an unparalleled edge in rolling out AI infrastructure, according to experts, corporate executives and on-the-ground evidence.

Zhang Qin, former deputy chairman of the China Association for Science and Technology, said, "AI is creating an explosive demand for energy, and the solution lies in safe, reliable and green power. China has great strength in that area.

"Apart from solar and wind power, nuclear energy can also be a crucial power supply for data centers if safety is guaranteed. This is a technological frontier to be conquered rather than a barrier to AI adoption," Zhang told China Daily on the sidelines of the recent World Internet of Things Top 500 Summit in Beijing.

The comments came as China embraces AI with open arms. The value of its core AI industries exceeded 1.2 trillion yuan ($177 billion) last year, maintaining annual growth of more than 20 percent for several consecutive years, according to the Ministry of Industry and Information Technology.

Furthermore, the country is turning AI into real-world applications and a driver of economic growth. In the first half of this year, China's industrial output of AI-related sectors, including integrated circuit manufacturing and in-vehicle intelligent equipment manufacturing, grew more than 30 percent year-on-year.

Wei Zhigang, chairman of China National Nuclear Corp Hainan Nuclear Power Co, said that as AI triggers unprecedented global demand for computing power, "the ultimate constraint has become clear: the need for vast, stable electricity supplies."

Tech giants such as Microsoft and Google are exploring the use of small modular nuclear reactors, or SMRs, to power AI data centers.

Wei added that CNNC's Linglong One — the world's first onshore commercial SMR — is nearing completion, and the company plans to begin commercial operations later this year.

One key advantage of SMRs, he said, is their flexible placement, allowing them to be deployed right where computing power is needed for close load-matching.

Wei said that the company is exploring the construction of a nuclear-powered, zero-carbon industrial park in Hainan province's Changjiang Li autonomous county, with the aim of creating a "nuclear energy plus computing power" pilot zone, in order to help solve one of the most pressing challenges the world faces in applying AI.

Zhou Ji, a senior researcher at market consultancy Pangoal, said that "China's energy system has an edge in three dimensions: scale, structure and system, which constitutes solid support for AI computing infrastructure at the foundational level".

Regarding scale, China ranks first in the world in total electricity generation and also leads in incremental growth. In terms of structure, the country has the world's largest installed capacity of new energy, with the share of green power continuing to rise, Zhou said. As for the system, China's ultra-high-voltage grid technology enables cross-regional optimization of power resources, while the country is also adopting a pioneering approach that aligns computing infrastructure with clean energy sources, he added.

Power strain

Looking ahead, the Beijing Institute of Technology's Center for Energy and Environmental Policy Research projected that by 2030, China's data centers will consume about 525.76 billion kilowatt-hours of electricity, accounting for 4.8 percent of the country's total societal electricity use.

In the United States, meanwhile, the amount of electricity consumed by data centers is expected to cause significant tightening across the national power market.

That strain is already feeding through to the real economy, as big tech data centers are driving up power bills for manufacturers across the US Rust Belt, Reuters reported. For example, the Belden Brick Co in Sugarcreek, the US state of Ohio, a 141-year-old enterprise, saw its electricity costs surge by 90 percent last year, largely due to rising power demand from regional data centers.

Its monthly capacity charge has recently increased from $1,600 to $12,000. Belden Brick is not alone, as Reuters added that many manufacturers across the US heartland are facing similar cost increases as power-hungry AI data centers proliferate.

China, by contrast, is taking a coordinated, long-term approach. Wang Hongzhi, head of the National Energy Administration, said that China's energy system will become significantly more secure, resilient and structurally optimized by 2030, driven by over 20 trillion yuan in planned investments for key energy projects and emerging sectors.

Localized needs

To accommodate the surging electricity demand driven by AI, Wang said China will closely align national computing hubs with large renewable bases in the western regions, advancing the synergy between computing and electricity, while simultaneously promoting distributed power, microgrids and virtual power plants to meet localized computing needs.

A virtual power plant is a network of small, distributed energy resources, such as solar panels and electric vehicle chargers, which are connected and managed as one.

Song Yunbo, chief sustainability officer of GCL Group, a major Chinese new energy company, said that China "already generates over one-third of its electricity from renewables — a global leader — and the country's energy surplus makes it an ideal testing ground for AI-driven applications". She noted that Chinese energy companies can incorporate AI into management to increase efficiency without massive job cuts.

Miao Qing, chairwoman of Yingli Group, another Chinese new energy company, said that AI-powered virtual power plants are reshaping electricity systems by aggregating distributed renewables, storage and flexible loads to smooth peaks and valleys while dynamically responding to real-time market prices.

Miao said China has already built more than 530 virtual power plants, and the prospect for wider application is "very promising".

According to Miao, Yingli has built its own multi-energy coordinated control platform and is exporting microgrids to Southeast Asia and Europe and building integrated photovoltaic solutions for use there.