How power equipment transforms into an AI infrastructure play



  • AI data centers drive surging power demand, creating both challenges and opportunities for energy resources
  • The power equipment supply chain is transforming into a new favorite in AI infrastructure
  • Exploring different segments of the AI value chain helps capture long-term AI potential

Long viewed as a relatively 'unexciting' sector, power equipment shares have emerged as an integral component of the AI value chain amid the technology's rapid advancement.

AI computing relies on massive data processing, driving the proliferation of data centres and creating both challenges and opportunities for power resources. A single data centre typically consumes about 50-150MW of electricity. Based on the calculation that 1MW can power roughly 1,000 households, the electricity demand of one data centre is equivalent to that of tens of thousands to more than 100,000 homes. The market is anticipating the construction of numerous hyperscale data centres, with electricity demand comparable to that of 500,000 households — approaching the load of a major city. According to data, there are currently over 1,000 hyperscale data centres worldwide, with several hundred additional projects underway. How will the market respond to the surge in power demand driven by AI?


Power Equipment steps into the spotlight as AI infrastructure pillar


The power grid remains the primary source of electricity for AI data centres, but expansion takes time, and voltage can be volatile. To ensure undisrupted AI operations, the electricity supply must be stable and reliable. As countries rush to expand generation capacity, companies are turning to on-site generators as backup power for data centres, elevating the power equipment sector into an AI infrastructure darling.

The power equipment supply chain has three main components: fuel, turbines, and energy storage system (ESS). When the grid experiences a power disruption or fluctuation, on-site turbines convert fuel into electricity. But before that happens, the ESS discharges electricity within a millisecond to stabilise the power supply, averting disruptions to computing operations.

Power equipment companies are spread across the globe, including in the US and Asia. A large-scale US natural gas producer last year secured two mega power projects in Pennsylvania, including the provision of electricity to data centres. The company said that selling directly to end customers would support sustainable organic growth.

As of today, natural gas is gaining traction as a fuel for on-site power generation. Converting natural gas into electricity requires the use of gas turbines. Currently, about two-thirds of global turbines are manufactured by three companies based in Europe, the US, and Japan. Buoyed by robust demand, a US power equipment firm said its turbine production capacity is fully booked through 2028, with under 10% left for 2029.

Within the power equipment sector, ESS plays a key role in discharging electricity promptly to stabilise power and ensure uninterrupted AI computing. A US digital infrastructure technology company said that, driven by strong data centre demand, it saw increases in quarterly sales and margins, among other metrics, and backlog, prompting it to raise its earnings forecast.

Although on-site power supply is becoming more prevalent, the grid remains the primary source for data centres. Reliable, high-capacity transformers that can manage spike loads are needed to deliver grid power to data centres. Supported by advances in AI, a South Korean manufacturer of AI transformers reported that this year's order target is 10.5% higher than last year and plans to expand capacity in South Korea and the US.

From AI innovation to energy infrastructure opportunities

To ensure seamless operation of AI systems, a stable and high-capacity power supply is critical. Demand for power equipment is likely to rise alongside accelerating AI penetration. Beyond focusing on cloud service providers and semiconductor designers, investors seeking to capitalise on AI's long-term potential could also examine other segments of the AI value chain to identify opportunities for allocation diversification.