Home > Business > Industry

Korean power, battery companies charge ahead on U.S. 'AI energy highway'

Published: 18 Mar. 2026, 18:10 Updated: 18 Mar. 2026, 19:54

A technician works at an Amazon Web Services AI data center in New Carlisle, Indiana, Oct. 2, 2025. [REUTERS/YONHAP]

Surging power demand from U.S. AI data centers is giving a lift to Korean power and battery companies as U.S. companies accelerate efforts to build what it calls an “AI energy highway” to deliver large volumes of electricity directly to data centers at high speed.

Korean power and battery companies are racing to secure local footholds in the United States as they target demand from AI data centers and seek an early lead in the infrastructure market spanning the full power supply chain, from generation to consumption.

The power infrastructure chain runs from electricity generation through transmission to final delivery to end users such as AI data centers.

Doosan Enerbility, which is targeting the power generation market, said Wednesday that it signed a contract with a local U.S. company to supply two 370-megawatt steam turbines and two generators for power supply to a data center.

Steam turbines are used in combined-cycle power plants, in which a gas turbine is driven by natural gas and the heat generated in the process is used to produce additional electricity, maximizing energy efficiency.

“We will use this order as a stepping stone to expand exports of combined-cycle power models to utilities and independent power producers in North America,” a Doosan Enerbility official said.

This photo provided by Doosan Enerbility shows a gas turbine unit produced by the Korean company. [DOOSAN ENERBILITY]

LG Energy Solution is also moving quickly to secure leadership in the market by expanding its North American energy storage system, or ESS, production bases to five.

The company announced Tuesday it plans to begin mass production in the second quarter after converting part of the Ultium Cells electric vehicle battery plant in Tennessee, which it established jointly with General Motors, into a production line for lithium iron phosphate, or LFP, battery cells for ESS applications.

An ESS stores electricity and supplies it to end users when needed. The technology is considered a key part of power infrastructure for easing imbalances in electricity supply and demand at AI data centers, where power loads can fluctuate sharply, and for improving supply stability.

“North American demand for ESS is rising rapidly as renewable energy expands and power demand from AI data centers surges,” an LG Energy Solution official said. “We have already signed supply contracts with global customers including Tesla and Terra-Gen.”

LS Electric, which is focused on power supply infrastructure, plans to invest a total of $240 million in production bases across North America by 2030 to complete a local value chain spanning design, production and service.

LS Electric's manufacturing plant in Texas, where it will produce medium- and low-voltage electrical equipment and switchgears for local clients. [LS ELECTRIC]

The company plans to invest a total of $168 million in MCM Engineering II, its switchgear manufacturing subsidiary in Utah, to expand the factory sixfold from 13,223 square meters (142,331 square feet) to 79,338 square meters and build three additional production facilities by 2030. The expansion gained momentum after the company recently received approval for tax incentives from the Utah Inland Port Authority.

In addition, LS Electric set up a sales and service office in Dallas, Texas, in the first quarter of this year and plans to add another base in Atlanta, Georgia, later this year.

“Through this campus expansion project, we will further strengthen our ability to respond to the North American market and accelerate growth in our global power infrastructure business,” an LS Electric spokesperson said.

The U.S. data center power market is projected to grow from $16.17 billion this year to $21.89 billion by 2031, according to market research firm Mordor Intelligence.

“New AI facilities in the United States currently have to wait four years to connect to the grid, while the replacement cycle for AI hardware is one year,” said Park Ki-hyun, an analyst at Kiwoom Securities. “This lead-time gap means the ability to secure power has become a key variable in determining whether a company falls behind technologically. The U.S. power market this year and next will enter its strongest period of demand growth since the early 2000s.”

BY KO SUK-HYUN [[email protected]]