In a Facebook post on Tuesday, Kim wrote that AI had evolved into a capital-intensive hardware industry, making scarce commodities such as graphics processing units (GPUs), memory chips, transmission lines and electricity far more decisive than software code.
"Intelligence spreads and is replicated quickly. Models are caught up to. Code proliferates. But power plants, transmission networks and semiconductor fabs cannot be copied overnight," Kim said.
Kim singled out what he described as a looming paradox for Asia's fourth-largest economy: SK hynix and Samsung Electronics produce the world's most advanced high-bandwidth memory (HBM) destined for Nvidia GPUs in overseas data centers, yet South Korea itself lacks sufficient large-scale AI computing clusters to harness the technology at home.
The policy chief stressed that while South Korea does not face an outright electricity shortage, the deeper challenge lies in delivering power at the scale and speed that AI demands.
Kim also championed the principle of local production and consumption of electricity, insisting that power-generating regions should share in the industrial benefits.
The remarks come as South Korea prepares to draft its 12th basic plan for electricity supply and demand, a 15-year blueprint covering 2026 through 2040 that will shape the country's energy mix amid surging demand from AI data centers.
The government earlier committed to constructing two large-scale nuclear reactors under the 11th electricity supply plan finalized in February 2025, signaling its intent to align energy policy with the power-hungry demands of next-generation industries.
Copyright ⓒ Aju Press All rights reserved.