Chipmakers’ rejection of prepaid power bills warrants reflection

Samsung Electronics and SK hynix rejected the Korea Electric Power Corporation's (Kepco) proposal that they prepay 25 trillion won ($18.6 billion) in electricity bills for five years. Kepco, burdened with debt exceeding 210 trillion won, planned to use the money primarily to expand the national power grid, accelerating transmission and substation construction for semiconductor clusters in Yongin, Gyeonggi, and the southwestern region.

The proposal appeared advantageous for the chipmakers as well. They could secure timely electricity supplies for their semiconductor complexes, while receiving interest on the advance payment above the yield on two-year government bonds.

The companies reportedly rejected the proposal because they questioned whether the semiconductor boom would last. In short, they could not be certain that today’s semiconductor supercycle would continue for five years.

Their caution contrasts with the Lee Jae Myung administration’s optimism. Former presidential policy chief Kim Yong-beom cited the irreversible nature of AI demand, arguing that demand for the technology, now industrial infrastructure, would persist. If memory and infrastructure demand in the AI era represents long-term structural change, he said, Korea could for the first time become a country generating sustained excess profits.

That optimism helped inspire the controversial proposal for a “national dividend.” Yet while companies with the clearest view of market conditions remain cautious, the government has been dreaming up policies based on current performance. Fiscal programs resembling cash handouts, premised on prolonged semiconductor prosperity and excess tax revenue, should be reviewed from scratch.

The semiconductor cycle is one concern, but China’s rapid pursuit is greater. As the JoongAng Ilbo's “China’s semiconductor offensive” series reported, Chinese chipmakers are catching up at extraordinary speed, backed by government support, a huge domestic market and a research ecosystem.

The technology gap between Korea and China in memory chips has reportedly narrowed to three years in high bandwidth memory, two years in dynamic random-access memory and one year in NAND flash. Korea must prepare meticulously for the possibility that Chinese companies, supported by a nationwide mobilization of resources, will eventually launch a price war.

This is no time to become intoxicated by the sweet dream of a semiconductor superboom. Above all, policymakers must soberly examine whether the rush to build a semiconductor cluster in the southwest has itself been driven too heavily by optimism about AI.

Samsung and SK hynix have effectively reminded policymakers of a basic principle: Companies investing their own money must prepare for downturns as well as booms. The government, which is spending taxpayers’ money and shaping long-term industrial policy, should be at least as cautious.