Korea Diverts Drought Water for Semiconductor Plants

The Water Behind the Wafer

South Korea is building a semiconductor megacomplex in the Honam region and needs 650,000 tons of water every day to keep it running. The government says it has a plan. But the numbers reveal something more unsettling: much of that water was already reserved for emergencies — drought relief for farmers and drinking water for Gwangju residents.

The Climate, Energy and Environment Ministry’s current proposal draws 50,000 tons daily from the surplus capacity of Juum Dam, another 50,000 tons from Dongbok Dam, 200,000 tons from recycled wastewater at Gwangju’s No. 1 Treatment Plant, and 100,000 tons from Naju Dam — agricultural water that was explicitly designated in a 2023 drought contingency plan as a last resort for the Yeongsan and Seomjin river basins. The remainder is supposed to come from a new dam under construction near Dongbok and from supplemental infrastructure for agricultural users.

That adds up to 650,000 tons. But the source material tells a different story than the ministry’s press releases.

Drought Plans, Revised and Reversed

The most striking detail in the documents obtained by Nation Assembly member Na Gyeong-won of the ruling People Power Party is that the government has already changed its plan once. When the three mega-projects were announced in June, the ministry promised 100,000 tons daily each from Jangheung Dam and Boseonggang Dam. Two months later, both were removed from the primary supply list and replaced with a wastewater-recycling plan. The reason: those two dams were also included in the 2023 drought mid-to-long-term strategy. You cannot simultaneously count the same reservoir twice — once as emergency backup and once as industrial supply.

The legislative research service went further. It confirmed that the 100,000 tons from Naju Dam falls squarely within the existing drought contingency framework, which specifically contemplates shifting agricultural water to municipal and industrial use only during extreme drought conditions. Meanwhile, the 50,000 tons attributed to Dongbok Dam’s surplus is, by the ministry’s own characterization in previous documents, Gwangju’s final emergency drinking water reserve.

In other words, the water the government is allocating to semiconductor fabs is water that was never meant to be freely available — it was savings account money, set aside for the worst case.

The Overlap Problem

There is also a second overlap that the ministry has yet to adequately address. The 50,000 tons daily from Juum Dam appears to conflict with the Gwangyang Stage 4 industrial water supply project, a KRW 212.8 billion undertaking that passed its preliminary feasibility study and is designed to deliver stable industrial water to the Yeosu and Gwangyang national industrial complexes during droughts. The ministry says the two allocations do not overlap. But the legislative research service raised the concern, and no independent reconciliation has been published.

This is not a minor accounting issue. It is a pattern. The government appears to be assigning the same water to multiple purposes and hoping that the overlaps cancel out rather than compound.

The Second-Order Effects Begin to Show

The implications stretch well beyond arithmetic. Korea’s semiconductor sector already accounts for roughly a third of the country’s total exports, and the Honam complex is intended to consolidate advanced logic and memory production in a single region. But water is not merely an input — it is a binding constraint that no amount of capital can substitute for. When the government treats drought reserves as ordinary supply, it sends a signal to manufacturers that water risk is being downplayed rather than managed. That signal can distort investment decisions. Companies planning expansions or process upgrades may proceed on the assumption that water will be available year-round, unaware that the underlying supply plan is itself contingent on favorable rainfall.

Farmers in the Yeongsan and Seomjin basins are already experiencing the early symptoms of this distortion. Irrigation allocations have been trimmed in recent dry seasons, and some paddy fields have been left fallow. The 2023 drought contingency plan was designed precisely to prevent further erosion of agricultural livelihoods. Diverting Naju Dam’s allocation to industry reverses that protection before the alternative infrastructure exists to replace it. If the Honam complex reaches full production while agricultural users are still waiting for supplementary facilities, the region could see a generational shift in land use — not through market competition, but through administrative reallocation.

Gwangju faces a different but equally concrete risk. The city’s population is approaching four million, and Dongbok Dam serves as both a routine supply source and an emergency backup. Removing even 50,000 tons from that reserve — what the ministry itself has called the city’s final safety margin — means that any extended dry period could force the municipality into staged rationing or costly emergency water imports. There is no regional pipeline network robust enough to reroute supply from other basins quickly. A drought that produces mild restrictions in the agricultural sector could produce drinking water crises in the city.

Who Pays the Price

If the Honam semiconductor complex draws on drought reserves during a dry year, the consequences will not be shared equally. Farmers around the Yeongsan and Seomjin basins have already experienced increasingly severe water shortages. Gwangju’s population depends on Dongbok Dam as a reliable drinking source. Neither group is represented in the ministry’s supply calculations — they are treated as variables to be managed, not stakeholders with claims.

Na Gyeong-won put it plainly: the government is engaging in what she called a makeshift, knock-kneed approach that takes from the bottom to prop up the top. She warned that unachievable plans amount to torturing Honam residents with false hope. Those are strong words coming from a member of the ruling party, and they signal that the political cost of this story is about to rise.

The political geography matters. Honam has long been treated as the periphery of Korean industrial policy, with investment directed toward the Taeyang corridor and the Seoul metropolitan area. A water diversion plan that sacrifices Honam’s agricultural and municipal security to fuel a semiconductor campus located within Honam itself is easily framed — and is already being framed — as extractive. It is not development of the region; it is development at the region’s expense.

What Happens Next

The ministry says it will secure 1.06 million tons daily overall, including the 650,000-ton semiconductor allocation, and that alternative facilities will protect agricultural supply if Naju Dam water is diverted. None of those alternatives have been built. None of the new dam near Dongbok has completed construction. The timeline for delivering 650,000 tons per day is therefore speculative at best.

Semiconductor fabs do not tolerate interruptions. A single week without process water can cost hundreds of millions in lost output. If the Honam complex ramps production during a drought year — and the government’s own contingency plans suggest that is possible — the region faces a choice between feeding chips and feeding people. There is no third option hidden in the arithmetic.

The deeper question is whether South Korea’s industrial strategy is being built on water that does not exist at scale. The government has already revised its plan once when the math stopped working. It may need to revise it again.

Until then, the Honam semiconductor corridor is being sold as a triumph of industrial policy. But the water behind every wafer produced there may well be borrowed from a future drought — and from the people who live in the basin.