“Whenever I meet someone, they ask me for memory chips. It would be great if I could give them, but production is fixed and we are struggling with supply shortages.”
That was what SK Group Chairman Chey Tae-won said on the 28th at the National Assembly. The complaint, coming from the head of SK Group, which controls SK hynix, was neither mere boasting nor self-pity. It was a scene that simultaneously revealed both sides of the rope 한국 industry is walking in the AI era.
At a policy seminar hosted by the Korea-China Parliamentary Forum that day, titled “Korea’s Growth Strategy Amid the U.S.-China AI Technology Hegemony Competition,” Chey took the stage.
Although he spoke in his capacity as chairman of the Korea Chamber of Commerce and Industry, the weight of his remarks was closer to the perspective of the head of the parent company of SK hynix, standing at the center of the semiconductor supercycle. What the chairman of a company posting record profits called a “happy dilemma” was ultimately a map of the bottlenecks Korean industry must solve.
◎A shortage of memory chips could even stop refrigerators
The first landscape Chey pointed to was the concentration in the memory market. With demand for high-bandwidth memory (HBM), which AI data centers are devouring, surging, a paradox has emerged: there are not enough memory chips left for home appliances, PCs, and cars.
His remark that “we cannot supply memory to refrigerators, TVs, and PCs on the other side” was read as a diagnosis that Korean manufacturing as a whole is becoming hostage to the semiconductor supply chain.
The core issue is not price but quantity. If memory prices rise too much, industries begin researching ways to “use less memory,” and in the long run Korea’s own demand for semiconductors could weaken, Chey warned. The message is that while enjoying the current boom, we must recognize it cannot last forever. It is a perspective looking beyond the peak of the supercycle.
When asked whether expanding production facilities would solve the problem, the answer was not simple. His short phrase, “land, electricity, and money are needed,” condensed the structural limits of Korea’s semiconductor industry.
From site acquisition to power supply and capital investment, Korea faces competition from the United States, China, and Taiwan on all three fronts. His remark that Samsung is in the same position signaled that this is not a management issue for individual companies alone.
◎If there is not enough electricity, AI stops too
Another bottleneck is power. Chey identified four areas whose supply is not smooth in the AI era: electricity, energy, GPUs, and memory chips. Memory and GPUs are issues of the global supply chain, but electricity and energy are problems Korea must solve itself.
Power shortages are no longer a temporary phenomenon during peak appliance seasons. The era has begun in which a single data center consumes as much electricity as a mid-sized city.
The United States is reviving decommissioned nuclear plants through Microsoft and Amazon, while China is bypassing the problem by building data centers directly next to renewable energy complexes.
Korea, meanwhile, has not fully prepared either path and is walking a tightrope between concentrated data center demand in the Seoul metropolitan area and the burden on transmission networks bringing power in from the Yeongnam and Honam regions.
When asked what he thought of a system that would offer discounts if industrial electricity charges were paid in advance for one-year or three-year periods, Chey avoided giving an immediate answer. His response was cautious: “It is the first time I have been asked that, so I will think it through and reply.”
What he emphasized more strongly, however, was the restructuring of the generation system itself. His question—whether it is sufficient for everything to be centrally controlled, dominated by one entity, and distributed from there—reflected skepticism toward the single power transmission and distribution structure centered on Korea Electric Power Corporation.
The keyword he presented as a solution was distributed generation. The idea is to use electricity where it is produced.
Instead of the current structure, in which electricity generated on the eastern coast must be transmitted to the Seoul metropolitan area, overloading the grid, the alternative is to locate data centers and factories near power plants. The key question is whether Korea can apply the model seen in Texas, where data centers are being attracted next to wind power complexes and becoming hubs of the global AI infrastructure.
◎A sharp rebuke to legalism
The final part of his remarks targeted legislative overreach. In response to a request to “say a word to the National Assembly,” Chey shot back: “Are you saying solve it with law, or make laws to restrict business activity? It would be nice if the public could know what the target is.”
He did not name any specific bill. But given that the comment came at a time when businesses are being shaken by the implementation of the Yellow Envelope Act, the message was clear.
He did not stop there. In response to Korea-China Parliamentary Forum chairman Kim Tae-nyeon’s emphasis on a “one-team Republic of Korea” in his opening remarks, Chey said, “You have to go to the site together.” He added, “You need to come as players and see it; if you look at it as a lawmaker, it becomes hard to solve,” in an oblique criticism of the distance between lawmakers and industrial sites.
It was a request to see directly how laws assembled on a desk can stop the gears of factories and research labs.
◎Translating complaints into policy
There is no need to overreact to a single seminar remark. Still, if Chey’s observations are translated into policy language, they sketch the list of tasks Korean industry must tackle over the next five years.
The memory supply bottleneck cannot be solved by simply expanding capacity. A package of support that combines faster site approvals, guaranteed power supply, and tax incentives is needed. That is how the United States is moving with the CHIPS Act, and how Japan is supporting Rapidus.
Korea is already pushing ahead with the Yongin semiconductor cluster, but progress on site confirmation and power infrastructure is slow. The supercycle will not last long enough for Korean companies to lock in market share.
The power bottleneck is a matter of redesigning both generation structure and demand locations. Chey’s emphasis on distributed generation ultimately requires easing the concentration of data centers in the Seoul metropolitan area.
For government incentives for data centers outside the capital region to work properly, telecom networks, labor supply, and local government cooperation must all follow. Bringing electricity production and consumption closer together is not merely an urban planning policy; it changes the country’s entire industrial map.
The balance between regulation and legislation is the most politically sensitive area. Protecting workers’ rights, preserving business freedom, environmental regulation, and industrial competitiveness are all values that cannot be sacrificed on either side. Still, as Chey pointed out, the habit of using law as the solution to every problem deserves reconsideration.
Work should first be done to distinguish between matters that can be handled through detailed enforcement decrees, guidelines, or industrial self-regulation, and those that require legal measures.
◎What awaits at the end of the “happy dilemma”
Chey’s complaint showed where the shadow of Korea’s current boom comes from. The memory supercycle is clearly a blessing, but a structure in which one industry depends on it while others suffer from a drought of memory chips cannot be called healthy.
Korea must also solve the contradiction that it cannot endure the AI era without building more power plants, yet the task of building them is becoming increasingly difficult.
As the United States and China clash directly over AI hegemony, Korea stands in a position supplying key components between the two sides.
That position is both an opportunity and a tightrope. To avoid falling, the gears of industrial policy must mesh precisely. How to regrind those gears before Chey’s happy dilemma ends will determine the next five years of the Korean economy.