“AI data centers and advanced semiconductors are important, but ultimately, the most important thing is electricity.”
This was emphasized recently by a business executive in South Korea’s semiconductor equipment industry. The remark, made during a discussion about the future of the artificial intelligence (AI) industry, identifies a fundamental requirement that can easily be overlooked amid the current investment enthusiasm. Companies are competing to make better semiconductors, secure more graphics processing units (GPUs) and build larger data centers. But a stable electricity supply is the basic requirement for operating all of them.
As AI transforms industry, the semiconductor market is experiencing an unprecedented boom. Samsung Electronics recorded revenue of 195.3 trillion won and operating profit of 107.4 trillion won in the third quarter of this year. SK hynix also achieved record quarterly results, with revenue of 104.6155 trillion won and operating profit of 83.3338 trillion won. The two companies’ combined quarterly operating profit exceeds 190 trillion won.
This reflects increased investment in AI infrastructure by global technology companies, which has driven demand for server semiconductors, including high-bandwidth memory (HBM). While semiconductor companies benefit from the boom, competition for investment is spreading to data centers.
Samsung SDS recently began construction of an AI data center in Gumi, North Gyeongsang Province. Its initial power capacity is 60 megawatts (MW), with plans to expand to a maximum of 120 MW. Completion is targeted for the end of 2028. A new AI-centered industrial ecosystem is rapidly taking shape, encompassing semiconductors, cloud services and data centers.
However, beyond the impressive results and investment announcements is a challenge that must be resolved: electricity.
AI data centers operate numerous GPUs simultaneously and must also remove the heat generated in the process. This is why they require enormous amounts of electricity and cooling equipment. Even with advanced semiconductors and trillions of won invested in construction, a data center cannot operate normally without a stable power supply.
The problem is that power infrastructure cannot easily keep pace with data center construction. Expanding transmission lines and substations requires considerable time and money. Site selection, permits and community acceptance issues can also cause project delays. Securing investment funding does not mean a company can secure the electricity it needs on time.
This raises the more fundamental issue of responsibility for the costs.
Companies build data centers to secure future growth, while central and local governments expect investment and job creation. But the assessment changes when the burden of grid expansion is also considered. If the boundary between costs to be borne by a data center developer and infrastructure investment for which the state is responsible remains unclear, the burden could fall on other industries and the public.
Society cannot bear all the costs solely on the grounds of promoting the AI industry. Conversely, infrastructure expansion for national industrial competitiveness cannot be left entirely to individual companies. Principles are therefore needed to allocate the benefits and costs of investment reasonably.
Electricity is not a challenge for data centers alone. Stable power supplies are also essential for South Korea’s main manufacturing industries, including semiconductors, batteries, steel and petrochemicals. If grid expansion is delayed while AI data center electricity demand surges, investment in other industries could also be affected. Efforts to develop future industries must not undermine the competitiveness of existing manufacturing.
In the United States, electricity bill pressures and community conflicts arising from growing data center power demand are also emerging as issues. South Korea cannot assume it will be an exception. Before promoting their success in attracting data centers, local governments must assess regional power supply capacity and responsibility for costs.
AI investment is, of course, important for future industrial competitiveness. Rising electricity demand could also create new opportunities for South Korea’s power equipment, generation equipment and nuclear power industries. What matters is that data center investment and power infrastructure expansion proceed together.
Rather than emphasizing only investment amounts and the number of GPUs secured, government and companies must also present plans for securing electricity, the costs of expanding transmission networks and arrangements for community acceptance. The competitiveness of the AI industry depends not on how many data centers are built, but on how reliably and efficiently they are operated.
The record results of Samsung Electronics and SK hynix demonstrate the enormous opportunity created by AI. But a semiconductor boom does not itself guarantee the success of the entire AI industry. Supporting power infrastructure is as important as technological competitiveness.
Semiconductors provide AI computing capabilities, but the industry cannot function without electricity
Government and companies must now compete not only in the speed of data center construction, but also in how quickly they prepare the power infrastructure needed to operate those facilities reliably.
AI leadership requires a reliable electricity supply
Original Korean article: inews24.com
English brief by Korea Wind Intelligence, koreawindintel.com
