SK Group is pushing ahead with construction of a cogeneration plant to supply power and heat to the Yongin semiconductor cluster led by SK hynix (000660.KS), bringing the project on track as the conglomerate concentrates on building an artificial intelligence (AI) value chain. With uncertainty over energy infrastructure essential to chip plant operations now resolved, fab expansion work is also expected to gain momentum, industry observers said.
According to industry sources Wednesday, SK Innovation (096770.KS) E&S, in partnership with Korea Midland Power, recently broke ground on the Yongin liquefied natural gas (LNG)-based cogeneration plant after completing an environmental impact assessment earlier this year. The plant will supply heat to SK hynix’s semiconductor facilities at the Yongin cluster. With a generation capacity of 1.05 gigawatts (GW), it will be able to meet all the heat requirements of the four fabs planned for the cluster. Completion is targeted for 2030.
Total project cost is estimated at approximately 3.3 trillion won. SK Innovation E&S and Korea Midland Power are negotiating the equity structure of a joint special purpose company (SPC), with a 50-50 stake split seen as the most likely arrangement. The two companies secured final government approval for the project in 2024, but actual construction began only this year due to extended preliminary work including the environmental assessment.
The project is expected to gain further momentum as a beneficiary of the National Growth Fund launched last year. The government is sequentially supporting seven projects through the fund, including energy infrastructure development within the semiconductor cluster. The LNG cogeneration plant produces both heat and electricity simultaneously, delivering energy efficiency about 30% higher than conventional thermal power plants.
With construction of the plant progressing smoothly, SK Group Chairman Chey Tae-won’s “AI full stack” strategy is also expected to move into high gear. AI full stack refers to the capability to provide all the layers needed to deliver AI services in an integrated manner, including infrastructure, hardware, and platforms. SK is concentrating on completing its AI value chain centered on key affiliates including SK Innovation (energy), SK hynix (semiconductors), and SK Telecom (017670.KS) (AI data centers).
Once the energy infrastructure is in place, construction of the semiconductor cluster can also speed up. With the global memory chip shortage deepening amid the spread of AI, Korean chipmakers are racing to build out their fabs. SK hynix broke ground on the second phase of the first fab building at the Yongin cluster in August and has reportedly recently finalized plans for a second fab. The opening of the first clean room, originally scheduled for May next year, has also been moved up by about three months.
SK Telecom is building a hyperscale AI data center (DC) worth approximately 7 trillion won in Ulsan in partnership with Amazon Web Services (AWS). The Ulsan AI DC will start at a 40-megawatt scale next year, with plans to expand generation capacity to more than 1GW in the future. The company has also set a mid- to long-term plan to achieve 1 trillion won in AI DC revenue by 2030.
As SK positions AI as the centerpiece of the group’s future growth, other affiliates are also seeking synergies in related businesses. In Ulsan, where SK’s AI DC will be located, LNG processed by SK Gas (018670.KS) is supplied to SK Multi Utility’s cogeneration plant for power generation. SK Ecoplant is expanding its business portfolio from semiconductor plant engineering, procurement, and construction (EPC) to AI DC facilities. SK AX is also strengthening its AI service capabilities by partnering with global Big Tech firms.
In addition, SK will hold a combined forum integrating its first-half strategy meeting and knowledge management platform “Icheon Forum” for the first time on Dec. 11-13 at the SKMS Institute in Icheon, Gyeonggi Province. Chairman Chey and CEOs of major affiliates will gather to discuss strategies for securing AI leadership.