SK Group Invests โHeavily in HBMโ Production to Meet Surging AI Demand, chairman Choi Announces
ULSAN, SOUTH KOREA – November 3,โฃ 2025 – SK Group is dramatically expanding its high-bandwidth memory (HBM) production capacity to address unprecedentedโค demandโ from artificial intelligence โleadersโค like OpenAIโข and NVIDIA, Chairman Taewon Choi โrevealed today at the SK AI Summit โข2025.โฃ the โขdeclaration signalsโค a major strategic shift for the South Korean conglomerate as it positions itself toโค capitalize on the rapidly growing AI market.
Choi disclosed plans for a newโ HBMโ fab in Cheongju, stating it will have a production capacity equivalent to “24 units ofโฃ Cheongju fab,” a response to โคthe growing cluster of โคAI โdemand centeredโข in Yongin. โคHe emphasizedโ the investment is criticalโข to “prevent supply shortage,” citingโ a specific request from OpenAIโข for “900,000 HBM per month for โtheโค ‘Stargate Project’,”โข a โขfigure he noted isโค “twice the โmonthly โproduction of HBM aroundโฃ the world.” The demand underscores the intense โcompetition for advanced โฃmemoryโ crucialโฃ for powering โฃlarge language โmodels โand otherโค AI โapplications.
Beyond expanding โproduction, SK HynixS technological โadvancements have reached a point where “NVIDIAโฃ no longer โdemands advancementโ speed from SK Hynix,” accordingโ to Choi, signifying industry-wide recognition โof โฃthe company’s โคcapabilities. SK Group’s broaderโฃ AI strategyโค focuses on efficiency, encompassing AIโฃ infrastructure development โขand the implementation ofโฃ AI in โmanufacturing processes.This includesโ buildingโฃ a self-designed data center infrastructure, aโข 1-gigawatt AIโ data center inโ Ulsan with AWS (scheduled forโค 2027), andโค a โnew โขdata center in Cheonanโฃ jointly developed โขwith OpenAI.
SK Hynixโข is also building a virtual โฃfactory based on NVIDIA’s ‘Omniverse’ to โขcreate a fully โขautonomous manufacturing system. Choi โคstated the โcompany will open this manufacturing AI solution to other industries, aiming to expandโ the AI ecosystem. He concluded by stressing โคthe importance of collaboration, highlighting partnerships with AWS, NVIDIA, and OpenAIโ asโข essential to โขbuildingโค anโ “AI efficiency ecosystem.”