Largest AI Investment in History! SK Group to Invest 2100 Trillion KRW Over 10 Years to Expand HBM Production and Build 15GW Data Centers

South Korea's SK Group has announced its 10-year AI development roadmap, planning to invest up to 2100 trillion won in semiconductor manufacturing and artificial intelligence (AI) data centers. Of this, 1100 trillion won will be used to expand the chip supply chain, and 1000 trillion won will be invested in building AI data centers. Leveraging its two core subsidiaries, SK Hynix and SK Telecom, the group aims to build an integrated industrial chain covering chips, computing power, and AI services, further strengthening South Korea's global competitiveness in the AI industry.

SK Group stated that the overall investment will be divided into two main areas. Approximately 1100 trillion won will be invested in strengthening the semiconductor supply chain, including production capacity for AI-critical memory such as high-bandwidth memory (HBM), next-generation DRAM, and NAND Flash. The remaining 1000 trillion won will be used to build AI data centers, aiming to transform South Korea from an AI technology user into an exporter of intelligent technology and AI infrastructure.

Chairman Choi Tae-won stated that South Korea should not merely export AI products but should export intelligent technologies themselves and cultivate a domestic AI market. Therefore, the group will make large-scale investments in 'AI factories'—ultra-large AI data centers—that will serve as a foundational pillar for future AI industry development.

In semiconductor development, SK Hynix will be the primary investment vehicle, investing 1100 trillion won to expand AI memory production capacity. The group plans to invest 100 trillion won in expanding facilities at its Cheongju site, 400 trillion won in building a comprehensive chip manufacturing base in South Korea's southwestern semiconductor cluster, and 600 trillion won in the Yongin semiconductor super cluster to accelerate the development of the group's core chip production base.

Notably, SK Group has advanced the completion timeline of the Yongin super cluster from the originally planned 2045 to 2033—12 years earlier—demonstrating the company's accelerated response to global AI chip demand.

Choi emphasized that the rapid growth of AI applications has led to explosive memory demand, with supply shortages already emerging in the market. He warned that if chip supply remains insufficient, not only will the memory industry be constrained, but the entire AI industry's development will also be affected.

Market research firm TrendForce recently noted that demand for AI servers continues to drive HBM shortages. SK Hynix remains the global leader in the HBM market and continues to supply AI chipmakers such as NVIDIA (NVDA-US) and AMD (AMD-US). High-end HBM product capacity has already been sold out several quarters in advance, and the market is expected to remain supply-constrained until at least 2027.

In addition to chips, SK Telecom will shoulder the responsibility of building AI computing infrastructure, planning to invest 1000 trillion won in AI data centers. The goal is to achieve a total installed capacity of 15GW by 2035, with the first phase completing 5GW by 2029, followed by an additional 10GW expansion based on market demand.

An initial investment of approximately 140 trillion won will establish an AI computing hub in South Korea's Yeongnam region. A 100MW ultra-large AI data center will be built in Ulsan, scheduled to launch in Q4 2027, later expanding to 900MW. Additional 1GW-class AI data centers are also planned in other parts of Yeongnam.

Chairman Choi stated that while traditional data centers primarily store data, AI data centers are 'factories that produce intelligence.' They will support industrial robots, smart manufacturing, and various physical AI applications, becoming a critical infrastructure for South Korea's AI industry.

SK Telecom CEO Jung Jin-ho stated that by integrating AI data centers with the group's semiconductor manufacturing capabilities, Yeongnam can become a key base for industrial AI technology verification and commercialization in South Korea.

This large-scale investment reflects the ongoing expansion of AI infrastructure by global tech giants. According to market research firms and Wall Street analysts, by 2026, major U.S. cloud service providers—including Microsoft (MSFT-US), Alphabet (GOOGL-US), Amazon (AMZN-US), Meta Platforms (META-US), and Oracle (ORCL-US)—are expected to collectively spend 600 to 750 billion USD annually on capital expenditures, primarily on AI data centers, GPU servers, and power infrastructure.

SK Group stated that while its investment scale is still smaller than that of U.S. tech giants, it is sufficient to establish its leadership position in the Asia-Pacific AI industry.

SK Telecom has been actively transforming from a traditional telecom operator into an AI platform company. It has already partnered with NVIDIA to launch GPU-as-a-Service (GPUaaS), aiming to become a key AI computing power provider in the Asia-Pacific region.

Meanwhile, SK Group is also advancing its global AI strategy. Earlier this year, SK Telecom announced a 738 billion won investment in AI Co., a U.S.-based AI holding platform. AI Co. was established in January this year, restructured from Solidigm, a subsidiary of SK Hynix, and headquartered in the U.S. It is responsible for the group's overseas AI investments and global industrial ecosystem development. Funding includes 10 billion USD from SK Hynix, 250 million USD from SK Holdings, and 380 million USD from SK Innovation.

Chairman Choi also revealed that over the next 10 years, SK Group will maintain an annual domestic capital expenditure of approximately 100 trillion won, continuously investing in AI data centers and semiconductor construction based on market demand. He stated that current order books indicate strong global AI demand, so the group will continue to ramp up investments to build a complete AI industrial chain.

Analysts believe that by integrating SK Hynix's AI memory advantages with SK Telecom's computing infrastructure, SK Group not only secures key positions in the AI supply chain but also has the potential to transform from a memory supplier into a comprehensive tech group offering chips, computing platforms, and AI services—positioning itself to compete head-on with U.S. tech giants in the global AI infrastructure race.