In a move that signals a tectonic shift in the global technology landscape, South Korean industrial titan SK Group has unveiled an audacious 2,100 trillion won ($1.36 trillion) investment roadmap. This multi-decade strategy is designed to cement the conglomerate’s dominance in the AI value chain, effectively transforming the group from a hardware-centric component manufacturer into a full-stack, "intelligence-exporting" powerhouse. By aligning its world-leading semiconductor production with massive-scale data center infrastructure, SK Group is positioning South Korea as the critical fulcrum of the Asia-Pacific AI ecosystem.
The Strategic Pivot: Exporting Intelligence, Not Just Chips
At the heart of this transformation is a philosophical shift led by Chairman Chey Tae-won. Speaking to the Yonhap News Agency, Chey articulated a vision that moves beyond traditional manufacturing exports. "We should not simply export AI products," Chey noted. "We need to export intelligence itself while building a domestic market for AI-driven intelligence. To achieve that, we will rapidly build AI factories in the form of large-scale AI data centers."
This strategy acknowledges that the future of the global economy will be defined by compute availability. By integrating the upstream production of high-performance memory with the downstream infrastructure of hyperscale data centers, SK Group aims to capture value at every stage of the AI lifecycle. This is not merely an investment in infrastructure; it is an investment in the intellectual capital of the nation.
Chronology of Ambition: The Roadmap to 2035
The deployment of 2,100 trillion won is not a singular event but a phased, sustained capital commitment spanning the next ten years. The roadmap is structured to ensure that South Korea’s domestic infrastructure matures in lockstep with the global demand for generative AI.
- 2026–2027: The Foundation Phase: SK Telecom initiates the construction of its first 100MW hyperscale data center in Ulsan, scheduled for completion in Q4 2027. Simultaneously, SK Hynix accelerates its capital expenditure to ramp up production of HBM3E and HBM4 memory modules.
- 2029: The Interim Milestone: SK Telecom targets 5 GW of operational AI data center capacity, setting the stage for the first wave of regional "AI hubs" across South Korea.
- 2030–2034: Scaling the Ecosystem: SK Group maintains an average annual capital expenditure of 100 trillion won ($65.3 billion), focusing on refining its GPU-as-a-Service (GPUaaS) offerings and integrating AI into domestic manufacturing processes.
- 2035: The Zenith: The group aims to reach its total capacity of 15 GW of AI data center power, solidifying South Korea’s role as a sovereign AI powerhouse capable of supporting massive-scale, localized LLM (Large Language Model) training and inference.
Semiconductor Supremacy: SK Hynix’s 1,100 Trillion Won Bet
SK Hynix, the group’s semiconductor crown jewel, is spearheading the hardware allocation with a staggering 1,100 trillion won ($706 billion) investment. As the global demand for High-Bandwidth Memory (HBM) skyrockets—driven by the insatiable requirements of NVIDIA’s GPU architectures—SK Hynix is moving to defend its market leadership.
The investment is targeted at three critical pillars:
- HBM Scalability: Expanding clean-room capacity for HBM3E and future iterations to ensure that the memory bottleneck—a common failure point in AI training—is effectively solved.
- Next-Generation DRAM/NAND: Moving beyond commodity memory to specialized, AI-optimized NAND flash components that provide the low-latency storage required for real-time inference.
- Process Node Optimization: Investing in advanced lithography and packaging techniques to reduce power consumption per bit, a crucial metric for the sustainability of massive AI data centers.
Telecommunications Evolved: SK Telecom’s "AI Factory" Vision
Perhaps the most radical departure from traditional corporate models is the evolution of SK Telecom. By allocating 1,000 trillion won ($642 billion) to physical infrastructure, the operator is shedding its identity as a legacy connectivity provider to become an "AI Factory."
During a public briefing with South Korean President Lee Jae Myung, SK Telecom CEO Jung Jai-hun emphasized the regional impact of this infrastructure. "The massive AI data centers could transform the region into a hub for the verification and expansion of manufacturing AI, when combined with the manufacturing capabilities in the region," Jung noted.
By focusing on the Yeongnam region, SK Telecom is creating a blueprint for "Industrial AI." This involves co-locating data centers with high-energy manufacturing zones, allowing for low-latency AI processing that can optimize everything from automobile assembly to chemical synthesis in real-time.
The Global Context: SK Group vs. The Hyperscalers
SK Group’s investment is occurring against the backdrop of an unprecedented global "arms race." US hyperscalers—Microsoft, Alphabet, Amazon, Meta, and Oracle—are forecasted to collectively spend between $600 billion and $750 billion in 2026 alone.
While SK Group’s total spend is spread over a longer horizon, its strategic concentration is unique. Unlike US tech giants that primarily serve global markets, SK Group is creating a sovereign AI ecosystem. By controlling the memory chips (SK Hynix) and the network infrastructure (SK Telecom), the group minimizes its reliance on foreign dependencies.
Furthermore, SK Telecom’s strategic alliance with NVIDIA for GPU-as-a-Service (GPUaaS) underscores its intent to become a regional provider for enterprises that cannot afford to build their own infrastructure but require the compute power of high-end NVIDIA chips.
International Ambitions: The "AI Co." Subsidiary
SK Group’s vision is not confined to the Korean peninsula. Last month, the conglomerate announced a significant infusion of capital into "AI Co.," a US-based subsidiary formed by repurposing Solidigm, the flash memory firm previously acquired by SK Hynix.
Backed by a $10 billion investment from SK Hynix and supplemental funding from SK Inc. and SK Innovation, this entity serves as the group’s strategic bridge to Silicon Valley. By establishing a physical and operational foothold in the United States, SK Group is ensuring that it remains at the table for global AI standards and collaborative research, effectively bridging the gap between Korean manufacturing efficiency and American software innovation.
Implications for the Future
The implications of this multi-trillion-won strategy are profound:
- Supply Chain Dominance: By unifying upstream chip production with downstream infrastructure, SK Group creates a vertically integrated ecosystem that is largely immune to the volatility of global third-party supply chains.
- Energy Transition: Building 15 GW of capacity in South Korea will necessitate a radical overhaul of the nation’s energy grid. This will likely trigger parallel investments in renewable energy and perhaps Small Modular Reactors (SMRs) to power the cooling and processing requirements of these "AI factories."
- Regional Economic Revitalization: The focus on the Yeongnam region is a clear signal of government-corporate cooperation to decentralize South Korea’s economy. By moving high-tech infrastructure out of the capital (Seoul) and into industrial hubs, SK Group is effectively "de-risking" the nation’s tech economy.
- Full-Stack Powerhouse: For the tech industry at large, SK Group is emerging as a "Full-Stack Intelligence" provider. Whether a client needs high-performance memory modules or a secure, low-latency GPU cloud for localized training, SK Group will soon possess the capability to fulfill both requirements.
Conclusion
SK Group’s 2,100 trillion won roadmap is more than a financial commitment; it is an act of industrial statecraft. By aggressively scaling both the physical and logical components of the AI revolution, the conglomerate is ensuring that South Korea remains a primary architect of the intelligence age. While the competition from global hyperscalers remains fierce, SK Group’s unique ability to control the "atoms" of the industry—the chips—alongside the "bits"—the data centers—provides a distinct competitive moat. As the world transitions toward an AI-first economy, SK Group has positioned itself not just as a participant, but as a foundational pillar of the next century of global technological development.
