The current surge in memory chip demand is not cyclical but a permanent structural shift caused by AI, which will create a supply-demand imbalance that persists until society integrates AGI.
The United States and China are pursuing fundamentally different AI strategies: the US focuses on high-end token quality, while China leverages its large talent pool to drive down token costs.
SK Group is committed to massive, escalating investments in the United States, viewing it as a critical market for capital, technology partnerships, and future growth in AI infrastructure.
The future of computing will involve individuals using hundreds of personal AI agents, which will drive exponential growth in demand for memory capacity far beyond current projections.
SK Hynix's core partnership strategy is to avoid competing with its customers, enabling deeper collaboration and long-term supply agreements in a capacity-constrained market.
Circa 2009
SK Group acquired Hynix approximately 15 years ago, when the company was under a bank-controlled workout program.
Recent Past
SK Group invested over $35 billion in the United States across its bio, battery, and semiconductor businesses, establishing a significant operational footprint.
Present
Chey Tae-won has led SK Hynix to a US listing to access American capital markets for its expansion plans. He has publicly articulated his analysis of the AI-driven memory demand boom and the divergent US vs. China AI strategies.
Next 3 Years
Predicts the cost per token for AI will decrease to one-fifth or one-tenth of its current level, while individuals will begin using tens or hundreds of personal AI agents.
Next 5 Years
Plans for SK Hynix to double its entire production capacity, which includes building a new fabrication plant in Indiana, to meet a fraction of the surging customer demand.
Next 10 Years
Announced plans for SK Group to invest nearly $1 trillion to build 20 gigawatts of AI data center capacity (15 GW in Korea, 5 GW abroad) to become a key infrastructure provider.
▶Aggressive US Investment and AlignmentJul 2026
Chey Tae-won is spearheading a major strategic push into the United States, building upon an existing $35 billion investment with plans for 'much, much bigger' commitments. This includes building new fabrication plants, like the one in Indiana, and establishing AI data center capacity to tap into the US capital markets and technology ecosystem.
This deep investment in US infrastructure signals SK Group's strategy to align with the primary AI market, mitigate geopolitical risks in Asia, and position itself as a critical partner in the American technology supply chain.
▶The AI-Driven Memory SupercycleJul 2026
Chey believes the AI era has fundamentally and permanently altered the memory chip market, shifting demand from being tied to consumer devices to the near-limitless needs of AI. He forecasts that supply will not be able to meet this demand until humanity reaches a 'settlement with AGI,' justifying massive, ongoing capital expenditures.
By framing the current demand surge as a permanent, structural shift rather than a cyclical boom, Chey is making a long-term bullish case for the entire memory sector, suggesting sustained pricing power and growth for key suppliers like SK Hynix.
▶Geopolitical AI Strategy AnalysisJul 2026
Chey frequently analyzes the global AI landscape, differentiating between the US strategy of focusing on high-end, quality tokens and China's strategy of leveraging its vast human capital to lower the cost per token. This framework appears to be a core component of his global business planning.
This analysis reveals that SK Group's strategy is not just about technology but also about navigating the US-China tech rivalry by positioning itself as a key supplier to the high-value, quality-driven US market.
▶Pivoting to AI Infrastructure ProviderJul 2026
Beyond chip manufacturing, Chey is outlining a vision for SK Group to become a major AI infrastructure player, evidenced by a planned investment of nearly $1 trillion over 10 years to build 20 gigawatts of AI data center capacity. This is complemented by exploring future business models like 'memory as a service'.
This strategy indicates a move up the value chain from being a component supplier to an integrated infrastructure provider, aiming to capture a larger share of the profits from the AI boom and create more stable, service-based revenue streams.