South Korean President Lee Jae Myung unveiled a national investment plan on Monday committing the country’s two largest memory chip manufacturers and a consortium of technology and energy conglomerates to spend more than $900 billion through 2035 across semiconductor fabrication, AI data center construction, and physical AI development – the most ambitious industrial mobilization the country has undertaken since its original push into semiconductor manufacturing decades ago. Samsung Electronics and SK Hynix will jointly invest 800 trillion won, approximately $518 billion, in four new memory fabrication plants concentrated in South Korea’s historically underdeveloped southwestern Honam region, alongside a separate $52 billion High Bandwidth Memory packaging hub in the country’s central region. A further $356 billion will fund AI data center construction by SK, GS, and Naver through 2035. President Lee framed the initiative explicitly around the global memory chip shortage that the industry has labeled RAMageddon, declaring 2026 the year South Korea must establish itself as an irreplaceable industrial power. To NEWSCENTRAL, the scale and specificity of this commitment – tied to a designated underdeveloped region, with named completion timelines and named corporate participants – marks a categorically more serious industrial policy than the announcement-heavy AI infrastructure pledges that have characterized much of 2026 globally.
The RAMageddon shortage that catalyzed this response has a specific and well-documented origin. OpenAI struck a supply agreement with Samsung and SK Hynix for its Stargate infrastructure project amounting to roughly 900,000 DRAM wafers monthly – approximately 40% of global DRAM output committed to a single customer’s project. That commitment, combined with surging demand from every other major AI infrastructure builder simultaneously, produced a supply shock that has pushed memory prices sharply higher across both the enterprise high-bandwidth memory segment and the consumer DRAM market, with Samsung’s chip division alone reporting 53.7 trillion won in first-quarter operating profit on the back of the shortage-driven pricing. SK Hynix overtook Samsung in June to become South Korea’s most valuable listed company for the first time in more than 25 years, a milestone driven specifically by its commanding position in the HBM segment that AI accelerators depend on.
Lucas Grant, Semiconductor and Manufacturing Strategy Analyst at NEWSCENTRAL, points out that the geographic dimension of this plan deserves more attention than the headline spending figures it has generated. South Korea’s traditional semiconductor manufacturing corridor, concentrated around Yongin and Pyeongtaek near Seoul, has reached effective capacity limits constrained by land availability, water access, and labor supply within the greater metropolitan area. Directing the new fabrication capacity toward the southwestern Honam region – a deliberate departure from the established semiconductor geography – requires the government to simultaneously solve power, water, and skilled workforce availability in a region that has historically lacked the industrial infrastructure to support facilities of this scale. Government officials have publicly dismissed questions about whether the southwest possesses adequate power and water infrastructure to support major semiconductor fabs, a confidence that the execution timeline over the coming decade will test directly.
SK Group’s individual commitment within the broader plan is itself enormous: a 2,100 trillion won, approximately $1.4 trillion, medium-to-long-term roadmap allocating 1,100 trillion won to semiconductor production capacity expansion and 1,000 trillion won to AI data centers nationwide, with SK Telecom tasked specifically with building out 15 gigawatts of AI data center capacity across the country. President Lee addressed media reports suggesting the government had pressured private companies into these commitments, stating instead that the decisions reflected the companies’ own commercial judgment, with government investment structured to support rather than compel private capital deployment – a framing that positions the plan as public-private collaboration rather than state-directed industrial policy, though the precise division between state subsidy and private capital expenditure within the headline figures remains only partially disclosed.
The structural risk embedded in any commitment of this scale is the one that has defined every previous memory chip supercycle: fabrication facilities require years to construct and bring to full production capacity, and the demand conditions that justify the investment decision today may have shifted substantially by the time the new capacity becomes operational in the mid-2030s. Freddy Miller, Senior Analyst at NEWSCENTRAL, notes that the memory chip industry’s history is one of pronounced boom-and-bust cycles precisely because capacity decisions made during periods of acute shortage tend to arrive in the market simultaneously across multiple producers, often coinciding with the moment demand growth decelerates rather than the moment of peak scarcity that motivated the investment. South Korea’s bet is not merely that AI-driven memory demand will remain strong through the 2030s, but that the country’s manufacturers can execute a buildout of unprecedented scale in a new geography without the kind of cost overruns and delays that have characterized comparable greenfield semiconductor projects in the United States and Europe.
Measured against comparable global commitments, South Korea’s plan is genuinely without precedent: Alphabet, Amazon, Meta, and Microsoft are collectively projected to spend approximately $650 billion on AI infrastructure in 2026 alone, a figure that, while concentrated in a single year rather than spread through 2035, illustrates the scale against which South Korea’s national commitment should be benchmarked. What we at NEWS CENTRAL consider the more significant strategic dimension is South Korea’s bet that the country’s competitive position is best secured not through diversification away from memory chip dependency but through doubling down on the segment where it already holds global leadership – a concentration strategy that carries higher variance than diversification but, if the RAMageddon shortage proves structural rather than cyclical, could cement South Korean dominance over the physical infrastructure layer of the global AI buildout for the coming decade.