South Korea Bets $80B on AI Chips and Data Centers

South Korea has committed approximately $80 billion in public funds to anchor a $576 billion semiconductor and AI infrastructure drive, positioning Samsung and SK Hynix at the center of the global AI supply chain.

Samsung Electronics and SK Hynix will invest 800 trillion won ($518 billion) with suppliers to build four new chip fabs across southwestern South Korea.
South Korea targets 18.4 gigawatts of AI data center capacity by 2035, backed by more than $648 billion in combined public-private investment.
SK Hynix supplies roughly 70% of Nvidia's HBM4 memory, anchoring South Korea's critical position in the global AI chip stack.

Lead

South Korean President Lee Jae-myung on June 29, 2026 unveiled the country's most ambitious industrial initiative, framing it as a "great leap forward" centered on what he called the "triple axis" of semiconductors, physical AI, and AI data centers. Seoul committed approximately $80 billion in direct public investment and policy financing to underwrite the strategy, designed to expand South Korea's share of the global AI supply chain while de-risking semiconductor geography. Combined with private-sector pledges from Samsung, SK Hynix, and allied conglomerates, total committed capital exceeds $576 billion — the largest technology infrastructure bet in the nation's history.

What Happened

Samsung Group Chairman Jay Y. Lee and SK Hynix Chairman Chey Tae-won both appeared at the June 29 launch to commit their companies to the initiative. Samsung announced it had selected Gwangju as the site of its new chip cluster; SK Hynix indicated it needed additional time to finalize a southwestern location. The event formalized a decisive shift in South Korea's industrial policy toward AI-era infrastructure, moving investment capital beyond the established Seoul metropolitan corridor into historically underinvested southwestern regions.

The government's framework channels approximately 30 trillion won ($20 billion) in direct investment and policy financing in 2026 alone, across six priority sectors: AI, robotics, future vehicles, defense, semiconductors, and secondary batteries. Cumulative public capital commitments are expected to exceed 120 trillion won (~$80 billion) through the mid-2030s. Regional governments in Gwangju, South Jeolla, and Chungcheong provinces will contribute an additional 5 to 20 trillion won each to anchor local fab ecosystems.

The Investment Architecture

Samsung Electronics committed 425 trillion won for two new memory fabs and a national AI computing center in the Honam region, while Samsung Group as a whole pledged 1,000 trillion won (~$649 billion) across a decade spanning semiconductors, AI data centers, and robotics. SK Group pledged 470 trillion won for two fabs and a 1-gigawatt AI data center.

On the data center side, SK Telecom, GS Group, and Naver anchored the buildout with a combined 550 trillion won ($357 billion) commitment targeting 8.4 gigawatts of capacity by 2029. A second phase running through 2035 adds 10 more gigawatts, bringing total planned AI data center capacity to 18.4 gigawatts and cumulative investment to more than 1,000 trillion won ($648 billion). An 81 trillion won advanced chip-packaging cluster will be constructed in the Chungcheong area near Seoul, integrating back-end packaging with the new fab network.

AI and Technology Angle

South Korea's dominant position in high-bandwidth memory (HBM) gives it an asymmetric stake in the AI chip economy. HBM chips are essential components of the AI accelerators produced by Nvidia and AMD, and both Samsung and SK Hynix are competing intensively for next-generation supply contracts.

SK Hynix, which currently supplies approximately 70% of Nvidia's HBM4 memory for the Vera Rubin platform, has seen its shares surge 288% in 2026 as AI infrastructure demand accelerated. Samsung is closing the technology gap, having completed HBM4 qualification tests and commenced shipments to major AI customers. Both companies are scaling dedicated HBM packaging fabs, with Samsung targeting a 50% increase in overall production capacity this year and SK Hynix quadrupling its infrastructure investment from previously announced levels.

South Korea's Korea tech policy also targets the development of homegrown neural processing units (NPUs), custom server hardware, and advanced cooling systems — with an ambition to export complete turnkey AI data center architectures globally once the domestic stack is commercially validated. The Ministry of Trade, Industry and Energy has set a target to double DRAM output within five years.

Strategic Context

The initiative reflects intensifying competition over semiconductor geography. The United States has redirected chip investment domestically through the CHIPS Act; Taiwan remains the global center for advanced logic through TSMC; and Japan has attracted TSMC to its Kumamoto facility. South Korea's response emphasizes memory and advanced packaging — segments where it retains established technical leadership — while constructing AI compute infrastructure to anchor demand for domestically produced chips.

The plan carries a political dimension. Opposition lawmakers have criticized President Lee's decision to locate new chip clusters in the Honam region, traditionally an electoral stronghold of his liberal Democratic Party, arguing industrial logic was subordinated to regional politics. Lee has dismissed the criticism, describing the megaprojects as multi-decade investments in economic diversification rather than a political exercise.

South Korea also positions itself as a trusted supplier for American and European AI infrastructure builders, particularly as export controls on advanced semiconductors have intensified scrutiny of Chinese-linked supply chains, opening a strategic lane for Seoul.

Outlook

South Korea's semiconductor and AI data center megaprojects represent the country's most concentrated industrial wager since it built its electronics manufacturing base in the 1980s. Four new fabs must reach volume production, 18.4 gigawatts of data center capacity must be commissioned, and homegrown NPU ecosystems must mature to global-export standards — all within a decade. Execution at this scale has no domestic precedent. Whether the commitment translates into durable market position will determine whether South Korea consolidates its lead in AI infrastructure or cedes ground to faster-moving rivals in the next critical phase of the global AI buildout.

Mentioned tickers: 005930.KS, 000660.KS, 017670.KS, 035420.KS, NVDA, AMD, AMZN