South Korea’s government and its two largest chipmakers have announced what may be the most consequential national AI investment bet of 2026. At a presidential briefing on June 29, President Lee Jae Myung unveiled a multi-program national plan backed by Samsung Group and SK Hynix, with combined spending commitments approaching $1.3 trillion over the next decade.

The figures are staggering, but the logic behind them is straightforward. South Korea makes the memory chips that every AI data center runs on. As AI infrastructure spending accelerates globally, the country is moving fast to lock in its position as the indispensable supplier.

What Was Announced

Samsung Group pledged 1,000 trillion won, roughly $648 billion, over ten years, making it the largest single corporate investment commitment in South Korean history. The figure spans Samsung Electronics’ semiconductor factories and AI data centers, Samsung Display’s manufacturing operations, Samsung SDI’s battery expansion, and Samsung Electro-Mechanics’ advanced packaging production.

The core of the national program is a semiconductor and AI cluster that Samsung Electronics and SK Hynix will jointly anchor. Under this initiative, each company will build two new fabrication plants in South Korea’s southwest, as part of an 800 trillion won national semiconductor ecosystem project covering the Honam, Chungcheong, Yeongnam, and Incheon regions.

Of Samsung Group’s total pledge, more than 350 trillion won is earmarked for AI infrastructure, primarily data centers. The Yongin semiconductor site, already one of the world’s largest chip campuses, will receive 60 trillion won in new investment. This is not a consolidation play. Samsung is expanding geographically across the country while deepening its existing sites.

Why This Matters for AI Infrastructure

These two companies are not peripheral players in the AI supply chain. Samsung and SK Hynix together supply roughly 80 percent of the global high-bandwidth memory market. HBM is the specialised memory architecture that feeds data to AI training accelerators. Every major AI training cluster, from hyperscaler data centers to sovereign AI projects, depends on it.

South Korean semiconductor exports hit $37.2 billion in May 2026, up 169 percent year on year, setting an all-time monthly record. That number reflects the structural demand created by the global AI buildout. The domestic investment announced today is the government and its chipmakers responding to that demand signal with a decade-long supply commitment.

For businesses building or buying AI infrastructure, this investment means HBM availability and pricing will improve over time as new capacity comes online. The current constraints on AI training runs, which are often memory-bound rather than compute-bound, will ease as Samsung and SK Hynix scale up their new plants.

The Physical AI Angle

The investment is not limited to semiconductors and data centers. President Lee’s announcement explicitly includes physical AI, encompassing robotics, industrial automation, and AI-enabled manufacturing equipment.

This reflects a broader trend. The same wave of AI investment that is reshaping software and cloud infrastructure is beginning to reach physical operations. Samsung, with its manufacturing footprint across consumer electronics, displays, and batteries, is positioned to apply AI agents and automation at a scale that few companies in the world can match.

The combination of in-house AI investment and world-class manufacturing creates a competitive loop: Samsung builds better chips, uses those chips to run AI systems, uses those AI systems to improve manufacturing processes, and repeats. It is a compounding advantage that is difficult for competitors without integrated chip and manufacturing capability to replicate.

What This Means for Business

Most business leaders will not be directly affected by semiconductor fab construction in South Korea’s southwestern provinces. But the downstream effects of this investment are relevant to any organisation building an AI strategy.

AI infrastructure costs will continue to fall. Increased chip supply, improved yields from new fabs, and competitive pressure between Samsung and SK Hynix will put downward pressure on the cost of AI inference over the next three to five years. Workloads that are currently expensive to run at scale will become cheaper.

Sovereign AI is becoming a real category. South Korea is not the only government treating AI infrastructure as a national priority. The US Stargate project, European AI factory investments, and now this Korean commitment all point toward a world where AI compute capacity is treated as critical national infrastructure. Businesses operating across borders will increasingly see AI policy as part of their operating environment.

The decade-long bet signals confidence. Samsung Group’s 1,000 trillion won commitment runs through 2036. That is not a speculative punt. It reflects genuine conviction that AI infrastructure demand will sustain itself for a decade. When the company that supplies the memory chips to the global AI industry makes that call, it carries weight.

The gap between AI leaders and laggards is about to widen. Every trillion-dollar infrastructure commitment of this kind accelerates the pace at which capable AI systems become available and affordable. Businesses that have not yet built the internal capability to use those systems will find themselves further behind with each passing year, not because the tools are hard to access, but because their competitors will have built the skills and processes to use them first.

If you want the playbook other teams are using with Claude and Codex right now, grab the free Working With Claude field guide. Download it here.