South Korea is committing $1 trillion to a coordinated push across memory chip fabrication, AI data centers, and humanoid robots, with President Lee Jae Myung framing the three together as the country's strategic axis. Samsung and SK Hynix will lead a $585 billion fab buildout, SK Group, GS Group, and Naver will spend $357 billion on AI data centers, and Hyundai Motor Company is putting $5.8 billion into a robot manufacturing facility and AI data center. The announcement, delivered in a televised speech on June 29, 2026, ties the country's industrial policy directly to the AI boom that has driven record profits at its two largest chipmakers.
The plan is explicit about pace. The government's goal is to double South Korea's DRAM production within five years, a target that would reshape global memory supply at a moment when AI buildouts have created shortages and pushed up prices on consumer hardware including Apple's Macs and Valve's Steam Machine. The Hyundai-Boston Dynamics line aims to produce 30,000 Atlas humanoid robots per year by 2028, with commercial deployment targeted across 10 major industries.
The chip portion is the heaviest lift. Samsung and SK Hynix's $585 billion will fund new fabs in South Korea's southwestern provinces alongside expanded construction in the Seoul capital region. The southwestern plants alone will need 6.3 gigawatts of electricity and 650,000 tons of water, with another 8 gigawatts earmarked for the new AI data centers, according to South Korea's Ministry of Climate, Energy and Environment.
Key facts
- 01South Korea's government and top tech firms are committing $1 trillion across memory fabs, AI data centers, and humanoid robots.
- 02Samsung and SK Hynix will spend $585 billion to double DRAM output within five years.
- 03SK Group, GS Group, and Naver will put $357 billion into new AI data centers across four provinces.
- 04Hyundai is investing $5.8 billion to build a robot factory and AI data center, scaling Boston Dynamics' Atlas to 30,000 units per year by 2028.
- 05New southwestern chip plants will require 6.3 gigawatts of electricity and 650,000 tons of water, plus 8 gigawatts more for the data centers.
Power mix is the open question. Nuclear and coal each accounted for more than 30 percent of South Korean electricity generation in 2024, and natural gas covered nearly 25 percent — a dependency that has left the country exposed during the ongoing Strait of Hormuz crisis. Officials said renewables and nuclear would help cover the new industrial load, alongside continued fossil fuel use.
Timing is the second open question. SK Hynix Chairman Chey Tae-won noted that the company's existing Yongjin cluster in the Seoul metropolitan area took nine years to build out. If new southwestern fabs run on a similar timeline, the relief for global memory pricing arrives well after the current AI capital cycle, not during it.
“it took nine years for the company to build a cluster of chip manufacturing facilities in Yongjin within the Seoul metropolitan area”— Chey Tae-won, SK Hynix Chairman
The data center leg of the plan spreads across South Chungcheong Province in the west, Gangwon Province in the east, and the North and South Jeolla Provinces in the southwest. The $357 billion commitment from SK Group, GS Group, and Naver is designed to keep Korean-trained models and Korean-served inference inside the country rather than routed through US hyperscalers, a sovereignty argument that has shaped similar buildouts in Austria and the Gulf.
On the robotics side, the government has designated physical AI as a national strategic industry and is targeting a Korean general-purpose foundation model based on a world model within three years. Hyundai's $5.8 billion Saemangeum facility in North Jeolla Province will anchor Atlas production, using the South Korean supply chain that the automaker has been building out since acquiring Boston Dynamics in 2021. The plan also calls for training 10,000 workers as AI robotics specialists over the next five years.
Labor is not on board. On June 25, Hyundai Motor's union overwhelmingly approved a potential strike as it negotiates profit-sharing and job protections tied to the Atlas deployment, and a state mediation committee granted the union the legal right to strike after suspending arbitration. The dispute is the clearest signal yet that mass humanoid deployment inside one of the world's largest automakers will be fought factory by factory, not waved through.
Political pressure on chipmaker profits is also building. South Korean officials have urged Samsung and SK Hynix to share AI-era gains with workers and smaller suppliers. In May, the presidential chief of staff for policy floated a national dividend funded by excess tax revenue from AI-driven corporate profits, an idea the government later walked back as a personal view rather than policy.
The skeptics' case is straightforward. Nine-year fab timelines, 14.3 gigawatts of new electrical demand split between fabs and data centers, a strike-authorized union at the country's largest automaker, and a political environment debating how to redistribute chipmaker profits all sit between the $1 trillion headline and the 2028 targets. The plan assumes AI demand stays high enough through the build cycle to justify doubled DRAM capacity — a bet that looks safe today but has no margin for an AI capex pause.
What South Korea is doing here is treating memory, compute, and embodied AI as one industrial stack and funding all three at once. For Samsung and SK Hynix, this locks in another decade of memory leadership against Micron and China's CXMT. For Hyundai and Boston Dynamics, it underwrites the manufacturing scale needed to actually price Atlas competitively against Figure and Tesla's Optimus. And for the global AI buildout, it concentrates a meaningful share of next-decade DRAM supply inside a single country — useful when the fabs come online, fragile if any one of the political, labor, or power constraints slips.
Working on something we should cover, or seeing a story we missed? Send leads, documents, or feedback to hello@aichatdaily.com. For sensitive tips, see our secure tips page for Signal and PGP options.
Spotted an error? Email hello@aichatdaily.com with the URL and the issue, or read our full corrections policy.
