Brijesh Desai 
NVIDIA partners with SK Hynix, Naver, and Doosan for South Korea’s AI infrastructure. Learn about gigawatt-scale AI factories, physical AI, and memory partnerships totaling tens of billions.
NVIDIA SK Hynix Naver Doosan AI partnership just became South Korea’s biggest AI infrastructure deal, announced Monday during CEO Jensen Huang’s visit to Seoul. The chip giant struck multi-year partnerships with all three companies for AI data centers, but here’s the kicker: global chip stocks sold off sharply on the news the same day, even as NVIDIA locked in deals worth tens of billions of dollars.
What NVIDIA Actually Signed With Each Partner
Let’s break down who got what:
This isn’t just “we’ll buy your chips.” It’s deep technology integration across memory, infrastructure, and robotics.
SK Hynix: NVIDIA’s Memory Backbone
SK Hynix is officially NVIDIA’s largest memory partner under a deal spanning more than two years with extension options. Jensen Huang said NVIDIA already buys billions of dollars from SK Hynix every year, and that volume will grow substantially.
The partnership covers:
- Advanced memory for AI data centers
- Memory for NVIDIA’s future platforms: Vera Rubin supercomputer, Vera CPU, RTX Spark PC, Jetson Thor robotics
- Chip design acceleration using NVIDIA’s CUDA-X libraries and Omniverse tools
- SK Hynix building “digital twins” of its fabrication plants toward fully autonomous operations
SK Group Chairman Chey Tae-won attended the announcement event in person, showing how serious this is.
Naver: Building a 1-Gigawatt AI Factory (That’s Insane)
Naver’s deal is the most ambitious. Construction begins in 2027 with a 55-megawatt phase, scaling to 200 MW by 2028, with an ultimate goal of 1 gigawatt.
To put that in perspective:
- 1 gigawatt = ~4x the capacity of “Gak Sejong,” Naver’s largest domestic data center
- Could house hundreds of thousands of NVIDIA’s latest GPUs
- Cost:Â tens of billions of dollars
The facilities run on NVIDIA’s DSX platform, fused with Naver’s homegrown GPU cluster expertise to reduce token costs. This is “sovereign AI”—Naver building its own AI infrastructure independent of U.S. cloud providers.
Doosan: The “Physical AI” Pivot
Here’s where NVIDIA’s strategy gets interesting. Doosan’s partnership focuses on physical AI and robotics across Doosan Robotics, Doosan Bobcat, Doosan Enerbility, and Doosan Corporation Electro-Materials.
This marks NVIDIA’s shift from pure cloud toward physical artificial intelligence. The partnership includes:
- GPUs for robotics
- Jetson edge AIÂ platforms
- Omniverse for digital twin simulations
The kicker: Doosan produces materials used in NVIDIA’s Blackwell chips. This is a supply chain relationship, not just a customer one.
SK Telecom: The Hidden Gigawatt Partner
One more player: SK Telecom (part of SK Group) will build a gigawatt-scale AI cloud in Korea using NVIDIA’s DSX platform. The first data center becomes operational in 2027.
So SK Group has two AI arms: SK Hynix (memory) and SK Telecom (infrastructure).
Why “Physical AI” Matters Right Now
Jensen Huang has been pushing “physical AI” hard—AI that interacts with the real world through robotics, not just chat. This NVIDIA-Doosan partnership is a concrete example: robotics, autonomous equipment, and energy systems powered by edge AI.
Physical AI includes:
- Humanoid robots in research labs
- Autonomous vehicles and equipment
- Industrial robotics for manufacturing
- Smart energy systems
SK Hynix is supplying memory for Jetson Thor robotics platform, which is NVIDIA’s flagship for physical AI.
The Market Reaction: Chip Stocks Sold Off
Despite the billion-dollar deals, global chip stocks tumbled sharply on Monday, June 8, 2026. Investors might be worried about:
- Supply chain risks in South Korea
- Capital expenditure requirements for gigawatt-scale builds
- NVIDIA competition from other chipmakers
- Economic slowdown concerns affecting AI demand
Either way, the market didn’t celebrate. Yet.
What This Means for South Korea’s AI Position
South Korea is now a critical AI infrastructure hub alongside the U.S. and China. The SK-Naver-Doosan-NVIDIA ecosystem covers:
- Memory manufacturing (SK Hynix)
- AI cloud infrastructure (SK Telecom, Naver)
- Robotics & physical AIÂ (Doosan)
- GPU compute (NVIDIA)
This is a full-stack AI economy, not just a customer relationship.
NVIDIA’s Global Strategy: Beyond U.S. Data Centers
These deals show NVIDIA’s strategy to:
- Diversify geographically beyond U.S.-only infrastructure
- Build sovereign AIÂ in key markets (South Korea, Japan, Europe)
- Integrate vertically with memory suppliers and robotics manufacturers
- Lock in long-term partnerships spanning multiple years
The Naver deal spans Asia, Middle East, and Europe, creating a global AI factory network.
Quick Summary
NVIDIA locked multi-year partnerships with SK Hynix (memory for AI), Naver (gigawatt-scale AI factory starting 2027, 55MW→200MW→1GW), and Doosan (physical AI & robotics). SK Hynix is NVIDIA’s largest memory partner supplying Vera Rubin, RTX Spark, Jetson Thor. Naver joins the Nemotron Coalition building sovereign AI on DSX platform. Doosan produces Blackwell chip materials and powers robotics with Jetson edge AI. Total investment: tens of billions. Ironically, global chip stocks sold off on the news. This marks NVIDIA’s shift from cloud-only to physical AI, with South Korea becoming a critical AI hub covering memory, infrastructure, and robotics.