NAVER and NVIDIA announced in July 2026 a landmark partnership to build gigawatt-scale sovereign AI infrastructure anchored in South Korea and stretching across Asia, the Middle East, and Europe. At the heart of the deal: NAVER's next-generation GAK Sejong data center in Sejong City will be outfitted with NVIDIA's DSX™ (Data Center-as-a-Service) platform, turning it into a full-stack AI factory capable of handling the world's most demanding AI workloads while keeping data within local regulatory boundaries.
What Is NVIDIA DSX?
NVIDIA DSX is not just a chip deal — it is a complete end-to-end platform covering accelerated computing hardware, high-speed networking, AI software stacks, and foundational models. Designed to let organizations rapidly design, build, and scale large AI factories, DSX has already been deployed across Europe and the Middle East with telecom carriers and national AI enterprises. NAVER's adoption of DSX at GAK Sejong brings high-density accelerated computing, energy-efficient cooling, advanced automation, and robust disaster-response capabilities under one integrated platform.
Phase 2 Target Late 2027 — 100 MW
Phase 3 Target 2028 — 200 MW
Long-Term Goal Gigawatt-scale (domestic + overseas)
Platform NVIDIA DSX™ Full-Stack
AI Models Next-gen HyperCLOVA X · Seoul World Model
Markets South Korea, Europe, Middle East
HyperCLOVA X and the Seoul World Model
This partnership directly accelerates NAVER's flagship AI model development. For the next generation of HyperCLOVA X, NAVER is fine-tuning NVIDIA's Nemotron 3 Ultra open model with its proprietary Korean-language data and training expertise — creating a more capable, culturally fluent model for Korean and global enterprise customers. The Seoul World Model and NAVER's agentic AI services will also be built and trained on this new infrastructure, pushing both products to new capability levels.
A Multi-Region Alliance, Not Just One Data Center
The scope of this deal extends far beyond Sejong. NAVER already operates sovereign AI cloud services in Europe and the Middle East, and the combination of its hyperscale data center capabilities with NVIDIA DSX gives it the infrastructure muscle to serve local data-sovereignty regulations in those markets at greater scale. NVIDIA CEO Jensen Huang called the partnership "a milestone in building global sovereign AI infrastructure anchored in Korea."
| Phase | Timeline | Capacity | Scope |
|---|---|---|---|
| 1 | Early 2027 | 55 MW | GAK Sejong DSX initial launch |
| 2 | Late 2027 | 100 MW | Domestic expansion + overseas demand |
| 3 | 2028 | 200 MW | Accelerate Europe/Middle East sovereign cloud |
| Long-term | TBD | Gigawatt-scale | Global AI factory alliance complete |
Why This Matters
This is bigger than a data center expansion. NAVER — Korea's largest internet company — is joining forces with the world's dominant AI hardware and platform provider to offer government agencies and enterprises a credible, Asia-based sovereign AI cloud that does not route through US or Chinese hyperscalers. For Korean public institutions focused on AI sovereignty, and for enterprises across Asia, Europe, and the Middle East seeking compliant alternatives, this partnership opens a new path: world-class AI compute power with local data control.
The roadmap from 55MW to gigawatt-scale also signals that NAVER and NVIDIA see sovereign AI demand not as niche but as mainstream — and are betting that the infrastructure to serve it will be one of the defining infrastructure races of the late 2020s.
- NAVER and NVIDIA officially announced a sovereign AI infrastructure alliance in July 2026
- GAK Sejong data center expands using NVIDIA DSX: 55MW (early 2027) → 100MW → 200MW (2028) → gigawatt-scale
- Next-gen HyperCLOVA X will be trained by fine-tuning NVIDIA Nemotron 3 Ultra on NAVER's proprietary Korean-language data
- Partnership extends across Europe and the Middle East for sovereign AI cloud services
- Part of a broader Korea–NVIDIA ecosystem push including SK hynix memory co-development and government AI agreements