On July 25, 2026, NAVER, NVIDIA, and Brookfield Asset Management announced in San Francisco the expansion of South Korea's sovereign AI factory: the initial 55-megawatt NVIDIA DSX deployment at the GAK Sejong data center in Sejong will grow to 200 megawatts by 2028 — more than tripling the original capacity announced just six weeks prior. The total funding envelope is $10 billion.
The announcement was made during South Korean President Lee Jae Myung's visit to the AI Summit in San Francisco, where he met with the CEOs of NVIDIA, OpenAI, and Anthropic. The geopolitics are explicit in the timing.
The structure of the deal
NVIDIA: a $1 billion equity investment in NAVER. Brookfield: a commitment of up to $9 billion — but in a non-binding letter of intent. The distinction matters to the market: the big number is not yet committed capital.
The infrastructure will use the NVIDIA Vera Rubin and Blackwell platforms, establishing a dedicated pool of resources for emerging AI companies, providing compute, software, and support to develop and deploy competitive models and applications at scale.
NAVER's long-term goal: 1 gigawatt of sovereign AI infrastructure — which would place South Korea among the few countries with national computing capacity at this scale.
Why this matters beyond Korea
"AI factories are the infrastructure nations need to compete and innovate in the age of intelligence," said Jensen Huang, CEO of NVIDIA. "This is how nations and companies are building their futures in the AI era."
The pattern repeats: national AI champion + full-stack NVIDIA infrastructure + regionally branded model. The same structure appeared in France (SoftBank/Sesterce), in the UK, and now in South Korea. NVIDIA is building the operating system of national AI sovereignty — and each deal strengthens the lock-in of the DSX platform as a global standard.
NAVER is also developing the "Seoul World Model" under this initiative — an AI stack focused on the Korean language, hosted in Korea, that bypasses American hyperscaler infrastructure for inference or training. It is the operational definition of AI sovereignty.
What the Brookfield data reveals
Brookfield's non-binding $9 billion commitment signals strategic interest, not guaranteed capital. For the AI infrastructure market, this is relevant: the actual committed number today is $1 billion from NVIDIA. The rest depends on execution, regulation, and demonstrated demand.

