Naver Secures $270M Loan from South Korean Government for AI Data Center Expansion in Sejong
April 24, 2026
Naver Secures $270M Loan from South Korean Government for AI Data Center Expansion in Sejong
South Korean AI cloud giant Naver has secured a 400 billion won ($270 million) loan from the country’s Public Growth Fund to support the expansion of its flagship AI data center in Sejong, marking a significant step in the nation’s push to reduce reliance on global big tech for artificial intelligence infrastructure. The loan, approved by the South Korean Financial Services Commission (FSC), will fund the development of the second and third phases of the facility, known as Gak Sejong, as reported by local media Chosun Biz. The government-backed financing underscores Seoul’s strategic focus on building domestic AI sovereignty amid intensifying global competition for advanced computing capacity.
Naver’s Gak Sejong campus, spanning 290,000 square meters (3.1 million square feet), is designed to ultimately reach 270 megawatts of power capacity and host up to 600,000 servers running Naver’s cloud services. The first phase of construction began in November 2023, and the company revealed plans in November 2025 to accelerate development by building the second and third phases simultaneously. The facility, which spans five floors—three below ground and two above—is engineered to withstand an earthquake of magnitude 7.0. Naver aims to complete Phase II by 2028 and Phase III by 2029, with a total of six phases planned for the site.
The FSC’s decision to fund the expansion was driven by the need to enable Naver to advance its own large language models and deploy the latest GPUs on-site. “Korean corporations have had a high level of dependence on global big tech corporations in developing AI models due to a lack of AI infrastructure facilities,” the FSC stated, adding: “To help the domestic AI industry reduce its dependence on global big tech and secure its own AI sovereignty, we decided to provide funding for domestic corporations’ large-scale AI data center construction projects.” This move reflects a broader trend among governments worldwide to invest in sovereign AI capabilities as a matter of national competitiveness and security.
Earlier this year, Naver Cloud revealed it had completed a cluster of 4,000 Nvidia B200 GPUs, though it did not specify which data center housed the cluster. In October 2025, Naver also announced plans to deploy some 60,000 Nvidia GPUs in partnership with LG AI Research, SK Telecom, NC AI, and Upstage. Naver Cloud currently operates six cloud regions in South Korea, along with one each in Japan, Germany, Singapore, and the U.S. West Coast, with future regions planned for the U.S. East Coast, Vietnam, Taiwan, and Thailand. The company also has plans for a 10MW site in Charlotte, North Carolina, a total of 40MW across Sweden, and a 500MW data center campus in Morocco.
In a separate development, Finland-based AI cloud company Verda, formerly known as DataCrunch, has completed a $117 million funding round to accelerate its international expansion and AI cloud infrastructure. The round comprises equity funding led by Lifeline Ventures, with participation from byFounders, Tesi, Varma, and other investors, alongside debt financing from a group of Nordic financial institutions. Verda, founded in 2020, currently operates data centers in Finland running on 100 percent renewable energy, as well as one in Reykjanesbær, Iceland. The company secured $64.6 million in a Series A round in September 2025, during which it noted plans for an AI data center in Latvia, following a $13 million seed round in October 2024. “We’re building the next generation of AI cloud infrastructure for pioneering teams across the globe. This funding allows us to double down on development and accelerate our expansion across Europe, the US, and Asia. We’re seeing very strong momentum, and we’re excited for what’s ahead,” said Ruben Bryon, founder and CEO of Verda. The company is planning to launch in new markets, including the UK and the US, in 2026.
Source: datacenterdynamics
