South Korean cloud company Naver Cloud has built a cluster of 4,000 Nvidia B200 GPUs.

The company revealed today, January 8, that it had completed the cluster, calling it the largest of its kind in South Korea.

Naver said in its announcement that the cluster uses cooling, power, and network optimization technologies based on its previous experience with deploying an Nvidia SuperPod.

Internal testing suggests that with the new cluster, Naver will be able to train a 72 billion-parameter model in 1.5 months as opposed to the 18 months it would take with the company's current cluster of 2,048 A100 GPUs.

Naver CEO, Choi Soo-yeon, said: “This AI infrastructure construction is meaningful in that it goes beyond a simple technology investment, and it secures a core asset that supports national AI competitiveness and AI independence and sovereignty.”

She added: “Team Naver will create real value by more flexibly applying AI technology to services and industrial sites based on an infrastructure that allows for rapid learning and repeated experiments.”

The company has not stated which data center the cluster is housed in, though it is one of the company's South Korea locations.

In recent months, Naver has revealed plans to expand its data center in Sejong, South Korea, aiming to reach 270MW.

Naver Cloud currently lists six cloud regions in South Korea, as well as one each in Japan, Germany, Singapore, and the US (West Coast); future regions are planned in the US (East Coast), Vietnam, Taiwan, and Thailand. It previously exited Hong Kong. The company also lists plans for a 10MW site in Charlotte, North Carolina, and a capacity totaling 40MW across Sweden.

In October 2025, it was revealed that Naver would deploy some 60,000 Nvidia GPUs in partnership with LG AI Research, SK Telecom, NC AI, and Upstage.