Naver's Lee Hae-jin bets big on AI with defense push, Nvidia meeting

Naver Chairman Lee Hae-jin is making a bold AI bet. The company has designated defense AI as a core next-generation business and set up a dedicated organization to pursue the security market. Amid that push, Lee is set to meet Nvidia CEO Jensen Huang on Thursday.

The two companies have steadily expanded their collaboration around sovereign AI. Because Nvidia's GPUs and AI infrastructure can also serve as the backbone of defense AI systems, the industry is watching closely to see how far Thursday's meeting will go in fleshing out cooperation on defense AI, physical AI and AI infrastructure.

Naver Cloud said Sunday it is launching a task force dedicated to defense AI transformation, or defense AX.

It is the first time Naver has formed an AI organization focused exclusively on defense. Naver Cloud CEO Kim Yu-won will personally lead the defense AX task force.

A defining feature of the task force is its front-line deployment of Field Deployment Engineers, or FDEs — specialists who design and implement customized AI solutions directly at defense sites. FDEs support the construction and operation of AI systems on-site with customers, an approach also associated with how U.S. firm Palantir expanded its work with defense and intelligence agencies. The task force will operate as a cross-functional unit, with teams covering AI model development and business development working in close coordination.

Naver plans to deliver AI solutions tailored to the defense and security environment by drawing on its proprietary foundation models, cloud infrastructure and sovereign AI capabilities. The company is also pursuing ways to apply its omnimodal AI technology — which can combine and process text, images, audio and other data types for inference and generation — to the defense sector.

Sung Nak-ho, Naver Cloud's head of hyperscale AI technology, previewed the direction at a public-sector AX strategy seminar held in Sejong in April. "Naver Cloud is the only domestic company with omnimodal AI capabilities that can combine and process diverse data types — text, images and audio — for inference and generation," he said. "We will concentrate our capabilities on analyzing complex battlefield situations and supporting the military's strategic decision-making."

Industry observers see the task force as a strategic move by Naver to secure technological self-reliance and data sovereignty in the defense sector. Because defense work involves sensitive data including military secrets, the ability to build and operate AI services and infrastructure independently — rather than relying on foreign AI — is critical.

Lee himself is lending direct support to the effort. On Thursday, he will meet Jensen Huang after the Nvidia CEO wraps up the company's annual AI event in Taiwan and travels to South Korea. The industry is watching whether the meeting will serve as a catalyst for Naver to sharpen its strategy across defense AI, physical AI and AI infrastructure.

Lee has already met Huang multiple times in South Korea, Taiwan and elsewhere over the past several years to discuss AI cooperation. Those talks led to an agreement in October last year to jointly develop a physical AI platform. Last month, Madison Huang, Jensen Huang's eldest daughter and a senior director at Nvidia, visited Naver's 1784 headquarters to hold in-depth discussions on technology and business collaboration.

During what the industry dubbed the first "Ganbu meeting" — Huang's visit to South Korea in October last year — Naver secured 60,000 GPU units, 10,000 more than the 50,000 each allocated to Samsung Electronics, SK Corp and Hyundai Motor Group.

Naver holds its own large language model HyperCLOVA X, along with cloud infrastructure, digital twin and robotics technology. Analysts say combining these assets with Nvidia's GPUs and industrial AI platform could extend the scope of their partnership well beyond sovereign AI into physical AI and defense AI.

"As AI use grows in recent conflicts, the technology is becoming increasingly important in the defense sector," a Naver Cloud official said. "Naver's omnimodal AI can process multiple types of data simultaneously — text, video and more — making it applicable to battlefield information analysis and decision-making support."

The official added that Naver's ownership of its own data centers and cloud infrastructure gives it a competitive edge on data sovereignty. "We will be able to leverage these capabilities to build AI services and infrastructure that account for the unique requirements of the defense sector," the official said.