Naver has become the first Korean company to join NVIDIA’s open-source Nemotron consortium, gaining access to open AI model development and fine-tuning capabilities to strengthen its HyperCLOVA X sovereign AI platform.
Naver has joined NVIDIA’s artificial intelligence ecosystem as the first Korean company to become a member of the NVIDIA Nemotron open-source large language model (LLM) consortium, marking a significant step in its sovereign AI ambitions.
Under the partnership, Naver and NVIDIA will collaborate on the development and utilisation of open AI models, while jointly researching optimisation and foundation technologies for large language models. Naver plans to fine-tune Nemotron3 Ultra using Korean-language data and operational expertise to enhance its HyperCLOVA X AI platform.
Unlike proprietary models developed by OpenAI and Anthropic, Nemotron is an open model that can be customised and adapted using locally sourced data, enabling Naver to build AI systems tailored to Korean requirements.
The collaboration also extends to infrastructure. NVIDIA announced that Naver will build a gigawatt-scale AI Factory based on NVIDIA DSX, an integrated AI Factory platform that combines modular open-source software, APIs, AI computing infrastructure, design frameworks and a partner ecosystem. The partnership spans infrastructure, AI models, physical AI, data centres and agentic AI.
Naver plans to launch a 55MW AI Factory in 2027, expand overseas capacity to 200MW by 2028 and ultimately develop a gigawatt-scale facility.
“We are participating as a global core partner where both Naver and NVIDIA will jointly bear the business outcomes and risks,” Naver said in a company statement.
Beyond sovereign AI, Naver plans to enhance its Seoul World Model digital twin using NVIDIA Cosmos and establish a Korean AI agent platform based on NVIDIA NemoClaw.
