Naver Cloud Joins NVIDIA Nemotron Coalition: Collaboration Toward Sovereign AI

Naver Cloud is newly joining the Nemotron Coalition led by NVIDIA. This coalition is a global collaborative body in which open model builders and AI developers share research achievements, data, and computing resources to advance open frontier models.

The core of this collaboration is the advancement of HyperCLOVA X. Naver Cloud plans to leverage NVIDIA's open LLM, Nemotron 3 Ultra, to develop its next-generation model, and both companies plan to jointly research optimization and foundational technologies for large language models.

Nemotron 3 Ultra, the starting point of this collaboration, is a model designed for autonomous execution agents, with the following characteristics:

• A Mixture-of-Experts (MoE) structure with 550 billion parameters
• Activating only about 55 billion experts per token, achieving an efficient balance between inference quality and cost

Naver Cloud is participating not merely as a technology user but as a co-developer, sharing technical improvement feedback on the Nemotron architecture with NVIDIA during the model training process. Through this, the goal is to be the first to deliver top-tier sovereign AI specialized in the Korean language and Korean culture.