Naver built HyperCLOVA X as South Korea's flagship sovereign language model, leveraging Naver's search, news, Knowledge iN, maps, and commerce data for Korean linguistic depth that foreign GPT-class models struggle to match. In 2026 Naver expanded GAK Sejong with NVIDIA DSX infrastructure (55 MW scaling toward 200 MW and a 1 GW path) and joined the NVIDIA Nemotron Coalition, planning future HyperCLOVA versions on Nemotron 3 Ultra open bases with proprietary Korean training. Enterprises access it via CLOVA Studio on Naver Cloud Platform; regulated sectors often require this domestic stack instead of US APIs.