Nvidia’s $1B Naver Bet: AI Data Centers in South Korea
Nvidia backs Naver’s 200MW AI data center expansion in South Korea—what it means for builders chasing capacity.
The story of the day
Nvidia’s $1B investment in Naver’s AI data center expansion isn’t just capital—it’s a capacity play. The South Korean internet giant will scale its planned AI data center from 55MW to 200MW, a move that directly addresses the chronic GPU shortage plaguing developers. For builders, this means one less bottleneck in the race to deploy models at scale. The deal also signals Nvidia’s strategy: control the stack by financing the infrastructure that runs its chips. Expect more such partnerships as cloud providers scramble to secure compute.
New tools & launches
Agent Console — Local dashboard for managing Codex and Claude Code workflows. Open source. [GitHub]
adCasa OS — AI marketing workspace with Bayesian attribution for ad spend optimization. [adcasa.io]
Models & research
NVIDIA Cosmos-H-Dreams — Proprietary framework for real-time generative simulation in surgical robotics. Targets latency-sensitive medical applications, leveraging Nvidia’s GPU stack. No open weights. [Nvidia Blog]
Industry pulse
China’s state media clarified its stance: support for open AI models has limits. According to Bloomberg, the message targets domestic developers relying on foreign open-source tech, reinforcing Beijing’s push for self-sufficiency. The subtext for global builders: open-source may not stay open in China-aligned markets.
Stat of the day
CXMT’s Shanghai debut surged 472%, valuing China’s top memory chipmaker at $487B. The rally underscores investor bets on AI-driven demand for domestic memory—despite U.S. export controls. [The Information]
Get the full analysis — subscribe to Gradient Pro.
