Nvidia Bets $100M on Korea’s Physical AI and Agentic Startups

The story of the day

Nvidia is funneling new capital into Korea’s AI ecosystem, targeting startups in physical AI, foundation models, and agentic workflows. At the MSIT-NVIDIA AI Startup Partnership Forum, the company announced investments in Point2 (RF-based AI interconnects for data centers), Motif Technologies, RLWRLD, and Upstage, with Samsung and Hyundai Motor Group already integrating NVIDIA’s OpenShell runtime for secure agent deployment. The move signals Nvidia’s push to own the full stack—from chips to agent safety—while Korea’s manufacturers adopt AI physics for semiconductor and robotics workflows. For builders: expect tighter coupling between Nvidia’s hardware and Korean industrial AI stacks, with OpenShell emerging as a de facto safety layer for enterprise agents.

New tools & launches

NVIDIA OpenShell — Secure runtime for agentic AI with sandboxed execution and policy-based controls. Integrates with Hyundai’s workflows and AIM Intelligence’s Starfort platform. [Nvidia AI Blog]

Point2 RF Interconnects — Scalable radio-frequency links for AI clusters, offering an alternative to copper/optical for data center scale-up. [Nvidia AI Blog]

0rwx Framework — Open-source framework for AI usage tracking and governance. [Hacker News AI]

Models & research

No new model releases or research papers reported in the past 24 hours.

Industry pulse

Google’s UK data center faces local resistance — Protests led by bikers and environmentalists highlight the growing backlash against AI infrastructure’s physical footprint. As data centers proliferate, expect more regulatory and community pushback, particularly in dense regions. [Wired AI]

Stat of the day

$100M+ — Nvidia’s new investments in Korean AI startups, focusing on physical AI and agentic workflows. [Nvidia AI Blog]

Get the full analysis — subscribe to Gradient Pro.