Foxconn's CoDoctor AI platform — now packing agentic AI agents — and its Nurabot nursing collaborative robot have moved from pilot programs into full clinical operations across Taiwanese hospitals, powered by NVIDIA NemoClaw as part of the government-backed "Healthy Taiwan" initiative.
From pilot to production
The rollout brings NVIDIA's physical-AI stack — GR00T-H vision-language-action models, Cosmos synthetic-data pipelines and the Rheo hospital digital twin blueprint — into a country-scale clinical environment. Nurabot, jointly developed by Foxconn and Kawasaki, handles patient transport, supply runs and support tasks that let nurses redirect attention to bedside care. CoDoctor's agents sit above clinical workflows, drafting notes, triaging alerts and coordinating multi-system tasks that used to require full-time care coordinators.
Anchor customers and scope
Chang Gung Memorial Hospital and other participating medical centers are the anchor deployments. Taiwan's Ministry of Health and Welfare is backing the program under the Healthy Taiwan framework, which pushes Foxconn's agentic and physical AI stack into public-hospital operations. NVIDIA describes the setup as the largest hospital-network-scale deployment of its clinical AI stack outside the U.S., a template it wants to replicate in Japan and Korea.
Same rails as the rest of physical AI
The clinical stack shares plumbing with NVIDIA's factory-floor rollout: Cosmos 3 Edge for Japan's robotics giants, Kawasaki + Dexterity's Mech humanoids and the NVIDIA-Hyundai Atlas alliance. What's different in healthcare is the regulatory perimeter: every workflow that touches a patient runs into a distinct approval process, and Taiwan is functionally the first jurisdiction to open that door at scale.
What to watch
Whether Foxconn can hit its Nurabot production targets, how quickly Chang Gung's CoDoctor agents extend beyond triage into billing and diagnostics, and whether Japanese and Korean health systems adopt the same NVIDIA + national-champion pattern.
Reporting based on coverage from NVIDIA Newsroom and NVIDIA Blogs.