NVIDIA expands sovereign AI and manufacturing work with South Korea

NVIDIA and South Korea’s Ministry of Science and ICT are expanding cooperation across the country’s sovereign AI ecosystem, according to NVIDIA. The company said the work spans AI semiconductors, data-center infrastructure, talent, startups, models, agents and security. It also creates a path for Korean NPU and XPU companies to explore joining NVIDIA’s NVLink Fusion ecosystem, NVIDIA said.

The program includes training and certification, hackathons and technical seminars intended to strengthen use of GPUs, according to NVIDIA. Under South Korea’s AI for All initiative, NVIDIA said it will work with three selected consortia and four Korean foundation-model builders, using its Dynamo software to help optimize and scale AI inference. NVIDIA also announced a collaboration with SK Telecom and Rebellions, supported by the ministry, to develop sovereign AI infrastructure for Korean models and workloads using NVLink Fusion.

Separately, NVIDIA described efforts by Korean companies to apply accelerated computing, AI and simulation in manufacturing, robotics and autonomous vehicles. It said Samsung Electronics is expanding work with NVIDIA on semiconductor engineering, including materials research, process development and device engineering.

For technology computer-aided design, Samsung is using NVIDIA cuISS and cuDSS to accelerate calculations involving sparse linear systems, and the companies are also working on GPU-accelerated process and optical simulations, NVIDIA said. Samsung’s sMOL platform has completed selected quantum-chemistry calculations up to five times faster end to end using NVIDIA cuEST, according to the company. Samsung is also exploring NVIDIA cuEquivariance for tensor operations in atomistic AI model training and inference, and developing chip-scale thermal and mechanical analysis workflows with NVIDIA PhysicsNeMo.

NVIDIA reported that one AI homogenization case combining AI-predicted material properties, coarser grids and GPU acceleration achieved an approximately 1,600-fold speedup. The company presented the work as an example of AI and simulation moving into industrial production.