AI in Taiwan

Built on NVIDIA Isaac for Healthcare, Foxconn's Nurabot collaborative nursing robot is field-tested at Taichung Veterans General Hospital, where it has cut nurse workload by 30% and saved nurses two to three hours per day on transport-related tasks. The robot uses Jetson AGX Orin for edge inference, Holoscan for real-time sensor processing, and Isaac Sim with Omniverse digital twins to validate deployment 40% faster, while Cosmos handles task scheduling and predictive route optimization at 98% navigation accuracy.

Taiwan's National Science and Technology Center for Disaster Reduction deployed NVIDIA's Earth-2 CorrDiff generative AI weather model for typhoon forecasting, with the model being 500x faster and 10,000x more energy-efficient than traditional CPU numerical weather prediction.

ASE Inc deployed Qualcomm Dragonwing AI On-Prem Appliance with Cloud AI 100 Ultra cards (870 TOPS INT8 per card) in cleanrooms, running vision inference for wafer defect detection and an LLM-powered Maintenance Copilot (aMA) at the edge. 3.8x better perf/W vs NVIDIA A100.

AIBIZ Technology deployed AMD Instinct GPU-accelerated Lean AI for semiconductor wafer yield optimization in Hsinchu production lines, combining edge inference on process sensor streams with ROCm-based GPU clusters for real-time anomaly detection.

TSMC deployed NVIDIA cuLitho for computational lithography and Omniverse for FabTwin digital twins across production fabs, achieving 20-50% cost or cycle-time gains in mask synthesis and virtual fab design. Partnership covers design, manufacturing, and equipment layout validation.

Pegatron built PEGAVERSE, an Omniverse digital twin platform with six virtual factories, and an AI factory on NVIDIA AI Enterprise and DGX. Its VSS-based Assembly Guiding Agent flags assembly errors such as missing screws in real time, cutting defects 67% and labor costs 7% per line.

Taiwanese furniture brand MR. LIVING put Crescendo Lab's CAAC AI Agent on its LINE account, which gets 16,000+ messages a month. The agent answers 75% of routine questions, sends sensitive cases to humans via topic control and keyword detection, and saves an estimated 160 agent-hours a month.

Compal Electronics (仁寶電腦), a top Taiwanese contract electronics manufacturer, scaled AI across more than 100 production lines by tackling AOI false positives. A "Data PK" head-to-head test exposed that the actual human error rate was ~5% (vs. the 0.02% that line staff believed), which built internal trust. Outcomes: AOI false-positive rate improved by 97%+ (stabilised at 0.5%-1%), assembly AOI test coverage rose 30%+, and AI now runs across 500+ global sites via a hub-and-spoke model that onboards sister company AcBel Polytech on a shared multi-agent platform running on-premise.