AI in Semiconductors

Kioxia Yokkaichi Plant, one of the world's largest flash memory fabs (Fab7 operational since fall 2022), processes 3 billion data points daily from thousands of tools. Uses AI/ML to infer device failure root causes and detect defect signs automatically across the wafer lifecycle.

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.

SK Telecom applied digital twins with NVIDIA Omniverse libraries to SK hynix semiconductor fabs, validating large-scale physical AI. Plus SKT's Agentic Digital Twin Modeling using NVIDIA Agent Toolkit automates 3D manufacturing data processing, targeting SK hynix Autonomous Fab by 2030.

Verkor's Design Conductor AI agent autonomously generated a layout-ready GDSII RISC-V CPU (VerCore) in 12 hours from a 219-word spec, clocked 1.48 GHz with CoreMark 3,261. Verified in simulation only — not yet fabricated on silicon; 5-10 human experts still needed for production-grade quality.

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.

Synopsys and AMD Honored for Generative and Agentic AI Vision, Leadership, and Impact Synopsys and AMD were recently selected by the World Economic Forum for inclusion in the WEF's MINDS (Meaningful, Intelligent, Novel, Deployable Solutions) AI program, recognizing their leadership and real-world impact in applying generative and agentic AI to semiconductor design and engineering.

Cognichip deploys deep learning models alongside semiconductor engineers to design next-generation computer chips, aiming to reduce chip development costs by over 75% and cut design timelines by more than half.