Major semiconductor industry players including TSMC, Cadence, KLA, Siemens, and Synopsys are rapidly advancing chip design and manufacturing processes by adopting NVIDIA Blackwell GPUs, Grace CPUs, high-speed NVLink fabrics, and CUDA-X domain-specific libraries such as cuDSS and cuLitho. These technologies are delivering significant improvements in computational lithography and device simulation, key steps in producing next-generation chips.
TSMC has highlighted the benefits of leveraging CUDA-X and the Grace Blackwell platform, stating that computational acceleration enabled by these tools expedites process development and makes complex simulation more cost-effective. NVIDIA´s cuLitho library, in combination with Blackwell GPUs, speeds up lithography processes by up to 25 times, enabling faster detection and correction of potential manufacturing issues before physical production. In parallel, Cadence has launched its Millennium M2000 platform exclusively on NVIDIA Blackwell architecture, providing a scalable solution for accelerated electronic design automation (EDA) workflows optimized for hyperscale workloads and sophisticated Artificial Intelligence inference.
The momentum continues with Siemens integrating CUDA-X and Grace Blackwell into its Calibre platform, achieving notable speed and accuracy gains for optical proximity correction, physical verification, and manufacturability analysis. Synopsys is using Blackwell and CUDA-X in key EDA tools, reporting performance scale-ups of up to 20 times on advanced simulations and projecting dramatic acceleration of other applications, such as PrimeSim and Proteus. KLA, a key manufacturer of process control equipment, is also collaborating with NVIDIA to advance its physics-based Artificial Intelligence systems and is poised to evaluate Blackwell-based solutions for further accelerating defect detection algorithms. Altogether, embedding NVIDIA Blackwell technology across EDA, manufacturing, and process control is enabling the semiconductor industry to deliver faster, more efficient, and higher-performing chips, meeting the increasing demands for complex, rapid, and high-volume production cycles.
