Siemens has expanded its collaboration with NVIDIA to accelerate computational lithography for next-generation AI chips, delivering up to 12x faster full-chip optical proximity correction (OPC) workloads. The companies are integrating NVIDIA’s cuLitho GPU acceleration library with Siemens’ Calibre manufacturing software to tackle one of the biggest bottlenecks in advanced semiconductor manufacturing. The development is particularly relevant to chipmakers and foundries moving toward High-NA EUV lithography, where increasingly complex patterns demand significantly more computing power. It could also help European semiconductor manufacturers improve production efficiency and accelerate the introduction of advanced AI processors and memory devices. Overcoming the lithography bottleneck As semiconductor manufacturing moves toward smaller process nodes, computational lithography is becoming increasingly demanding. Traditional OPC techniques struggle with the processing requirements of advanced patterning, particularly when using inverse lithography technology (ILT) to optimize complex, curvilinear mask patterns. These computational challenges can limit full-chip optimization, forcing manufacturers to balance processing time against patterning accuracy. The resulting compromises can increase the risk of lithographic hotspots, reduce process margins and ultimately affect chip yields. By combining Calibre with NVIDIA cuLitho, Siemens aims to remove these limitations and make computationally intensive patterning techniques practical at manufacturing scale. “Computational lithography has traditionally been one of the most compute-intensive stages of the design-to-manufacturing flow,” said Ankur Gupta, executive vice president, EDA IC Portfolio, Siemens. “By combining Siemens’ signoff-grade Calibre technology with NVIDIA’s GPU acceleration, we are giving our customers the ability to deploy advanced patterning strategies like curvilinear ILT at scale. This isn’t just about speed; it’s about enabling the precision required for the next generation of AI infrastructure.” Up to 17x faster lithography simulations According to Siemens, verified engineering data shows acceleration ranging from at least 10x to 17x across key lithography simulation workloads, alongside the reported 12x improvement in full-chip OPC. The performance gains could allow chipmakers to run more optimization iterations, detect potential manufacturing defects earlier and improve photomask accuracy before tapeout. Importantly, the acceleration also makes full-chip curvilinear ILT more practical for High-NA EUV lithography, a technology expected to play a key role in future leading-edge semiconductor production. Faster path to advanced AI silicon The collaboration targets both semiconductor manufacturers and photomask suppliers, helping reduce turnaround times for complex patterning workflows. For manufacturers developing high-performance AI accelerators and advanced memory, shorter simulation cycles could translate into faster manufacturing readiness and improved process margins. The integration also highlights the growing role of GPU computing beyond AI training and inference, with NVIDIA technology increasingly being used to accelerate the engineering workflows needed to manufacture AI hardware itself. The post Siemens and NVIDIA speed up AI chip lithography by 12x appeared first on eeNews Europe .