AI starts at the nanometer scale

ZEISS SMT has been helping answer that question for more than 50 years: Optical lithography has long been the key technology behind microchip manufacturing. With DUV technology (deep ultraviolet light) ZEISS SMT helped combine the growing demand for high-performance chips with cost-efficient production, laying an essential foundation for the digital age. But AI is only as powerful as the chips it runs on. As AI applications become more advanced, the demands on semiconductor manufacturing increase dramatically. More transistors have to fit into ever smaller spaces – with absolute precision, repeatability, and quality. Here, the wavelength of light becomes critical: the shorter the wavelength, the finer the structures that can be patterned onto a wafer, the thin silicon disc that serves as the basis for microchips. With DUV technology, the potential for further shortening the wavelength had largely been exhausted. That changed with EUV technology (extreme ultraviolet light). Together with its strategic partner ASML, ZEISS SMT helped usher in a new era of chipmaking by reducing the wavelength used in lithography from 193 to 13.5 nanometers. This made it possible to image structures on a wafer that are around 5,000 times thinner than a human hair. That level of precision created the conditions for the AI applications we know today. The impact is significant: around 80 percent of all microchips produced worldwide are manufactured using ZEISS technologies. In other words, the future of AI and the future of the semiconductor industry are inseparably linked.

AI is also becoming an increasingly important part of the entire value chain. From software development to procurement, AI-supported systems help accelerate processes and create space for better-informed decisions. One particularly compelling example is the Machine Agent – an intelligent assistance system that is transforming collaboration between people and machines.
The Machine Agent connects a wide range of knowledge sources across the company and makes them accessible quickly and efficiently. In practice, that means when a disruption occurs, the system can analyze the situation, identify possible causes, and guide users step by step through the resolution process until the machine is running again. But its value goes far beyond troubleshooting. Machine operators can quickly access the right work instructions, technologists can build on proven solution paths, maintenance teams receive structured guidance, and development teams can analyze complex error patterns across system boundaries.
The system is continuously improving. With every interaction, documentation input, and user feedback loop, it becomes more precise. What once required time-consuming research across different systems can now happen in seconds. The result is faster problem-solving, shorter downtime, and employees who can apply their expertise more effectively where it matters most.