AI Is Moving From Writing Code To Designing Physical Products

AI is rapidly moving into one of the most demanding areas of industrial engineering: designing, simulating and testing physical products. Companies including Nvidia, Siemens, Cadence and Synopsys are developing AI systems that can accelerate complex simulations, automate engineering workflows and help design everything from semiconductors to aircraft.

This week, Siemens unveiled new AI-powered factory capabilities and French AI company Mistral is expanding its collaboration with Nvidia to bring open-weight AI models, physics modeling and autonomous engineering agents into the mix, highlighting an industry-wide race to transform how physical products are designed and built.

The implications could be substantial. Engineering simulations that traditionally take hours or days to run can limit how many designs manufacturers evaluate before committing to expensive prototypes and production. AI-powered physics models promise to compress those cycles, allowing engineers to test more alternatives, catch design problems earlier and cut development costs.

For aerospace, automotive, semiconductor and other complex manufacturing industries, the opportunity goes beyond faster engineering. AI could fundamentally change how products move from initial concept to finished design.

But as AI moves from coding and content generation to making decisions about physical systems, the consequences of getting things wrong become much more serious. Can AI agents reliably handle complex engineering tasks without introducing costly or dangerous errors? Will open models and Europe’s growing sovereign AI efforts challenge established industrial software platforms? And who will control the models, simulation technology and engineering workflows that become central to designing and manufacturing physical products?

AI Is Taking On Engineering’s Slowest Work

Getting such systems to work reliably is a much harder problem than describing what they should do.