Nvidia is moving deeper into the chip-design stack, teaming up with Siemens to run the German giant’s electronic design automation (EDA) tools on Nvidia GPUs.
Announced at CES 2026 during Siemens’ keynote in Las Vegas, the deal aims to speed up one of the most compute‑hungry steps in building modern processors.
Nearly every chip on the market today is designed with EDA software. As features shrink and transistor counts rise into the billions, those tools chew through massive compute cycles. Most of that work still runs on CPU-based servers.
Nvidia wants to change that by shifting key workloads onto its GPUs, where the company claims it can crunch EDA simulations faster and more efficiently. Siemens, one of the big three in EDA, is the latest partner to sign on.
The collaboration goes beyond making today’s chip flows run faster. Nvidia and Siemens say they want to build detailed digital replicas of entire systems — from individual chips to full racks — and test how they behave before any hardware is built.
“What we are hoping for, and the reason why we’re partnering so closely together, is so that we could build that Vera Rubin in the future as a digital twin,” Nvidia CEO Jensen Huang said on stage at the Siemens keynote.
The Vera C. Rubin Observatory, a massive telescope project in Chile, has become shorthand in the industry for large, complex systems that push hardware and software to their limits. Huang’s point: if you can simulate something at that scale, you can de‑risk designs long before they hit the fab or the data center.
Digital twins are already common in manufacturing and heavy industry. Bringing that same idea down to the level of chips and server racks would let engineers probe performance, power, and reliability virtually, then iterate before they commit to silicon or steel.
For Nvidia, the Siemens deal reinforces a broader message at CES 2026: GPUs aren’t just for AI training anymore. They’re becoming general‑purpose accelerators for everything from scientific computing to, now, the tools that create the chips themselves.
For Siemens, tapping Nvidia’s GPU stack is a way to keep its EDA portfolio relevant as design workloads explode and customers push for faster time‑to‑market.
If the partnership delivers, the next generation of chips — and the racks they run in — may be extensively debugged and optimized long before anyone powers them on in the real world.
