Skip to content
hw.dev
hw.dev/signal/siemens-tsmc-ai-eda-collaboration
SignalSiemens

Siemens and TSMC Expand AI Automation Across the Full Design Stack

Siemens and TSMC certified Fuse EDA AI Agent across N2P, A16, and A14 nodes while extending 3DFabric and silicon photonics support -- a systematic push to make agentic EDA the default, not a beta feature.

Thesis connection
validationcoordination

Certifying an agent that drives Calibre DRC, Aprisa digital design, and 3DFabric thermal analysis at N2P and A16 means production signoff flows -- historically the most human-iteration-heavy step -- are being claimed as agent-runnable.

#eda#ai-hardware#semiconductor
Read Original

Siemens and TSMC announced expanded collaboration at the TSMC Technology Symposium, certifying the Fuse EDA AI Agent across TSMC's N3A, N3C, N2P, A16, and A14 process nodes. The practical scope covers DRC automation and physical verification through Calibre, digital design acceleration with Aprisa, 3D IC thermal and connectivity analysis through 3DFabric integration, and joint development for TSMC-COUPE silicon photonics.

The detail that matters is the process node breadth. Getting an AI agent certified at N2P and A16 is not trivial -- these are the nodes where timing closure, power grid analysis, and DRC complexity have historically required the most human iteration. Certifying Fuse Agent there means Siemens is claiming the system can handle real production signoff flows, not just RTL synthesis demos. That is a materially different bar than what the EDA vendors were shipping 18 months ago.

The counterpoint: collaboration announcements between EDA vendors and foundries at technology symposia are a genre unto themselves, and the gap between "certified and available" and "engineers are actually using this in production tapeouts" is wide. Siemens launched Fuse EDA AI Agent at GTC in March; how many teams have taped out with it is the number that will matter. Watch the DAC 2026 customer case studies for signal on actual adoption velocity.