Agentrys raises $24.5M to automate chip design with AI agents

Agentrys sits at the intersection of two leverage points in the AI stack: chips and agents. If AI can reliably automate large chunks of chip verification and physical design, it shortens the iteration cycle for new accelerators and specialized hardware. That, in turn, feeds back into how quickly labs can spin new architectures for both training and inference.

The company’s “Agentic Design Automation” framing matters. Instead of one monolithic model, it is building multi-agent workflows that operate over existing EDA tools, customer data and evaluation signals. Those are exactly the kinds of structured, feedback-rich environments where agentic systems can move from toy demos to production utility and where emergent failure modes can be studied at scale.

For the race to AGI, the more that AI systems design the next generation of AI hardware, the tighter the optimization loop becomes. That could accelerate capability growth if it makes it cheaper and faster to explore chip design spaces that humans would not practically search. It also raises new safety questions about verification and interpretability when both the models and the hardware they run on are co-designed by AI.