ChipAgents and NVIDIA Launch Build-A-Chip-Bench to Test AI Chip Design Agents

ChipAgents introduced Build-A-Chip-Bench in collaboration with NVIDIA to evaluate whether AI agents can execute end-to-end digital chip design from specification through physical implementation and verified GDS. Initial benchmark evaluations showed that tested AI agents passed functional verification on ~40% of tasks on average, while only approximately 10% met the required physical performance, power, and area (PPA) targets. While existing hardware-AI benchmarks primarily stop at register-transfer level (RTL) generation and verification, Build-A-Chip-Bench extends testing across logic synthesis, floor planning, placement, clock-tree synthesis, routing, timing and power signoff, and physical verification. The benchmark was created by Surya Gunukula, Jerry Yicheng Wang, and Kevin Zhang alongside technical expertise from NVIDIA to measure how well autonomous systems preserve validated behavior and operate specialized engineering tools across physical design stages. "The real challenge is carrying engineering intent through a long sequence of interconnected decisions until the design satisfies both functional and physical requirements." — William Wang, Founder and CEO of ChipAgents.