EDA Automation Requires Formal Proof and Audit Trails
The article argues that AI's biggest opportunity in EDA is workflow-level orchestration, not just improving individual tools. Trust requires evidence, semantic continuity, and independent verifiability through deterministic verification and audit trails. The emerging model is human-guided, agentic automation where engineers retain control over specifications and signoff.
The article describes semiconductor design reaching a scale where coordination between tools matters as much as the tools themselves. Multi-die assemblies and chiplets create interdependencies that fragmented point tools struggle to manage, while advanced-node design costs reaching hundreds of millions of dollars make errors extraordinarily expensive. A single respin can jeopardize an entire program.
Siemens EDA's Ankur Gupta notes customers are demanding at least a 2× reduction in time to results. The economics of AI orchestration also factor in — poorly written tool interfaces can drive up token and orchestration costs, so vendors are doing integration work in advance to let customers use either vendor-supplied or proprietary agents efficiently.
This shift toward AI-orchestrated EDA workflows could reshape who can participate in semiconductor design. If automation reduces reliance on scarce expert engineers, smaller companies may gain access to advanced-node design capabilities previously out of reach. However, dependence on agentic systems with audit trails may also concentrate design knowledge within a few large EDA vendors, potentially affecting workforce dynamics and the accessibility of chip design expertise across the broader technology industry.