Tech news in 3 minutes

CEO Interview with Suresh Vasudevan of Clockwork.io

35 d ago

A recent panel session moderated by Ed Sperling (Semiconductor Engineering) featured Walden Rhines (Silvaco), Vincent Wong (Verific), Dave Kelf (Breker Verification Systems), Shelly Henry (MooresLab AI), Ann Wu (Silimate), and Cindy Cui (ChipAgents). Hosted by the Electronic System Design Alliance (ESDA), a SEMI Technology Community, the discussion focused on AI’s role in semiconductor design. The panel concluded that AI is no longer experimental in electronic design automation (EDA) but a core workflow component. Current AI tools act as assistants—generating RTL, testbenches, and documentation. The next phase is agentic AI: autonomous systems executing multi-step tasks across specification, architecture, verification, and signoff, potentially compressing chip cycles from years to months. However, trust and correctness remain paramount. Unlike consumer AI, chip errors are catastrophic, causing months of delays and millions in costs. Panelists stressed that generating output is easy; verifying correctness is hard. Thus, human-in-the-loop approaches are essential—engineers will supervise AI, not be replaced. Their roles shift from creators to reviewers and orchestrators. The industry lacks standardized benchmarks to evaluate AI quality, coverage, and reliability, hindering adoption. Design space exploration may be AI’s greatest value, allowing evaluation of far more architectural alternatives than manual methods. AI could also democratize chip design, enabling smaller organizations to develop custom silicon. On workforce impact, panelists agreed that entry-level tasks will automate, but demand for experienced engineers will persist with new skills: systems thinking, AI supervision, and critical evaluation. The key challenge is ensuring future engineers develop deep expertise. Ultimately, agentic AI will transform semiconductor design, but trust, accountability, and engineering discipline are critical enablers.