Synopsys Says Chip Engineers Must Become 'System Validators' in AI Era

What happened

Frank Schirrmeister, vice president of strategic programs for system solutions at Synopsys, said in a recent interview with THE ELEC that as AI drives increasingly sophisticated automation, chip design engineers must take on the role of "system validators," assessing whether AI-generated results are accurate and safe; Synopsys sees such validation capabilities as a key competitive advantage for South Korea, which has expertise in high-bandwidth memory (HBM) and semiconductor packaging. He expects it will take more time before AI can autonomously handle the entire chip design process: agentic AI-based chip design will spread across individual workflows rather than being introduced all at once, and the stage where AI coordinates the entire engineering process will take shape over the next several years. Synopsys considers current AI adoption in semiconductor design to be at the "orchestrated agents" stage, where AI agents divide tasks within a single domain such as functional verification (including specification interpretation, RTL generation, test planning, bug fixing and verification completion); the next stage, "multi-domain agents," involves AI agents seamlessly coordinating tasks across design, verification and simulation. Multi-domain agents are needed not only to improve productivity but also to address the increasing complexity of advanced semiconductors as processes shrink: at angstrom-scale nodes, physical effects such as power leakage, signal delays, thermal deformation and electromagnetic interference increase exponentially, and transistor scaling alone is becoming insufficient to deliver PPA improvements; similar challenges arise in HBM and multi-die packaging, where heat and interference from stacking chips or placing them close together can adversely affect performance and lifespan. Synopsys combines Ansys' multiphysics analysis technology with its EDA engines to manage these interactions from early development stages, via the Synopsys Autopilot platform, which deploys multiple AI agents for design and verification that can independently reason, plan and execute EDA tools, supporting both Synopsys' specialized agents and external AI models. Schirrmeister said companies adopting the Autopilot platform are seeing initial benefits in design verification, implementation and PPA optimization; Synopsys previously said it was validating the platform and specialized AI agents across more than 50 customer projects, including Samsung Electronics, TSMC, Intel, Microsoft and MediaTek. Synopsys said an autonomous verification workflow developed using Nvidia AI technology expanded verification coverage by 20% and reduced the time required to obtain verified RTL by as much as 50 times, and Fujitsu reported a 10% to 30% improvement in RTL code generation productivity using Synopsys' verification AI agents.

Impact and signal

Affected components and suppliers

Source and time

Source: The Elec · 11/4/2026 · View original source