VC-backed EDA founders argue AI agents limited by chip design's long tool loops

ai · x · 2026-09-29

A VC highlights Nojan Sheybani and Kevin Gubbi as a rare credential-plus-learning-rate combo building in the legacy-heavy EDA space.

The quoted thread carries the substance: chip design cycles are extremely long, and one small upstream RTL change can cascade into huge downstream impact. The best fixes are speed-oriented — shortening simulation wall time or using heuristic methods for fast estimation along the way. Notably, the author points out that tool time dominates EDA so much that AI agents can't iterate fast — like compiling code taking 9+ hours each run — and long feedback loops significantly lessen AI's impact.

Original post →

More from Companies & People

Companies & People channel →