Summary:

Twenty-five years ago, SEMATECH first alerted the world to a concern known as the design productivity gap: the observation that the ability to manufacture complex chips had started outpacing the capability of designers to create them by more than a factor of two. This concern was subsequently reiterated in the ITRS report of 1999 and discussed and reported on in many articles during the past two decades.

In recent years, generative AI in general and natural language processing more specifically have taken the world by storm, opening-up a wealth of possible applications, including chip design.

But will it be enough to finally start closing the productivity gap, where continuous improvements in EDA and the application of IP- reuse have done nothing more than decelerate its growth somewhat? This article presents a comprehensive overview, showing that generative AI will indeed finally close the design productivity gap and even enable us to fully automate hardware design.