TSMC has announced 16nm university program. You guys have to be real fast to keep the community with efabless. Otherwise, 16nm vs 130nm will be lost battle.

Let me know when TSMC are opening their 16nm PDK...

They are probably not publishing that PDK without an NDA... Its probably the same as other shuttle programs with NDAs. The Sky130 PDK was primarly intended for developing and sharing actual concepts, as well as developing free / open EDA Tools... And still a 130nm Node ist still far ahead of most university fabs... But I would love to be proven wrong...

Yes, I doubt they will too, my comment was highlighting why efabless and tsmc aren't really in competition

The reason I have gone with Efabless is the Open Source CAD Tools. The last time I checked TSMC, they do not support Open Source CAD Tools

I think they actually support LTSpice, which would be at least freeware... But I might be wrong on that...

what does this mean? Are they providing Free Tapouts?

Sadly... No... It means, that you can just download the software for free... btw. I got notified by another user, that it is actually not allowed to use LTSpice for VLSI...

Mask set cost of 130nm process: $100K. Mask set cost of 16nm process: $2.5M. (Rough estimates).

Rakesh, The points that I think others are making above are valid. We also do clearly see the benefit of an educational advanced node and would be delighted if one could be brought to the open source market but it is not something that can be afforded by the majority for several reasons. Mask costs as Tim is pointing out are the largest barrier Making it untenable as an undergraduate program offering and very rare even for post graduates. We can only hope that sponsors would be willing to solve that financial problem.

The entire industry has been obsessively focused on performance, as indicated by the industry obsessing over Moore's Law. Sadly, in the attempt to stuff more transistors per area onto silicon, it has been forgotten that there are vast unexplored areas of how to do clever things with transistors. I read in a news article recently that ChatGPT is costing OpenAI several cents per conversation to operate. When you have to build several power plants just to run a piece of software that's still not quite as clever as something that evolution managed to stuff into a volume of 1200 cubic centimeters, you have missed some very important points along the way. . .