The main problem is that somehow there just isn't much money to be made in EDA tooling. Go to any EDA conference and ask anyone in a business booth.

There's quite a bit of money, a license for a small design shop will easily span from mid six figures to millions. There is rampant theft though, see the recent http://norcalrecord.com/stories/511081584-synopsis-claims-ub... as an example

The real reason innovation died is a complex and meandering sadness, but basically the business leaders prioritized head count and comfort over whatever it takes to lure creative, intelligent, academic, etc people into an industry and do non-linear things. I think there are similar parallels to computer architecture and operating systems, but these are a bit more widely approachable so there's still a small influx of academics and new companies.

> There is rampant theft though, see the recent http://norcalrecord.com/stories/511081584-synopsis-claims-ub.... as an example

This is not theft, but unauthorized use of the software.

What I heard from them is most of the money comes from a small percentage of customers doing expensive work. They don't want to take chances on unproven tools since mistakes cost too much. So, they keep using the Big 3. There's occasionally companies like Magma that show up delivering good value for a fraction of the price. Plus cheaper or better alternatives to Big 3 functionality such as Tanner. They usually get acquired, though, by companies that dislike effect of competition on profits. The integration of those features make the competition look even worse.

Those successful in EDA seem to mostly be niche players doing one job a lot better than Big 3.

Seems to be a general problem with design tools in many industries, isn't it?

When the jobs disappear it's hard to sell picks and shovels.

I think part of it is that the cost of messing up and having an undetected design fault in deployed hardware can be so huge for the big players. The big players are where the EDA vendors make most of their money and so derive their mindset. You make money by being reliable, and the easiest way to be reliable is to be conservative. Once the design process works, don't mess with it in case something breaks.

There were some startups in the 90's doing analog synthesis, but all flopped.

Tools for MMIC & RFIC design have become pretty advanced (specifically AWR).