I would think patenting the idea is the way to go, rather than verifying your design on an ASIC. Chances are that’s much more expensive, though, especially for a worldwide patent.
It depends. How many billion dollars are you willing to pay TSMC for fab access? If it's not at least $5 bln then you might as well pack it up and go home, Nvidia and Apple can take things from here.
I would think patenting the idea is the way to go, rather than verifying your design on an ASIC. Chances are that’s much more expensive, though, especially for a worldwide patent.
Patenting it would have been a good idea a decade ago before I first started talking about it in public
> How does that lead to actually realized profit?
It depends. How many billion dollars are you willing to pay TSMC for fab access? If it's not at least $5 bln then you might as well pack it up and go home, Nvidia and Apple can take things from here.
Small scale fab access is in my budget, enough to prove the point. (Less than $500)