Then your overarching thesis is not very clear. Is it simply ‘don’t invest in hardware capital, software always makes it worthless’?

It's more like: don't invest $100B in capital when there's still orders of magnitude improvements in software to be made.