New products or features are of an acceptable performance only, sometimes not even that. Additional work on performance tuning is often followed up in a later version. It's really about priorities that have to be decided, and deadlines usually win.

Those that prioritize performance upfront can find all that work thrown out if the design needs to change for some reason.

But I think that performance by design upfront should be done where possible. This is where experience helps a ton.

Allowing (encouraging even) the paying down of technical debt is the real solution. Build the prototypes fast and broken, if they catch on then fix them. This was the way things were once done, and how most of the billions were originally made.