Excellent points.

I think that you ought to try to tune your code, you ought to go back and refactor it when you can. However deadlines, and time pressures can force you to compromise. At some point, it does make sense to throw hardware at it. In web development, where you can deploy another server and scale out, I think that it's harder to spend a lot of time tuning when for $2-3k you can throw a lot of hardware into 1 or 2 more servers. If a developer costs $70k, then you can eat up $2k of time quickly.

The issue with many web developers is also talent. So many programmers don't know how to tightly tune code.

In SQL Server, you have a single server. To scale up, it's replace a server, often with much more hardware and it can be $10k. So it makes more sense to spend time tuning. There are still limits. You can still spend too much time tuning and not get new development done, but it's more worth it since a new server costs more $$$. You also will run into scale limits. At some point you can't buy a bigger DB server to handle the load from poor code.

In either case, developers (SQL or otherwise) need to learn more about better coding techniques. Learn how to write better code that performs better with less resources.