1) When they bought these wafers, Intel 4, Intel 3, and 20 A were all just a hope and dream. Intel was burned for years by refusing to use TSMC. Imagine what would happen if those nodes failed AND Intel wasn't using TSMC. Intel would have actually been doomed as you like to say. If that happened, then Intel would neither have a foundry or CPUs worth buying. This is and always was about risk reduction--avoid Intel's past mistakes.
2) Intel is now free to use whatever foundry is right for the right tile. Intel now buys the wafers from the best known node at the time the decisions are made. This means Intel can now use nodes that are the same or even better than AMD has access to (for cases when Intel's nodes beat TSMC in the future).
3) As a foundry, Intel's eggs are all in 18A. Using 3 nm class TSMC nodes has nothing to do with competing as these are not comparable nodes. Dozens of companies are signing up for 18A. That is when Intel starts competing with TSMC to be a foundry. This is like the Tesla is using a Ford van meme. Tesla is not competing with Ford in the van space.
https://www.google.com/search?sca_e...kEHY3zBTcQtKgLegQIDRAB&biw=1500&bih=839&dpr=2
4) It drives up costs and drives down production for AMD as Intel takes TSMC's limited capacity away from AMD.
5) This frees up Intel to sell foundry services to outside companies and actually become a foundry for others. Without out this, Intel would have to outbid other companies for Intel's limited wafers. This would harm Intel's goal to become a foundry when wafers aren't available AND drive up costs of Intel's CPUs.