Their goal is also to stay in the race. It's an arms race much like the old US vrs the Soviet Union. TSMC must spend to stay in the race or Intel will gobble up the high margin customers and leave TSMC with the table scraps.
TSMC makes incredibly good money on what you are calling table scraps. Probably more margin than what they make in trying to stay in the N-generation process arm's race.
Assuming they continue to fall behind as we project, I still see TSMC as the largest foundry in the world, and even without a bunch of N generation wafers, they will be a very successful company.
That doesn't mean that "staying in the race" is unimportant. It's a huge incremental opportunity for TSMC. So there's no reason for them not to invest in it. It's just not necessary for them to have an otherwise successful business.
Market Data 
Markets 
AI
