Samsung Foundry exists, and Nvidia used them for the 30 series iirc. TSMC just doesn't have that much pricing power as a result. Their best processes are obviously good, but if they try to gouge too much customers probably just take the hit for a year. If they have to they can always sacrifice size/power for performance on an older, cheaper process.
The real question is why doesn't TSMC charge more? How can a customer using the most cutting edge fab get 60% margins? (Similar for other customers).
Clearly there's a lot of pricing power on the fab side that's not being asserted.