Yes - this is true. A program will either halt or not, and if you could pluck that answer from somewhere, you would know.

But my point is in practice, this is just not possible, because the problem spaces are too large.

A related problem is the Busy Beaver Problem - how many steps can a program take as a factor of the number of turing instructions it contains. If we could determine that, we would "solve" the halting problem, in that we would know a program that is still running after N steps will never terminate.

However, even for very small programs, N is so large to be incomprehensible; ie, all the stars will have burned out even with all the compute in the world running until that happens.