The existence of uncomputable problems does not mean the things we know are not from computation.

I would submit that a computation that never completes is probably still a computation.

If you show me a TM with an infinite length memory tape I’ll accept that.

Presumably you mean the halting problem for Turing Machines? It's undecidable for other Turing machines, but that doesn't mean it's undecidable for all machines.

Is there a physical system that solves the halting problem?

> "yet it is implemented in the universe"

Or dividing by zero