Excuse any ignorance I may have and please consider that I am attempting to use Cunningham's Law here; posting the wrong answer and waiting for someone to correct me.

Could modular exponentiation, some form of modular math (a concept that I’m new to), or some other mathematical computation be used to infinitely scale bitcoin?

Im this case, a block wouldn’t be an aggregate of the block data but a mathematical computation that (when solved) reveals the data of the block. Similar to a trapdoor function. Bc the data requirement for a single function is much smaller than the result of the aggregate block data (esp. as it grows) and the computation required is “simple” when the trapdoor is solution is available. So, a theoretically infinite amount of transactions could exist in a single computation. The difficulty in finding this trapdoor might be prohibitive, but I could it be reasonable?