In the name of avoiding a semantic debate, I'll merely point out that many things that are deterministic cannot be modeled via models that are simpler than the system which they are intended to model (such as many aspects of economics). So functionally, not all systems can be modeled and not all states can be predicted.

Either way, are you trying to determine the total energy used by the Bitcoin network, or the energy used by the particular minor that finds a block, since the last block? If it's the first, the fact that you're posing the question rather than simply using publicly available estimates implies that you need more accurate or precise data and I'm curious if public companies consider that to be a trade secret, and if it's the first or second, I could see that requiring hardware updates, again, depending on the accuracy and precision that you need.

I apologize if I seem dismissive or critical; in my field it's regularly joked that ideas are worthless because basically all of the value is in the execution and I get the impression from you that you have pretty specific data requirements, but occasionally there are revolutionary ideas that are worth some serious effort!