In short, the transaction is “paid” in two ways.
The originator of the transaction pays an amount of ETH based on the complexity of the transaction. The client has so-called “nodes” executed on the blockchain, each (full) node in the network performs this transaction, so the result can be validated. These transactions (contracts) require minimal computing power, and you can so to speak on a Raspberry Pi (think, for example, storing a piece of data in the blockchain, or transferring a token from wallet A to wallet B, and that kind of “transactions/ requests”).
This is independent of the energy costs (the second type of “cost”) required to keep the current Proof of Work network running, the so-called “mining,” which remains with BTC, for example. The latter becomes cheaper, because heavy calculations are no longer required for mining blocks.
In order to make executing on-chain transactions (as a client) also cheaper, you should instead consider an intermediate layer (called Layer 2) that executes transactions and stores only the end result in the blockchain, so that not every node contains all the data. Perform operations within nodes .
I’ve tried to describe it in Jip-en-Janneke language, the whole process is a bit more complicated than the way I explained it.
“Coffee buff. Twitter fanatic. Tv practitioner. Social media advocate. Pop culture ninja.”