This is what the lightning network is for: scalable, instant, low-cost, and off-chain TXs. All transactions between the opening and closing of a payment channel are consolidated into two TXs. One to open it another to close it. This means energy cost per TX goes down pretty significantly with adoption.
Similarly, we've had folks transacting personally with Bitcoin in Venezuela and Zimbabwe. It tends to work out fairly well for them - at least their money holds value. But transaction costs in Bitcoin during the 2017 bubble were about $20/transaction, which makes it a pretty high-friction experience for groceries. There's a lot of market pressure to create other solutions (Lightning Network? Stellar? Local fintech companies?) for payments when Bitcoin costs $20 for everything.
Aren't we losing the assumed federation in this system? A country whose banking system is under strain (Nigeria) or has collapsed (Zimbabwe) will have a bad-faith government acting to constrain the free-flow of money (because they want a transaction-fee cut).
Wouldn't that still lead to me wanting to go buy my groceries with BTC? Do we create a localised version/model where people can transact with each other without touching the blockchain (because of the low throughput), or how do we address this use-case; as it seems to be the most attractive one?
Has the energy requirement peaked, or is it still going to increase? Linearly or not?