In general I have come to treat anything that cites Alex de Vries with suspicion. Amongst other things, the repeated discussion of the number of transactions and the power expenditure. The power expenditure does not scale by number of transactions. Associating the two without mentioning one is not a factor of the other creates a misleading impression. Doing so repeatedly suggests bad faith.
Bitcoin does use a lot of power though. On the other hand, the reward halving does mitigate some of this. If you are confidant that Bitcoin won't increase on average 20% every year forever then you should be happy that the problem will fix itself.
The current problem comes from the value rise being greater than the reward halving. To sustain that it has to double in value every 4 years _forever_. The early growth has been higher than that, but to just sustain current power levels BTC would have to be worth millions in a few decades.
But in the future, the amount of energy consumed for a single transaction will tend towards exactly the fee paid for that transaction multiplied by the cheapest electricity price in the world.
So either the transaction fees will be outlandish, or for a transaction that costs $0.10 you will consume roughly 2kWh.
Trip to the shops, but fuel, some take away and a few things online when you get home and you’ve used 10-16 kWh of energy, multiple that by a population (even a small one), and even the massive investment into renewables that there is now will be quickly eclipsed.
Adding transactions that are so intensive to people’s everyday energy consumption would be a huge step backwards IMO.
Making a bitcoin transaction is like recasting gold bars - not something you want to do often with small amount of gold. This will be done only for large individual payments or for large settlement payments aggregating thousands or millions of normal payments. Everyday payments will be performed on second or higher layers (e.g. Lightning Network)