It's also about the systems it's replacing, which all the careful studies I've seen have rated as even less efficient. It also doesn't take into account future improvements in energy efficiency, which in the case of Ethereum are near and profound.
Proof of work pays for network security, not Tx volume - the most secure super-computer ever invented.
Transaction processing is a free side-benefit.
With Lightning, TX Processing scales to meet any arbitrary needs.
People using LN are not transacting using Bitcoin. They are denominating their transactions in Bitcoin, and trusting LN to eventually convert to Bitcoin (at the huge bitcoin transaction cost) if/when they eventually need it. Even worse, most people will not use LN directly, they will use yet another third party, one that will operate a bitcoin wallet and LN channel for them (likely the same wallet and channel for many, many different people).
My understanding is that they are onboarding something like 20,000 new users per day through community trainings and bootcamps.
Users can participate with any Lightning wallet such as the free and open BlueWallet.io app or any other Lightning app. No one is obligated to use Strike.
That is the beauty of free, own and interoperable standards such as Bitcoin or Lightning.