[mcjshciejebcu]

That's getting close, and it doesn't look good for bitcoin, but the estimates of visa's energy useage are from a visa published report, and they seem to omit the energy usage needed to sustain employees. Of course, the bitcoin energy estimates also omit the energy usage for employees, but it's easy for me to imagine those are dwarfed by mining costs, and it's not easy for me to imagine something similar for visa.

[jcranmer]

The energy usage to attribute to employees should be proportional to the number of employees more than anything else. There's no a priori reason to expect that a "Bitcoin employee" uses more or less energy than a "VISA employee."

So as far whether or not that matters... do you expect that there is a substantially different number of employees need to maintain the VISA system than the Bitcoin system, on a per-transaction basis? Even if you're generous and say that VISA needs 10× the employees that Bitcoin would (were it the same size as VISA), it would still require that an employee use 100,000× the energy cost of a transaction to overcome the fact that Bitcoin transactions require 1,000,000× the energy of VISA.

[mcjshciejebcu]

I would expect visa to need more employees to maintain their systems, but that's just my intuition, and it might be wrong. You're right, too. Even though humans are very resource intensive (they need > 2kWh of energy per day), they're unlikely to outweight many orders of, magnitude difference.

I'd like to see an estimate of visa's energy usage using the same method the source you linked used to estimate bitcoin's energy usage. Take visa's revenue and multiply it by some fraction of revenue used for energy use (they used 52%). Some very rough calculations I did put it about 1000 times more energy expensive than visa's own report. That's still 1000 times better than bitcoin, using the same method, but it'd interesting to know where the discrepancy comes from, and whether that should also be considered for bitcoin's energy use.