I don't buy this energy figure at all - where are you getting it from?

120 GWh / 500 million is 240 Wh per person.
The download is 4GB. Downloaded over a 100 megabit connection, it takes about 5 minutes.
240 Wh / 5 minutes is 2.88 kW.

What equipment is consuming an extra 2.88 kW due to me downloading something?

This sounds like one of those pants-on-head "calculations" about how sending an email consumes five billion gigajoules or whatever. They typically add up the energy usage of a data centre and divide by the number of emails going through it. Is that what's happening here?