The numbers in this study are very back-of-the-envelope and assume a worst case: widespread adoption of Bitcoin and not much improvement in Bitcoin mining activity, along with long replacement cycles for older, less efficient mining rigs.
Blockchain isn’t the only distributed ledger system, but it is the most power-hungry one. Other systems, like Merkle trees, can handle global-scale transaction ledgers on a couple of blade servers.
The results show that in an optimistic scenario, the increase in electricity consumption of the bitcoin network compared to now is not shocking, from around 350 MW to around 417 MW, but still on the order of one small power station. If things play out a little less favorably, however, the bitcoin network may draw over 14 Gigawatts of electricity by 2020, equivalent to the total power generation capacity of a small country, like Denmark for example.
Even in the optimistic scenario, just mining one bitcoin in 2020 would require a shocking 5,500 kWh, or about half the annual electricity consumption of an American household. And even if we assume that by that time only half of that electricity is generated by fossil fuels, still over 4,000 kg of carbon dioxide would be emitted per bitcoin mined. It makes you wonder whether bitcoin could still be called a virtual currency, when the physical effects could become so tangible.
Source: Sebastiaan Deetman @ Motherboard