Is bitcoin’s energy use really excessive when compared with other currencies?

There’s been a lot of discussion lately about how much energy it takes to mint 1 Bitcoin. Most of the answers to this question involve estimates based on the efficiency of Bitcoin miners and the cost of electricity. In this post I take a different approach by examining the break-even energy value of minting the next Bitcoin and show that there is a very significant difference between the cost of minting a Bitcoin, and cost of using a Bitcoin. I want to extend this idea to other currencies and make the observation that Bitcoin is actually about as efficient as US dollars in terms of energy use needed to support the observed transaction volume.

This is where the Energy Parity Principle comes in. The current price of a Bitcoin is about 50,000 \$/BC. In most electricity markets where mining is affordable, the price of electricity is not more than 20 \$/MWh. Using this alone we can estimate the energy value of one Bitcoin must exceed 2,500 MWh/BC. For perspective, this is roughly the typical monthly energy use of 1000 US households. This seems like an astonishingly high energy marginal cost for a currency.

Let’s put this in perspective. Consider the breakeven cost of mining gold, which is basically the price of gold, e.g., 1,000 \$/oz. Assuming labor costs are small relative to other variable costs and capital costs are sunk so that most of that marginal cost of producing the next ounce of gold goes into the energy used to run the machines that mine gold. Here again, gold mining is usually done in places where energy is affordable given the efficiency of the mine, so let’s use the same cost of electricity, i.e., 20 \$/MWh. In that case the energy cost of gold is 50 MWh/oz. To have parity with Bitcoin, consider what it costs to mine \$50,000 worth of gold, i.e., 50 oz, under the same conditions as Bitcoin, and not surprisingly we come up with the same number: 2,500 MWh.

It looks like there might be a pattern here. Indeed, it doesn’t matter what the currency is, or even how it’s produced. If its value is directly tied to energy use, either directly in minting it, or indirectly by tying its production cost to energy use, then it will have the same breakeven energy value when created under the same energy cost conditions. This is the currency energy parity principle.

What’s the problem?

There are some important issues to consider. All viable currencies require scarcity to preclude inflationary behavior. In the absence of scarcity there is no limit on who or how much currency is in circulation and the value of a unit of currency would tend to zero. To ensure scarcity, some mechanism is required to present a barrier to freely minting new units. An energy-intensive process, like exclusivity of the sovereign authority, is just one way to induce the scarcity needed for a currency. In this sense Bitcoin is not as novel as one might think. Absent some intrinsic constraint on the supply of currency, the currency would fail in very fundamental ways unless some other mechanism were in place to put into effect some limit on the amount of currency in circulation. This is the problem solved by fiat currency when it replaced constrained gold and silver supplies as the basis for currency, and why fiat currencies only work when sovereignty is stable and secure.

The root problem with the energy-intensity argument against cryptocurrency is that it fails to account for the significant value in the transaction volume enabled by the currency. The utility of Bitcoin not its intrinsic value, but that it can be used to securely transfer something of value from one entity to another in consideration for something of value using digital infrastructure without a third-party settlement intermediary to address the double-spend problem. The question of how much it costs to create a new Bitcoin is not as important as the value of what is being transacted in the time interval during which the next Bitcoin is being mined to secure those transactions. At the current time, one Bitcoin is created every 10 minutes at an energy cost of about 2,500 MWh. During that time, approximately 500 BC are transacted, which is equivalent to about \$25M, or about 100 Wh/\$. That’s still a lot of energy, but it’s not nearly as scary looking as the result we got before.

In contrast, consider the daily US dollar transaction volume is about \$5T and is associated with about 100 TWh of energy use in the United States. This energy expenditure is necessary to support the value of the US dollar, without which the US economic collapses and the value of the Treasuries used to manage the fiat currency evaporates. This transaction volume’s energy demand is equivalent to about 20 Wh/\$, which is only about 5x the energy cost of transactions using Bitcoin. This seems to violate a would-be currency-energy parity principle. But consider that not all these transactions are actually in cash. In fact only about 25% of all transactions are in cash, and the balance are credit/debit cards which arguably are not subject to the same currency constraints. (These are constrained by the fees and transaction rate limits of the third-parties that provide the settlement services.) In this case, the energy cost of US dollar cash transactions is closer to 80 Wh/\$ and the currency-energy parity principle seems to apply, at least within a reasonable approximation.

Seen in terms of transaction volume, the energy cost of Bitcoin does not seem as high as one might expect. This difference can be easily justified by the relative benefits of using Bitcoin or US dollars cash, i.e., speed, digital, anonymity, international transactions, notwithstanding the restrictions, uncertainties, losses, niche applications, and general user skepticism associated with Bitcoin in particular, and cryptocurrencies more broadly.

Compare these currencies with gold, which is used for about \$11B of transactions per day. As we saw above, the energy breakeven energy is the same as Bitcoin, but the energy cost of gold transactions is about 220 kWh/$, which is much higher than US dollars or Bitcoin. This makes sense because transactions in gold are more often associated with trades for investment instruments or hedges against other asset volatilities. Gold transactions are much less frequent and therefore can afford to be significantly less efficient than cash or cryptocurrency transactions.