Bitcoin miners now use as much power as Massachusetts — and will need more

Bitcoin miners now use as much power as the state of Massachusetts. But they are very profitable and, in the near term, they will probably need more.

Let’s consider why. The energy consumed in Bitcoin mining protects the Bitcoin blockchain against hacking, it’s a deflector shield if you will, a feature of the protocol which secures the integrity of the distributed ledger.

As bitcoin grows in value, the rewards for mining become more lucrative which attracts more capacity to the system and increases the energy consumption. Conversely, if the price of bitcoin were to fall, mining capacity would contract to reflect the lower economic value of the Bitcoin blockchain, and so would the power consumed.

From time to time concerns are expressed about the amount of energy that mining consumes. Many economic activities consume substantial amounts of energy, and the amounts Bitcoin mining consumes reflects its economic importance, currently $200bn in terms of market cap. This is significantly smaller than, say, the balance sheet of JP Morgan ($2.5tn) or annual worldwide oil production ($1.5tn). Because of the design of Bitcoin, energy is a relatively big factor of production, comparable perhaps to salaries for JP Morgan or drilling gear in the oil industry. There is no more need to worry about the energy consumed by Bitcoin mining than there is to have concerns about the number of people employed at JP Morgan.

Nevertheless, it is interesting to estimate how much energy Bitcoin mining consumes. Cryptographic mining, or consensus by proof-of-work, is a completely new form of economic activity, it did not exist before Bitcoin. The overall increase in energy consumption is significant but so are the advances in mining technology. Mining has migrated from inefficient CPUs to GPUs to energy efficient ASICs and now pushes the envelope of what is possible in silicon.

The overall Bitcoin mining hashrate now stands at 12,000 petahash per second. Assuming only the most efficient processors are used, comparable to Antminer S9, the total power consumption of Bitcoin mining now stands about 1.2 GW. If only the least efficient processors are used, ie, those that barely break even at current prices,the total power consumption of Bitcoin mining is about 21 GW.

The latter estimate assumes miners pay competitive market prices for co-location and power, the actual number probably lies somewhere in between the upper and lower figure since miners deploy a range of systems and do not all enjoy the same commercial terms for co-location and power.

For comparison, the state of Massachusetts has maximum electricity generation capacity of 12GW (see www.eia.gov). This is approximately the median value between the lower and upper bound of our estimate. There is now way of knowing the actual total power consumed by miners for sure, but it would not be surprising if it was comparable to the total electric power of the state of Massachusetts.

Miners spend far less on energy costs than they collect in new bitcoins and transaction fees. Annualized total mining revenues at $12k/BTC are about $8bn, but the annual operational cost of 12,000 petahash using the most efficient mining chip is about $1bn which makes mining very profitable, even after the next block reward halving. It is therefore quite likely that mining capacity will continue to grow until margins shrink to lower levels.

(To those who wish to analyze the profitability of prospective mining deployments, a financial model including a reference calculation is available at https://github.com/sweyn/bitcoin-mining-profitability.)
 
We still enjoy the benefits of Moore’s Law, processors are becoming more efficient every year. At the same time, the energy industry is seeing rapid advances in solar power generation and in battery storage. These trends combined should make Bitcoin mining more energy efficient and less costly in the future. With each advance in processors, power storage and generation, the security of the Bitcoin blockchain will improve as will the efficiency of the miners. As for deflector shields, their greatest utility is in protecting centralized structures, and they won’t be necessary to secure distributed ledgers if and when they emerge!