Understanding Bitcoin Energy Consumption - Interview with Alex de Vries of Digiconomist.net

Martin C. W. Walker
5 min readMar 12, 2021


Alex de Vries is the founder of the Digiconomist website which since 2014 has analysed the environmental impact of cryptocurrencies. He is an internationally recognised expert in this area and has been quoted by Bloomberg, CNBC, the LSE Business Review, The Daily Telegraph, Money Week, the BBC, The Guardian, CNN, The Independent, Newsweek, The Washington Post and many others. His background in financial economics with many years of experience in data and risk analysis at some of the leading institutions in banking and accounting.

Q. Alex how did you become interested in the environmental impact cryptocurrencies and why do think it is an important topic to study?

I first read an article in 2015 that estimated the energy use for a single Bitcoin transaction could power 1 US household for 1.5 days, and I remember thinking “that’s insane — why isn’t anyone else talking about this.” I decided to dive into the topic, and realized there wasn’t any good information or heavily outdated at best. I subsequently created the Bitcoin Energy Consumption Index, as I feel that if we managed to create the world’s most inefficient system people should know about it. Especially since the end users of Bitcoin typically aren’t confronted with those environmental costs. It’s the miners that pay the electricity bills.

Q. How you estimate the energy consumption and CO2 emissions resulting from cryptocurrencies?

I generally use a combination of economic theory and information on miner income versus miner expenses. We know how much miners are earning, we got a good idea how much they’re paying for electricity, and we can estimate their electricity cost share. Nowadays we can also estimate where these miners are located, which makes it possible to attach a carbon intensity to any energy estimate (you need to know where their energy is being sourced from).

Alternatively, you can also estimate the total computational power in the network, and attach an energy efficiency per unit of computational power per unit of energy consumed based on a selection of available devices. Cambridge University follows this approach, and I use it myself to at least put a minimum bound on the energy estimate (you can’t do better than the most efficient machine out there). Both approaches have their own drawbacks, but I found an economic model more appealing since it has predictive value (see latest study publishing Mar 10).

Q. How do you get the data needed for your models and how much certainty can we place in the data?

The models are pretty generic so you can source almost anything you need from websites like blockchain.com. The certainty of the outcome depends on what you’re calculating. E.g., the network’s lower bound is 100% certain, but the real number is also guaranteed to be higher. Any estimate that gives a precise number is sadly almost unverifiable. There’s just no way of knowing how much energy the network is consuming today. Last year I looked into IPO data that contained information on almost all Bitcoin miners that had ever been produced up to the date considered, and found that both my BECI and Cambridge’s CBECI were still on the conservative side. Unfortunately, this type of information is extremely rare and only available for certain dates, but at least allows us to have confidence in the numbers being produced.

Q. There has been a great deal of criticism of your work from cryptocurrency investors. Has any of this criticism uncovered errors in either your data or your models?

There are many known limitations to the existing models; I don’t think that classifies as errors. e.g., both me and Cambridge assume rational agents. Even so, I observed (study above) that Bitcoin miners were being sold below their production cost back in 2019. That led me to overestimate the share of hardware costs in the total mining costs. You just cannot account for that type of behaviour. Another limitation is that both me and Cambridge assume a fixed price of electricity throughout the year, while we know this number varies seasonally. There’s no good way to model it, as it also depends on trends (e.g., Iran gaining popularity), so that adds to the numbers generally being on the conservative side.

Q. Many cryptocurrency enthusiasts claim the mining process is good for the environment because it encourages research on more efficient forms of energy generation and mostly uses renewable energy. What does the evidence show about these claims?

That’s an easy one; there’s been no support for that in independent research. If you survey miners, at best they will indicate they their energy is 39% renewable. That’s coming directly from miners, and still a minority. Meanwhile, the energy used for Bitcoin mining on average still has the same carbon intensity as if the whole network was running on natural gas (480–500 gCO2/kWh). Bitcoin mining isn’t clean at all, and there’s no way around it.

Q. Another claim they frequently make is that cryptocurrencies are no more wasteful that the conventional financial system?

Given that a single Bitcoin transaction has the same carbon footprint as 735,000 VISA transactions that’s somewhat hard to follow. Bitcoin mining will soon consume as much energy as all data centers globally (see latest study), with only a fraction of the utility. Wasting energy was part of Bitcoin’s core design, so this is the logical outcome of that.

Q. What is the impact of rising cryptocurrency prices on energy consumption and Co2 emissions?

The more income miners generate, the more they will spend on hardware and electricity to mine. It’s a very simple relationship; as long as it’s profitable to, miners will continue adding more devices to the network. That also means more energy consumption and related emissions.

Q. What do you think regulators and politicians should do about the environmental problems of cryptocurrencies?

Bitcoin is said to be a decentralized currency, but suffers from massive points of centralization within its ecosystem. The Bitcoin mining device production supply chain contains less than a handful of companies, and those miners subsequently all try to get access to the same sources of energy (so they all flock to the same locations). This makes them a pretty easy target; just like you can easily target Bitcoin itself by cutting it off from on-ramps. We’ve just seen Inner Mongolia using this to ban mining from the province. You cannot stay under the radar with 20,000 or more machines.