De Vries estimates that the Bitcoin network consumes “at least 2.55 gigawatts of electricity currently, and potentially 7.67 gigawatts in the future.”
“Economic models tell us that Bitcoin’s electricity consumption will gravitate toward the latter number,” comparable with Austria, which consumes 8.2 gigawatts, de Vries writes. “A look at Bitcoin miner production estimates suggests that this number could already be reached in 2018.”
De Vries is not alone in warning about Bitcoin’s appetite for gigawatts. The rise in popularity of Bitcoin and other cryptocurrencies has been met with growing concern about the energy usage required by the thousands of computing systems that power these virtual currencies.
Producing each Bitcoin is called “mining,” and it requires a fair amount of computer processing power to verify the transactions that create the coin as well as add them to the public ledger.
Various studies have attempted to quantify the energy used in bitcoin mining with relatively different results. If there’s two things all parties agree on, it’s that Bitcoin mining is energy intensive, and that it’s only expected to increase.
But there are a few problems with de Vries’ underlying assumptions, according to Koomey and other experts. They point to how de Vries calculates the value of the energy used in Bitcoin mining and the price paid for that energy.
“The worry is that those are two numbers that are picked out of the air,” Koomey said. “There may be some basis for them, but it’s a very unreliable way to do these kinds of calculations, and nobody who does this for a living would do it like that. It’s odd that someone would.”
The other problem is that much of that mining doesn’t happen out in the open — in fact, cryptocurrency miners tend to value their privacy — which means that the energy consumption data necessary to come to any firm conclusions is far from readily available.
De Vries could not be reached for comment for this article, but he said in a news release that his work does require some less-than-rigorous math and stressed the need for more research.
“We’ve seen a lot of back-of-the-envelope calculations, but we need more scientific discussion on where this network is headed,” said de Vries, who works at the Experience Center of PricewaterhouseCoopers in Amsterdam and founded the blog Digiconomist. “Right now, the information available is pretty poor quality overall, so I’m hoping that people will use this paper as a foundation for more research.”
Christian Catalini, an assistant professor at MIT’s Sloan School of Management who researches cryptocurrencies and block chain technology, also said researchers need data directly from Bitcoin miners.
“The only people with good data on this are the miners, so you do need that data to understand the electricity consumption,” Catalini said. “The main challenge is that this gear is scattered across the globe and faces different prices.”
And it’s not as though the servers that the miners use are particularly easy to keep track of, either.
“The equipment they’re using is customized to their use,” Koomey noted. “This is not the standard servers that are being sold. These are machines that are assembled using other components. That makes it more difficult. It’s a custom design and those kind of servers are not tracked as well.”
Some experts are also troubled that the de Vries model concludes that the annual energy consumption of Bitcoin miners would exceed that of major data companies like Facebook and Amazon.
“This debate keeps popping up, but it would be great if someone did some data sharing with the miners and got some good estimates,” Catalini said.
To Koomey, Bitcoin enthusiasts and detractors need to be a little more patient and wait for better research to come out — particularly a study based on conclusive data.
“We should try to get the numbers right, but these things don’t happen super fast,” Koomey said with a laugh. “It takes money to do research and a lot of time.”