Ever since the industrialization of Bitcoin mining in 2015 we've heard concerns about the concentration of hashpower within the borders of China. Why is hashpower concentrated there? For one, most of the world's semiconductor production facilities are in Asia and thus most of the mining chips are produced there. But perhaps more importantly, China has an abundance of cheap energy.
While there are dozens of countries with incredibly cheap electricity, in general they are not sufficiently industrialized or politically/economically stable enough to support large scale mining operations. China is one of the few countries with the combination of infrastructure and energy abundance.
Much of this mining power operates at 2.5-3c / kWh in Inner Mongolia during the dry season via cheap coal and wind power, but during flood season, these miners migrate significant capacity to Sichuan and Yunnan provinces to take advantage of 1c/kWh hydroelectric power. Quite simply, there is more power generated than there is demand for consumption, which results in very low prices.
Hashpower Distribution Estimates
The Cambridge Centre for Alternative Finance estimates that in Q1 of 2020 70% of the total network hashrate was in China. Note that these metrics are based upon several assumptions and their accuracy can't be verified.
Though the hashrate is spread throughout the provinces of China (not evenly of course)
A CoinShares Research report from December 2019 placed 65% of global hashrate in China.
In July 2020 a study that was commissioned by the Fidelity Center for Applied Technology released a report that estimated closer to 50% of global mining power capacity is likely in China. However, they noted that mining is a secretive industry and they believed they were only able to identify 15% of Chinese mining sites.
Why do people get concerned about more than half of hashpower being located in China? Because of the infamous "51% attack" against which Bitcoin is susceptible. The interesting thing about 51% attacks is that they don't break any of the rules of machine consensus. So what are the ramifications of pulling off a 51% attack? It somewhat depends upon the motives of the attacker.
Despite its scary-sounding name, a 51% attacker is limited in what they can do:
- They can't steal people's bitcoin arbitrarily
- They can only actually double spend their own bitcoin
- They can't change the consensus rules
- They can't make invalid transactions become valid
If a 51% attacker is seeking to maximize their profit then the juiciest target for a double spend is going to be an exchange. Mine a bunch of coins, send them to an exchange, convert them to a different censorship resistant cryptocurrency or stablecoin, withdraw those funds, then release a bunch of blocks that were mined in secret that send your original UTXOs that were spent in a deposit to the exchange back to your own wallet. The downsides to such an attack:
- Any exchange with decent liquidity to make them attack-worthy will likely have withdrawal limits
- Similarly, most such exchanges will require AML/KYC and thus you'd also need to compromise a verified exchange account with high limits
- The value of the bitcoin you still hold after the attack will likely have decreased substantially, thus a successful large attack could actually result in shooting yourself in the foot.
- You'd better not slip up while you're accessing the exchange you target. For example, one hacker returned $25M in stolen funds after leaking their IP address.
It's certainly safer to just use your hashpower to profit from securing the network rather than attacking it. But what if the attacker doesn't care about profitability? What if the attacker is a nation state that prints their own money?
A Nation State Adversarial Miner Attack
There are probably a thousand+ mining farms scattered throughout China; it would take some effort for the government to seize them all. I suspect it would be nearly impossible for the state to start seizing control of mining facilities without the news leaking to the rest of the world. If we heard of such activity taking place, you can be sure that Bitcoin stakeholders would start planning emergency actions.
An easier attack vector would be against mining pools; at time of writing it appears that over 70% of hashpower is being coordinated through fewer than 10 mining pools that are located in China.
Of course it's far more feasible for a nation state to identify and attack 10 targets within its borders simultaneously than it is for them to attack a thousand simultaneously.
On the flip side, switching mining pools is incredibly easy for miners. Once again it becomes a question of being able to perform an attack covertly. Given how many independent entities are observing activity on the Bitcoin network, it's practically guaranteed that within a matter of minutes an alert would be sounded and miners would start looking into taking action against malicious actors. At time of writing we even have Twitter bots that put out alerts about any orphaned blocks.
It's hard to imagine a scenario in which a state actor would be able to quickly and covertly seize enough hashpower to perform an ongoing attack that lasts more than a few hours. A worst case scenario in which state actors did seize all of the physical equipment could result in China only mining empty blocks and orphaning other blocks that did contain transactions, essentially halting all transaction confirmations on the network. Then it would become a game of either patiently waiting for them to give up or coordinating a code change that would make their hashing machines worthless.
Point being, any large-scale mining attack is going to be limited in its effectiveness for a variety of reasons and will be unlikely to disrupt network operations for more than a short period of time.
Matt Corallo designed Betterhash to further decentralize miner control away from pools. It looks like the first real implementation of this idea is occurring in Stratum V2.
Allowing miners to choose their own transaction sets moves some power from mining pools further downstream to the miners themselves, thereby increasing the censorship resistance of Bitcoin. This idea was originally put forth by Matt Corallo in BetterHash, and we felt it was very important to include in Stratum V2 as well because it has a meaningful impact on Bitcoin’s decentralization.
Over the very long term I expect we will see semiconducter foundries outside of Asia begin producing more mining chips and countries with even cheaper power sources will continue to become more industrialized, thus providing more competition when miners are seeking out new locations to set up shop. China's mining dominance is unlikely to last; I expect that this theoretical attack will become less and less likely.