One of the most common worries customers express in our mining pool's tickets is the future of GPU mining after Ethereum merges to proof-of-stake. Naïve articles in mainstream media often portray the end of ETH proof-of-work mining as a cataclysmic event that will bankrupt everyone and destroy the value of GPUs for eternity. That couldn't be farther from the truth, because Ethereum Classic has been, and continues to be, ready to expand the ETH universe.
Timing of the merge
Currently, the Ethereum developers project that mining will end on September 19. Given how many times this deadline has been delayed, meeting it is unlikely. However, because the two testnet merges that have occurred so far have merged successfully, I suggest there is reason to believe that the developers will indeed get the merge off successfully - but probably not until the end of the year. Vitalik Buterin has been extremely conservative with his management of the network, which has to date probably harmed Ethereum by focusing on perfection rather than ironing out the expected few bugs after release, as other coins often do.
If we believe that the merge occurs three months late, around December 19, that gives about 4 or 5 months until ethash mining becomes less profitable because Ethereum mining stops.
ETH mining before the merge
Until the merge, two trends will persist. The first is that the price of ETH will rise in anticipation of the upgrade, and the second is that people will unload graphics cards on eBay. As a result of the GPU fire sale, ETH's hashrate will not rise along with price, and GPU mining is therefore likely to become extremely profitable over the next few months. Don't be surprised to see profitability triple between now and the day before the merge.
Understanding resource usage with coins
I've stated frequently that every excess resource in the world will eventually be used to secure a coin. If you have spare CPUs, you can mine Monero, if you have spare GPUs, you can currently mine ethash, if you have spare disks, you can farm Chia, and so on. Eventually, someone will even figure out a way to use unemployed humans to do calculations to secure a coin. The reason every spare resource seems to get sucked up into cryptocurrency mining is simple: a secure way to transfer money is valuable. We know it is valuable because people will pay for it by buying those coins and in the transaction fees to send money in those networks. Once one of these categories of work is invented, it never just goes away; while Chia has declined in value, it is still popular and continues to function. Specific coins often decline, but other coins using the same type of work always replace them.
Why are these coins valuable? It's largely because the bitcoin developers limited the size of blocks to 1MB. Litecoin doesn't need to do anything more than bitcoin does; it's extremely useful for sending money between exchanges simply because it has the transaction capacity to do so. At times, it is cheaper to sell bitcoins at one exchange, use litecoin to send money to another exchange, and then buy them back there than it is to spend money on bitcoin network fees.
Over time, it's become clear that there is a market for two types of networks: those that just send money, and those that can execute contracts. Bitcoin is the largest "money" coin, and Ethereum is the largest "contract" coin. Ethereum, like bitcoin, is primarily limited by its blocksize, which causes high gas fees. In fact, sending money on Ethereum costs much more than sending money on bitcoin most of the time. There's a significant market for a "contract" coin that has cheaper fees. Several networks, like Solana, currently attempt to fill that niche, with varying degrees of success.
When the Merge occurs, there will be a huge number of GPUs that are available to secure some network, and the demand for a cheap and secure smart contract network will exist. The next question is: which network will that be?
"Useful work" tokens
There are several "useful work" tokens that have come about in recent times. Livepeer allows miners to be paid for transcoding video, and RNDR token pays miners for rendering video, like animated movies. The problem with useful work is that, strangely, there is less useful work to go around than contrived hashing work.
Livepeer, for one, appears to be limited by the amount of video there is to transcode. Its network stats show that usage increased dramatically for the first few months, but the daily payouts have stalled at an upper bound of around $5,000 per day, which isn't anywhere near enough demand to require a large number of GPUs.
I don't believe that useful work is the future of GPU mining because there isn't enough useful work to do.
The three "minor coins"
Currently, Ethereum mining represents about 94% of the total value of all GPU-mineable coins. A good (but outdated) summary of all the coins is available in a table at https://bitproit.com/gpu-mining-profitability-after-ethereum-merge/. Note that I don't agree whatsoever with the conclusions of that article because I believe the author has a shallow understanding of mining, but the percentages of marketshare were correct at the time it was written and are useful for this analysis.
The coins from that list to pay attention to are the top four other than ETH: Ethereum Classic, ERGO, Ravencoin, and Conflux. Let's focus on the three others than Ethereum Classic first.
I don't want to get into the specifics of how each of these coins works, because that would bog down the article. For example, Ravencoin is a coin that also includes a sort of smart contract platform like Ethereum, and the reasons why each was designed the way it was could require many paragraphs. The key piece to understand is that each of these three coins runs its own algorithm, with custom code needed to connect to each one. Mining software needs to be configured to receive unique data, mine data a different way, and return uniquely formatted responses for each one. If one mining project, like gminer, eventually supports all of these coins, it will basically be through "if" statements selecting what code to run based upon what the user selected at startup. Even though there are similarities in how they are mined, like how Conflux shares daemon commands with Ethereum, it's not possible to directly switch between them like Proswitching can with scrypt coins.
Eventually, after much optimization, each of these algorithms will have preferred GPU models, like how AMD cards might perform better on one algorithm and nVIDIA cards hash better on another algorithm. The lesson to take from this section is that if you didn't already think it's a mistake to panic sell all your graphics cards, all cards are not created equally. We don't know yet which models and brands of cards will be more valuable because all the optimization focus to date has been on ethash, not these other algorithms. Therefore, there will probably be a niche for specific card models on these coins, even if they aren't the dominant mineable coins.
But these three coins have a key shortcoming: they aren't Ethereum.
Ethereum Classic - the "major coin"
Ethereum Classic isn't Ethereum, but it's identical in every other significant way other than name and proof-of-work method, which is why it is the obvious choice to take over as the primary GPU-mineable algorithm.
After the Merge, Ethereum isn't going to magically become more capable, or have new features, or have more capacity. Those things require sharding, which is at least two years away in "Ethereum developer time," which likely means more in time as you or I define it. Meanwhile, gas prices will continue to trend upward and new DeFi protocols and smart contracts will come about. Currently, the solution to the gas problem is to move those contracts to Solana or Binance Chain, two highly-centralized platforms with lower fees. The centralization of Solana, in particular, has led to network-wide outages that have lasted entire days. Tether, for example, has representations of its token on multiple chains for use when ETH gas prices are too high.
There's a key difference between Solana and Ethereum Classic, though - one requires developers to write code for a different client, and the other uses the same Solidity language that Ethereum uses. An Ethereum contract can be dropped into ETC with minimal changes, even after the Merge.
ETC's previous shortcomings
One might wonder, then, why Ethereum Classic never achieved widespread usage like Ethereum has. One older reason had to do with the acrimonious split between ETH and ETC over the failure of The DAO, but six years has passed since then, and given how many other DAOs have followed, many don't even recall the incident.
The more recent problem is 51% attacks allowed by cloud mining rental provider Nicehash. I wrote about these attacks when they happened in another blog post, where I disproved Nicehash's false statement that the company could not prevent such attacks, which were conducted by allowing criminals to rent mining rigs from their service. The rented mining rigs were used to mine a "private blockchain" of Ethereum Classic, which was then suddenly released to the network, reorganizing the public chain and stealing money from exchanges. (In response to the repeated attacks against Ethereum Classic, all of which used Nicehash's services, Nicehash recommended that coin developers rent hashrate from Nicehash to fight against the criminals who were also renting hashrate from them.)
As attacks against ETC continued to occur, exchanges began to recognize Nicehash as a threat to their businesses and began requiring ridiculously long confirmation times for ETC - sometimes as long as 10,000 blocks, or 120,000 seconds. By comparison, some of these exchanges require just one block, or ten minutes, for bitcoin.
ETC is currently in a position where the only thing it needs to succeed is security. Security is provided by increased hashrate.
The cycle of ETC's growth
Here's how the path to ETC's growth will work. First, even though I don't think this step is necessary, mining pools that want to ensure a future for mining will make big investments in ETC. Antpool has already done this, with a $10m investment in ETC development. Digital Currency Group/Genesis/Greyscale has a significant investment in ETC with its stock-traded fund and I would expect them to also make investments.
Next, as is occurring now, people are speculating that what I'm describing here could happen, and are purchasing ETC. The increase in ETC price encourages a few miners to mine it. This increase in price will continue until the merge, when ETC likely rises from its current $40 price to $1,000. If ETC then settles $600, its profitability will be about half what ETH's was at the low point of this cycle, as ETC's block reward of 2.56 is higher than ETH's current reward of 2. Assuming the likely case that other coins also appreciate at the same time, ETC would be coin #6 (non-stablecoin #4,) behind Binance Smart Chain and ahead of Ripple, with a market capitalization of around $50b. If ETH hits $4,800 and then settles at $3000 after the Merge, that would be a 6:1 ETH/ETC ratio, which seems reasonable, especially given that 15% of ETC's supply is effectively unusable due to the DAO hack in 2016. I would expect this ratio to close even further after that first blowoff top, until sharding actually increases Ethereum's capacity dramatically.
One the price of ETC rises, hashrate will follow, even before the Ethereum Merge. Ethereum Classic's algorithm is etchash, a derivative of ethash that is possible for some ethash ASICs that became obsolete due to ethash's DAG size, like the E3, to mine. Since ETC started this run at $9 and is currently only worth $40, hashrate would be expected to increase by at least 50x from the lowpoint of the cycle. The actual increase will be higher, because a larger number of GPUs will be willing to mine at lower profitability levels because there are fewer coins to mine.
Once the hashrate of the network has increased, it will no longer be possible for Nicehash to enable attacks against the network using its services. Exchanges will reduce the number of confirmations required to receive ETC, and smart contract creators will be more confident that transactions will be unable to be reversed.
The final stage of the process is that after the network is secure and widely supported, developers and users will notice that gas prices on ETC are much lower than they are on ETH. The availability of ETC will roughly double the capacity of ETH-based smart contracts, so companies like Circle and Tether would likely deploy representations of their stablecoins on ETC for customers who want to send smaller amounts. Other contracts will duplicate or move quickly, because the two networks are completely code-compatible. In many cases, it is even possible to use the same addresses on both networks.
End result
In the above scenario, the amount of money being mined on Ethereum Classic would be about 40% of the current Ethereum network's mining payouts - one third based on price, and then a little more because ETC's block reward is higher. If the other three "minor coins" increase in price as a result of improved security, that means that perhaps 45% of GPUs would mine at the current profitability. But since many GPUs today are already making significantly more than breakeven, I suggest it's reasonable to expect that many of these GPUs that are flooding eBay will be purchased at firesale prices and redeployed, resulting in perhaps 50% of current hashrate being viable shortly after the Merge.
Therefore, rather than being a catastrophic event where every GPU becomes suddenly worthless, the Merge will be more similar to the hundreds of previous events when coins changed algorithms or were delisted from all exchanges. In those instances, as in the current instance, the big losers will be older equipment that can no longer compete with more power-efficient cards. Specifically, 2000-series nVIDIA cards will not be usable for mining after the merge.
Finally, as always occurs during the uptrend part of the cycles, there will be a period when seemingly obsolete equipment will suddenly become profitable again. This effect has always occurred during litecoin halving cycles; in the last cycle, some ASICs that were manufactured in 2013 actually became extremely profitable for mining litecoins and dogecoins for several months before fading back into worthlessness. If, when sharding is nearing release in 2025 or so, Ethereum surpasses $10,000 and drags ETC with it to the $2,000 range, even those 2000-series cards would be profitable again.
Conclusion
GPU mining isn't going to disappear catastrophically after the Ethereum 2.0 Merge occurs. Network security is valuable, which is why most computing resources are now used to secure one coin or another.
There are several coins that are candidates for GPU mining, but Ethereum Classic is the obvious winner. ETC is plug-and-play with existing ETH smart contracts, and the Merge isn't going to increase ETH's capacity. Gas fees on ETC are close to zero, and the only problem holding ETC back at present is 51% attacks allowed by Nicehash. Increased hashrate will eliminate this problem. Rather than requiring reliance on more centralized networks that require reprogramming like ADA or SOL, an attractive option for developers will be to simply use Ethereum Classic.
There won't be room for every miner, and lower-end GPUs will end up unprofitable, as always happens during coin transitions. But fears of a complete collapse are unfounded, and the next GPU mineable coin has been right here all along - it's Ethereum Classic.
Comments (1)
Archived discussion. New comments and replies are closed.