Ethereum's Merge Completed Without a Hitch, GPUs Are Free

In the early hours of today, September 15, 2022, the Ethereum community stood breathless. At 2:43 AM EST, there were over 41,000 people viewing an "Ethereum Mainnet Merge Viewing Party" via YouTube. The reason: a software upgrade to the Ethereum Virtual Machine (EVM), known as The Merge, the second most important event in Ethereum's history barring its creation. After the first validator node was successfully brought online (with more following suit), Ethereum's Proof-of-Work (PoW) ceased, replaced with Proof-of-Stake (PoS). Gamers and GPUs rejoice!

The Ethereum Merge has been a long time coming ever since its proposal (in December 2020) as a possible upgrade to Ethereum. The Merge's main feature is to enable Ethereum to transition from the energy-intensive PoW method to the far less demanding PoS. In PoW, validators/miners employed the best graphics cards and perhaps a few ASICs to crunch the cryptocurrency's Dagger-Hashimoto algorithm, securing the blockchain in the process and ensuring transactions are processed correctly.

PoS does away with the computationally intensive security method. Instead, validators will have to show they have a stake in Ethereum's future by holding the equivalent of 32 ETH (~$50,615 at time of writing) in their node. Holding these 32 ETH units theoretically means that these validators have Ethereum's well-being on their minds, since working against it or "poisoning" transactions would likely eat into Ethereum's market perception and value, in turn reduce the value of their staked Ethereum tokens.

One concern that's been raised against the new PoS method is that centralized exchanges can participate in the staking process: users will be able to stake their ETH directly with services such as Coinbase. This has raised questions regarding the true decentralization of Ethereum. Institutions and law enforcement having greater power over centralized exchanges than they do on individual/organized miners has been one of the Merge's discussed aspects. Lido, a community-run validator collective, controls over 30% of the stake on Ethereum’s PoS chain. Coinbase, Kraken and Binance — three of the largest crypto exchanges — own another 30% of the network’s stake. That reads as a few key players being trusted with the keys to the kingdom.

That said, Ethereum's $60 billion ecosystem of cryptocurrency exchanges, lending companies, non-fungible token (NFT) marketplaces, and other apps are now supposedly more secure and scalable.

Perhaps more important for our readers and PC enthusiasts (granted, some of which likely did plenty of mining), The Merge and the PoS transition finally put an end to GPU mining on the Ethereum network, which has been online since July 2015. Speculation prior to The Merge was that this would lead to a flood of used graphics cards from AMD's RX 6000-series and Nvidia's RTX 30-series hitting secondary markets. The result is that graphics cards that are already selling at a discount, such as the $680 RX 6900 XT, could fall even further.

Until yesterday, the Ethereum network counted around 900 TH/s of primarily GPU-driven computing power. That's the equivalent of roughly 9.5 million RTX 3080 cards, but more likely a large mix of slower and older GPUs were also participating — meaning there was probably closer to 20 million GPUs involved with Ethereum mining, give or take. This doesn't necessarily mean that a mix of 20 million cards are going to suddenly hit the market, of course. If miners do choose to offload their graphics cards, though, Nvidia and AMD could see some difficulty in selling new graphics cards at retail.

At least some graphics cards have been put to work on alternate mining cryptocurrencies such as Ethereum Classic (whose hash rate has already doubled since The Merge, up to 158 Terahashes per second) and Ravencoin (also almost doubling from 8.9 TH/s up to 15,9 TH/s). But the more miners turn to these coins, the higher the mining difficulty imposed by the network, which will drive profits down. Cryptocurrencies using PoW generally follow a formula that adjusts the difficulty of the algorithm to control the flow of new coins into circulation. This means more people are competing for a limited resource, and if price doesn't surge in lockstep with hash rates, the profitability of mining will plummet.

Above, you can see the data from NiceHash's Mining Hardware page, which is still based on the pre-Merge values. We ran some quick tests, using NiceHashMiner (and NiceHash's QuickMiner) to see where things stand right now. Prior to The Merge, a GPU like the RTX 3090 could gross around $2.80 per day and the RTX 3080 sat around $2.30 per day, almost entirely thanks to Ethereum mining. Now? Oh boy, how things have changed.

An RTX 3090 running NiceHash Miner decided Autolykos with the most profitable choice for mining. At 245 MH/s, it was bringing in BTC equivalent to around $0.65 per day, while the PC consumed 400W (about 330W from the GPU). Ethereum Classic meanwhile ran at 120 MH/s and consumed the same 330W, potentially bringing in $0.33 per day. At a baseline estimate of $0.10 per kWh, that's $0.80 in power used by the GPU per day, and $0.96 for the entire PC, meaning every coin right now is well into the unprofitable range with NiceHash.

Here's the full set of NiceHash Miner benchmarks for the RTX 3090, running the latest version 3.1.0.0 of the software. The GPU was tuned for memory intensive workloads like Ethereum, however, so these results should only be taken as a rough baseline of what could be achieved.

What about direct mining? WhatToMine's RTX 3090 data suggests you could gross up to $1.35 per day with mining Ergo (ERG) mining, which uses the Autolykos algorithm. GPU power might be tunable to as little as 260W, which means you could potentially net $0.70 per day. That's still far less than half of what the RTX 3090 was doing prior to The Merge, and it remains to be seen if any coin can emerge from the collective with sustainable mining profitability on GPUs in a post-Ethereum world.

The abandonment of GPU mining also means that Ethereum is improving its energy efficiency by leaps and bounds. Since graphics cards no longer need to run complex computations to power and secure the network, Ethereum's energy consumption footprint (and carbon footprint) has been reduced by 99.9%, simultaneously cutting worldwide power consumption by 0.2% (which is still much less than the worldwide energy consumption of electronics left on standby, by the way).

Interestingly, no price-action occurred for Ethereum post-Merge, positive or negative. That's partly because most exchanges put a freeze on Ethereum trading while waiting for the network transition to take place. It's also possible that speculation had already made its way into the pricing over the past few weeks, especially since the last successful testnet for the Merge, Goerli, occurred little more than a month ago.

While the main Ethereum network has done away with GPU mining, existing communities of miners may still attempt to keep their cash-cow running. Several proposals to copy the Ethereum blockchain while keeping mining capability (also known as a hard fork of the network, which we've seen happen with Ethereum Classic) have gained some ground within the mining community. Making a new coin based on an existing coin isn't difficult; the real problem will be convincing the cryptocurrency users of the utility of such a coin, to give it some perceived value.