I have discussed about the various demand drivers for ether on Twitter and Reddit many times before. There was never enough material to expand it into a blog post, but I have found a thought experiment that may be intriguing - how we can estimate what ETH’s demand profile looks like, i.e. why people are buying ETH, and how much for each reason. Unfortunately, I have zero programming skills and have no idea how to analyse on-chain data, but perhaps others can. Looking at you, Data Always! It is important to note that it’s not just Ethereum L1, but also L2s, sidechains, alt-L1s - all chains with ETH bridged count. There’s a fair bit of forensics involved, and it’ll never be perfect - but that’s alright we’re only estimating. There are many nuances here, and I’m simplifying it to 10 categories, leaving out some niche ones. Feedback welcome, as always! Here goes the list, in descending order of quality, in my subjective opinion out of 10:
Long-term reserve-asset (10): these are the people who have been holding ETH for X years and consider it their long-term store-of-value. As for what X is, it’ll be revealed on further analysis - I expect to see a sharp drop-off at a certain time duration. I believe Glassnode has a similar method for their LTH metrics. Of course, need to account for multiple addresses etc. What about CEX cold wallets? We may have to proportion those based on other findings.
Long-term stakers (10): Likewise, divide stakers by their time horizons. I’ll explain later while long-term stakers are significantly higher quality than short-term stakers. Of course, we’ll only be able to assess this in the months after withdrawals are enabled.
Economic collateral (9): Just look at all the ETH locked in DeFi protocols, or elsewhere. No issues if they are switching between protocols frequently, with a caveat - more below.
Unit-of-account (8): ETH changing hands for trading NFTs, ERC20s etc. paired against ETH. There may be spikes in mania markets, but consider the baseline.
Medium of exchange (5): This one’s challenging, because you have to look at people who are actually using ETH as currency. I also suspect it’s a pretty minor contributor, so we can afford a wider uncertainty range.
Speculation (5): Speculators play a key role, and is higher quality than I sometimes give them credit for. Short-term holders (STH, per Glassnode) will qualify, as will CEX warm wallets, and some proportion of their cold wallets, estimated from LTH/STH ratios.
L1 transaction fee burns (3): Gas fees burned on Ethereum L1. Aside from mania markets, not that much ETH is burned. It’ll be the financial hub for the cryptoeconomy for the foreseeable future, so it’ll remain a significant contributor. However, there are scalability upgrades planned for L1 in the long term - statelessness and zkEVM being the two massive ones - which will make this a less important demand vector over time. Why not priority fees and MEV? For simplicity, those can be considered direct demand drivers for staking, which leads to demand for ETH.
Speculative farming (3): Basically farmers who are constantly on the move, using their ETH to primarily collect and sell token incentives. While they do offer some quality to the ecosystem in terms of bootstrapping economic bandwidth, they are ultimately low quality. Note: there may be some overlap with this category and long-term holders, so we need to account for that.
Short-term stakers (2): Likewise, there’ll be short-term stakers who will be chasing yield in mania markets when MEV+priority fee rewards greatly exaggerate APRs, those who are most likely exit when the inevitable bear market unwind comes. We will probably need a few years to establish this.
L2 fee burns (2): Today, L2s make up between 2%-5% of total ETH fees burned, and it may continue to increase leading up to EIP-4844. However, after EIP-4844, L2s will get their dedicated space to settle data with its own fee market. As a result, this will drop significantly. Furthermore, this data layer has plans to be highly scalable, with techniques like expanded 4844 and eventually danksharding. Because data is a much more trivially scalable resource than execution on L2s, I expect this to be a relatively minor contributor in the long term, but it’ll still be worth noting if L2s are adopted en masse. Needless to say, we’ll need to wait for EIP-4844 to assess this. Stuff like bridging will still require L1 fees. Cross-L2 MEV will still be a demand driver for staking as far as they relate to L1 MEV, as I expect L1 builders to be integrated into L2 sequencing. L2 MEV will accrue to L2s.
My wish is to see a simple pie-chart that shows how exactly ETH is being used. If this interests you and you’d like to work on something to quantify these - please get in touch, tag me on Twitter! (I can’t keep up with DM requests, so please bear with me)