Disclaimer: This is not financial advice. Anything stated in this article is for informational purposes only and should not be relied upon as a basis for investment decisions. Triton may maintain positions in any of the assets or projects discussed on this website.
TL;DR
In light of the recent spot Ethereum ETF launches in the US, we double click on the underlying economics of Ethereum and how that dynamic is impacted by its modular rollup-centric roadmap.
The Headline Numbers
Ethereum is the second most valuable network behind Bitcoin, with a current market capitalization of ~$325B and another $100B of capital locked in contracts on top of it. Per Token Terminal, over the past 30 days it has facilitated almost $500B in trading volume, the native token Ether is held by 127M unique addresses, sees roughly 7M MAU generating $100M per month in fees (over $1B per year annualized), ultimately accruing over $600M per year in ‘revenue’ to token holders via burns. In its history, the network has generated almost $20B in fees and almost $13B in revenue to token holders. This growth hasn’t been free – networks fund growth through token issuance to miners or stakers, and Ethereum has issued a cumulative $33B in token incentives over its lifetime. For a completely open-source, crowd-funded project that is less than 10 years old, the magnitude of these numbers is impressive. In this post, we go deeper into all of this to get a better understanding of the dynamics at play and to put these numbers in better context.
The High-Level Model
Let’s start by delineating the economic model for Ethereum. At the absolute highest level, Ethereum generates fees when users, either directly or through use of applications or L2s, pay for transactions to be executed and settled on the network. In terms of expenses, Ethereum splits those fees earned with validators while also ‘paying’ for network validation and security by issuing token rewards to validators.
Fees, Please
Diving deeper, how does this actually look in practice?
Ethereum is a transaction-based state machine and as such is driven by users initiating operations (e.g. sending ETH to another address or creating a new smart contract) to be taken by the network that updates its state (e.g. adjusting address balances). The network bundles a set of these user operations into a block (hence ‘blockchain’) which are then submitted to the network to direct what the next state of the network should be. For example, if user A sends user B 1 ETH, the next state of the network should show user A with one less ETH and user B with 1 more ETH. A new block on Ethereum is posted every ~12 seconds. Any user around the world can submit an operation to be included at any time but not all operations can be included in every block – there is an upper limit on block size as determined by gas limits to avoid undesired actions like spam or infinite loops.
‘Gas’ is exactly what it sounds like: it is the resource required to run the network and is the base unit that measures how computationally expensive a certain operation on the network is. Different operations naturally have different computational requirements, so rather than limiting block capacity by a simple number of operations, block capacity is theoretically limited by the computational difficulty of executing that block (e.g. execution time). A simple ETH transfer requires a gas limit of 21,000 units while creating a contract requires a minimum of 32,000 and can rise quickly from there. Every opcode has its own predefined gas cost.
Where does 21,000 come from? Per Vitalik Buterin, the creator of Ethereum:
Some more gas was tacked on to account for transaction-specific overhead, bringing the total to 21,000.
Currently the block limit is 30M units; EIP-1559 technically sets the target at 15M, but there is capacity to include up to double if network demand is sufficiently high. Ultimately the total data requirement in any block is variable.
Source: beaconcha.in
Just as the cost of driving a car is equal to the amount of gas used multiplied by the cost of gas, so too is the cost of having an operation executed on Ethereum. While the amount of gas required is somewhat defined by the complexity of the operation and is out of the hands of the user, the ultimate amount the user is willing to pay per unit of gas is up to them. This is reflected in the gas price and is made up of 2 separate pieces since EIP-1559 went live in August 2021: the base fee and the priority fee.
Source: Etherscan.io
The base fee is programmatically adjusted block-by-block by the network based on network congestion and represents the minimum price for gas a user can pay to potentially have their transaction included. The priority fee is an additional amount above and beyond the base fee that a user can choose to pay to the validator to have their transaction prioritized, essentially a tip. As such, the total price per unit a user will pay for gas is the base fee plus the priority fee. The total cost of a transaction is thus the gas limit * (base fee + priority fee). These fees are denominated and paid in the network’s native token, ETH. Through these priority fees, validators can earn substantial payments arising from Maximum Extractable Value (MEV), but we will save those details for another post.
With EIP-1559, the network automatically burns the base fee while the validator receives the entirety of the priority fee. In times of high demand, this automatic burn turns the network deflationary and has resulted in a total burn of 4.4M ETH, equivalent to roughly $12.3B. During the massive market sell-off earlier this month, this dynamic was clear. As trading volume (e.g. selling) skyrocketed, so too did the amount of ETH burn from transaction fees:
Source: TheBlock
Dating back to Q4 2023, fees on Ethereum have totaled $2.4B, $400M of which have gone directly to the validators and the rest burned.
Source: TokenTerminal
Notice the change in makeup of total fee payments since EIP 1559 went live – from 100% to validators, down to roughly 15-20% on average with the remainder being burned. For the first several quarters, this was going to miners that were still securing the network through Proof-of-Work (PoW). It was not until ‘The Merge’ upgrade happened in September 2022 that the network shifted to its current Proof-of-Stake (PoS) mechanism, switching away from a miner-secured network to that of a validator-secured one.
Source: TokenTerminal
EIP-4844 (proto-danksharding) was a major upgrade in March 2024 that essentially switched L2 fee payments to a ‘wholesale’ model and away from a ‘retail’ model, creating a separate fee market for chains when posting data to mainnet. We go deeper into this in our next post, but it has had the effect of dampening transaction costs, and thus fee generation, on Ethereum. For example, Base paid $9.3M to post data to Ethereum in Q1, Arbitrum paid $20.1M and OP Mainnet paid $14.2M, combining for almost $44M between just those three networks. In Q2, the three networks combined for just $1.4M in payments to Ethereum, a more than 95% decrease.
While across all L2 networks this amounts to a sizeable piece of revenue, the nature of the fee pricing means it has a non-linear impact. That is, the impact of e.g. $50M reduction in fee loss to Ethereum is far higher because the reduced competition means the network-wide gas price also sits at a lower level naturally, thus reducing revenue from every other transaction that still happens on the network. Ultimately, equilibrium depends on elasticity of demand in the long run, and it is important to take a long-term view as to whether the reduced fees to transact on mainnet leads to a corresponding increase in usage to help restore those aggregate revenue levels.
For example – if gas is at 5 gwei and ETH is $2500, 1 million ERC-20 transfers (65,000 gas limit on average) would lead to ~$81M in fees. Now, imagine an L2 leaving that fee market that paid $10M in settlement costs also caused the price of gas to fall to 4 gwei from 5. Those same 1 million transfers would result in $65M fees. So, the $10M L2 fees ‘leaving’ the market really caused a $26M reduction in network fee generation overall from the reduced competition. Though occurring at a time when network activity fell too, the reduced gas price is clear:
Incentivizing Security through Network Issuance
Not only did The Merge upgrade cut the energy consumption of the network by 99.95% and solidify the foundation of Ethereum’s roll-up centric future, but it also dramatically reduced the rate of new network issuance. Pre-Merge, miners earned ~13,500 ETH daily from validating transactions, with a fixed amount of 2 ETH earned per block prior to the upgrade.
Source: Block rewards paid to miners prior to the Merge via Etherscan
Following the switch to PoS, new token issuance is now programmatically derived at a rate of (2.6*64) times the square root of the sum of staked ETH annually (though potential changes to this are now being discussed, rooted in Minimum Viable Issuance). Put another way, the rate of new ETH issuance falls as the amount of staked ETH increases: for example, 166K ETH is issued if 1M ETH is staked, and 1.66M if 100M ETH is staked. The goal of this is to incentivize a balanced participation rate of ~30M ETH staked, resulting in a yield of ~3.3%. With a base requirement of 32 ETH per individual validator, this translates to ~1M validators participating in consensus at desired steady state.
Source: Active Ethereum Validators have recently passed 1M via beaconcha.in
Currently, there are 1,055,017 active validators staking 33.6M ETH, earning 3.35%. Relative to the total supply of ETH circulating, that amounts to 28% of the entire supply staked to participate in consensus. This issuance serves a dual purpose. Primarily, it directly incentivizes staking by rewarding those who help secure the network. It also serves an inflation tax on those who do not, as their relative ownership of the network decreases as ETH supply grows. Over the past week, the network issued 18K new ETH to validators while burning 8.7K ETH from transaction fees, resulting in a net new issuance of 9.3K ETH. Annually, this amounts to a net supply growth of ~0.4% ($2.8B issued to validators). This is almost 10x lower than what would be the case if the network was still using PoW, with 3.7% supply growth, or ~$15B in new annual issuance.
Source: validatorqueue.com
Importantly, ‘validators’ essentially just means the number of nodes with 32 ETH. These are typically run by professional operating companies but also can be run by individuals. Either way, validators are not required to stake all 32 ETH themselves. Rather, many validators aggregate ETH from other holders and then run the node on their collective behalf, passing through the issuance rewards while taking a small fee of 5-10%. This democratizes staking to any holder of ETH regardless of amount held and has proven highly successful in growing participating share. This explosion of staked ETH has led to massive growth in liquid staking and restaking protocols such as Lido and Eigenlayer, with $26.5B and $14.3B staked deposits, respectively.
Bottom Line: Is Ethereum Profitable?
There are two lenses through which one can gauge whether Ethereum is profitable as a network in its current state. The difference comes down to whether one considers validator priority fees as being endogenous or exogenous to the network. If one considers them endogenous, then Ethereum has been net profitable in 6 of the 8 quarters following the Merge, with last quarter and the partial current quarter being the exceptions (Q3 may yet be profitable). In aggregate, these 8 quarters amount to overall network profits of $1.1B, or $138M per quarter. For comparison, the final quarter before the Merge saw a net ‘loss’ of $1.5B.
There are others that hold that the ‘revenue’ to Ethereum consists of only those fees that are derived from the base fees and burned, thus excluding validator priority fees. Naturally, if one uses that measure overall network profitability declines slightly (revenues are down 15-20%, after all), but the profitability picture remains dramatically different from historical levels. Since the Merge, that amounts to total profits of $350M to ETH holders, or $44M per quarter.
Source: TokenTerminal
In the long run, for the network to maintain profitability – most conservatively defined by the amount of ETH burned from transactions being greater than the amount of new ETH issued to validators – there will need to be either a) far greater transaction demand on the network, thus resulting in higher average gas fees, or b) reduced issuance, as mentioned earlier. Recently, average gas costs on ETH have been below 4 Gwei with discrete periods during which that price spikes. To be at a level where ETH turns deflationary from burns, gas costs need to be close to 24 Gwei, all else equal.
Source: Ultrasound.money
This Profitability Sets Ethereum Apart from Other Networks
Importantly, Ethereum only issues new tokens to validators in return for securing the network - this is its only expense (all of which is paid to ETH holders by definition). A combined outcome of its relative age (2015), user and developer share capture and its incentive design, Ethereum does not face the same disproportionally massive spend requirement that other networks face to generate activity. We went deep into the mechanisms underpinning airdrops/token programs and how they are used to incentivize activity in a recent post. While specifically focusing on Ethereum L2 Starknet then, all new networks (L1s, L2s and L3s) face the same major issue: overcoming the cold start problem to build out supply (developers) and demand (users).
Outside of Ethereum, Solana ($77B FDV) and TON ($27B FDV) are the next largest smart contract platforms in terms of network market capitalization. Both implement similar fee models to Ethereum, splitting payments into validator tips and burning the rest. The major difference is that up until a few weeks ago, both Solana and TON strictly burned 50% of transaction fees, instead of the variable amount as on Ethereum. In May 2024, the Solana community voted to change the model and send 100% of priority fees to validators and forego any burn, thus removing some of the ‘revenue’ that flows to token holders via the burn.
Source: TokenTerminal
Either way, neither Solana nor TON see near the same profitability as Ethereum. Though Solana’s network incentives are on roughly the same magnitude (e.g. $100s of millions per quarter), its fee generation and revenues are much lower. TON operates on an entirely different magnitude altogether, paying out just $10s of millions quarterly in token incentives. On TON, new issuance is ~10x that of the burn mechanism. Both of these networks are much earlier in their development than Ethereum and both have long-term plans to move validator rewards to a fee-driven structure and away from an issuance-driven structure. In the short term, however, network growth is highly reliant on endogenous incentives and fully fee-funded security is a long way off.
Interestingly, both Solana and TON are actually competitive with Ethereum mainnet in terms of pure active user counts (excluding L2s and L3s). Since late 2023, both have seen marked growth in MAU, especially relative to a somewhat stagnated Ethereum. Solana now boasts nearly twice as many users as Ethereum, while TON has roughly half (up from just a few % in early 2023).
Source: TokenTerminal
Despite this, Ethereum’s fee generation is far higher than the other chains. This largely derives from the difference in cost to transact on each and the nature of the activity. Average transaction fees on Ethereum are multiple orders of magnitude higher on Ethereum than on Solana or TON. Average fees on Ethereum are typically between $5-$10 but can spike substantially in times of high network congestion due to the mechanisms we discussed earlier. At the height of the 2021 cycle, a simple transfer transaction could be over $50. In recent hot markets, fees spiked to almost $30 on Ethereum. In comparison, TON’s average fee on that same day was 1000x cheaper at just $0.02 and on Solana was another 10x lower than even TON at just $0.003.
Source: TokenTerminal. The flat lines at the bottom reflect Solana and TON fees – tiny relative to Ethereum.
A major reason for this higher fee regime on Ethereum is that its fee market and mempool are global, meaning that if a single application (such as a high-demand NFT mint) results in a period of increased activity, all other uses of the chain are put in competition with those users trying to secure the earliest spot in line for the mint. In practice, that means a user trying to have a simple ETH send included will have to be willing to pay the same fee (base+priority) as a power-user trying to snipe the first NFT. Solana, meanwhile, implements local fee markets, meaning that high-demand events can stay isolated from the broader ecosystem, reducing any negative externalities on other users.
Naturally, the size of transactions on Ethereum and the nature of activity are of far higher value. Average revenue per user (ARPU) on Ethereum is typically $10-$20+, just $0.50-$1.00 on Solana, and $0.10-$0.15 on TON. For example, TON is almost entirely click-game-driven at this point and has just $500M in value locked on its platform, concentrated in essentially just 2 protocols. Relative to Ethereum’s $100B across hundreds of projects, the difference is clear. Blackrock chose to put its $500M BUIDL fund on Ethereum for a reason. At the same time, if retail users want to trade $50 of memecoins or a project wants to mass-mint NFTs or build out a DePIN network, Ethereum is prohibitively expensive and would make little sense to build an application like Solana’s pump.fun, Bubblegum, or Helium.
Modularity Matters
Many readers will note that there is a multi-billion-dollar hole in this explanation thus far, namely in the form of Ethereum L2s. Next week, we will continue this discussion by looking at the economics of Ethereum rollups, specifically Arbitrum, Optimism (OP Mainnet) and Base, how modularity differentiates Ethereum from the other top layer 1 networks, and the impacts of EIP-4844 on Ethereum’s fee structure.
