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.
Modular vs. Monolithic
Frequently a debate between communities about which approach is better, modular or monolithic are two differing strategies used to achieve the same goals of decentralization, scalability and security. At their core, blockchains can be broken down into 4 primary functions: execution, settlement, consensus and data availability. The Execution Layer handles the actual execution of transactions and is where applications and smart contracts reside. The Settlement Layer is where confirmed transactions are recorded and made immutable and where any canonical bridging occurs. The Consensus Layer handles transaction validation and is where the state of the network is agreed upon by nodes. And finally, the Data Availability layer ensures the requisite historical data is available to all nodes on the network.
Monolithic blockchains attempt to enshrine all functions in a single layer with nodes responsible for all 4 roles. Modular blockchains instead split some (or all) functions off into separate layers in order to optimize certain components, most commonly execution and increasingly data availability. Due to their prominence on Ethereum we will focus on rollups in this post, but there is a long list of networks that go about this in differing ways, falling along a spectrum between fully monolithic and entirely modular.
Examples of Monolithic chains include Bitcoin, Tron, Solana and pre-Merge Ethereum. Polylithic chains implement network-of-networks approaches, linking sovereign monolithic chains together through a shared communication layer. Examples of polylithic networks include Cosmos, Polkadot and Avalanche. Semi-Modular approaches outsource just a few functions to other networks but maintain the majority of functionality on the base chain. Post-Merge Ethereum, Near and Tezos are examples of this approach – Ethereum technically consists of a Consensus Layer (e.g. Beacon Chain) and Execution Layer (e.g. original ETH Mainnet). Finally, fully modular chains are those that outsource most or all of the functionalities to other networks. Rollups such as OP Mainnet, Arbitrum and Base fall into this category, as do function-specific networks such as Celestia and EigenDA. Increasingly there are new ‘frankenchains’ emerging such as Eclipse that bring together functions from different ecosystems to create a best-of-all-worlds solution. Eclipse specifically is an Ethereum-based Rollup (Optimistic and ZK) that uses Solana’s SVM as its execution layer and Cosmos-based Celestia for Data Availability. Also note that although a chain may be ‘monolithic’, that does not preclude alternative layers being built – the dozens of Bitcoin layer 2s coming to market highlight this fact well.
Scaling Ethereum
We have alluded to in previous posts the massive number of layer 2 and 3 scaling networks being deployed on Ethereum. Each of these is in essence trying to grow the broader Ethereum ecosystem by offloading certain functions from Ethereum mainnet, most often execution. Currently, there are 71 live layer 2s and 20 layer 3s, with another 82 in the pipeline.
Source: L2Beat
Optimistic Rollups and their successor ZK Rollups command the L2 landscape, but OP Stack Optimiums are becoming increasingly common. In our last post we highlighted how expensive Ethereum is to transact on relative to other leading networks such as Solana and TON. These Ethereum L2 networks largely target that discrepancy directly, focusing on improving the execution environment while leaving the security and settlement to Ethereum mainnet. Layer 3 solutions – typically built on Arbitrum’s and Optimism’s stacks – extend this separation further, enabling even more customization and improvement around execution, with settlement back to the layer 2 chain.
In essence, each layer in this stack submits a proof of a bundle of transactions for inclusion in a block on the underlying chain, termed ‘posting data’ to that chain. While this ever-growing stack starts to blur the line of what function happens where, the easiest delineation is what makes a layer 1 network: a blockchain that does not depend on another chain for security. Any network that derives its security from another chain is a layer 2 or 3 network. Security refers to the guarantee that any transactions recorded are correct, irreversible and final and is assured by the collective actions of all nodes on the network. For Ethereum, ‘security’ is derived from the 1 million nodes securing the chain with a ‘staked’ security budget of almost $85B.
Optimistic Rollups: the Path to a Brighter Future?
Despite the proliferation of scaling networks on Ethereum, the landscape is still very much dominated by a few main L2s: Arbitrum, Optimism’s OP Mainnet, and OP Stack-built Base. Combined, these three networks account for ~75% of all value residing on L2s. Each of these implements a similar “Optimistic” rollup architecture. In terms of functionality, each of these takes on Execution while leaving Settlement, Consensus and Data Availability to Ethereum. Each chain has its own sequencer that is ultimately responsible for ordering transactions on the network, the outcome of which (state root) is posted to Ethereum. For Optimistic Rollups, there is a 7-day window during which 3rd party verifiers can re-execute those transactions and challenge the result if it disagrees via a fraud proof. With each state root post there is also a bounty included and if the challenge to that post is successful, the verifier receives that bounty. If there is no challenge during this window, the state root is committed to Ethereum and is final. These rollups get their ‘Optimistic’ name due to the fact that transactions are assumed to be valid unless proven otherwise.
Now: The Economics
The preceding sections are more technically focused than most of our posts, but that additional context is required to understand the different economic models of these rollups and what that means for Ethereum. Arbitrum, Optimism and Base are not only the largest but also provide a unique look into three slightly different economic models being used.
Arbitrum is a straightforward example of an Optimistic Rollup. Through its sequencer, it receives fees from users to execute transactions on the network. These fees consist of 2 parts: the cost to execute on L2 and a fee to cover the cost of posting to Ethereum L1. Most rollups are centralized (with plans to decentralize going forward), which in essence means that the sequencer that is responsible for ordering transactions is run by the Foundation or development team. Arbitrum is no different and thus any fees paid to the sequencer are retained. One difference from Ethereum is that there are no priority fees on Arbitrum – the sequencer simply operates on a first-come-first-serve basis. The main costs for Arbitrum are two-fold: posting data to Ethereum L1, and token incentives to encourage development. Arbitrum does not have network validation and thus does not natively issue ARB tokens as is done on Ethereum, instead the token is strictly used as an incentive and governance device. All gas fees on Arbitrum are denominated and paid in ETH.
Optimism takes a very similar approach to Arbitrum, but with minor differences. It too runs a centralized sequencer through which it earns execution fees. There are priority fees on OP Mainnet (unlike on Arbitrum) that accrue to the sequencer as well. Uniquely, Optimism has also created a L2 framework called the ‘Superchain’, a network of chains that share security and communication layers built on a common open-source development stack, called the OP Stack. As part of this, partner chains contribute a share of their own revenues to the OP Collective. Currently, Base is the main example of a chain built using this framework, but there are dozens on the way. As such, Optimism receives fees from Base, equal to the greater of a) 2.5% of Base’s top-line sequencer revenue or b) 15% of Base’s net on-chain sequencer revenue (e.g. netted out after L1 settlement costs). In terms of expenses, Optimism incurs L1 settlement costs and token incentives via grants to fund development. Similar to Arbitrum, it does not natively issue OP to validators nor use it as the gas fee on the network.
Base is a unique implementation of an optimistic rollup and is Coinbase’s own L2 chain. It is built on the OP Stack and is the preeminent ‘Superchain’ to join beyond OP Mainnet. As such, it earns revenues through execution but must pay a share of that to the OP collective. Beyond that fee share, expenses are essentially entirely L1 settlement costs. Base has no native token of its own to use as an incentive mechanism and rather largely depends on the influence and draw of Coinbase to spur development on the chain.
Source: Arbitrum, Optimism, Base documentation
Given that L2s have no validators to pay or security budget to ensure, and that L2s incorporate L2 settlement costs into the transaction fees they charge, the key driver of L2 economics ultimately comes down to the cost to post data to Ethereum. Historically, this has accounted for ~70-80% of the cost structure of L2s and has been passed through to the users, resulting in higher transaction fees.
EIP-4844 (“proto-danksharding”) was an Ethereum upgrade that went live in March 2024. In essence, it created a new rollup-specific type of transaction called ‘blobs’ that gave rollups an alternative route through which to post data to Ethereum rather than using the more expensive calldata (essentially an arbitrary field in a transaction that usually specifies what functions to call along with any parameters). The primary difference is that blobs are ephemeral and last just long enough to ensure data is available if necessary to secure rollups but does not result in permanent storage on Ethereum. This effectively transformed L2s from ‘retail’ buyers into ‘wholesale’ buyers of Ethereum blockspace, reducing costs substantially and creating a separate fee market priced in ‘blob gas’.
Overnight, rollups went from spending 70-80% of revenues on L1 settlement costs to just a few %. Arbitrum’s relative cost remains slightly higher than OP and Base given it has no Superchain-type fee sharing, which reduces the relative share for Optimism (e.g. higher revenues separate from L1 fees) and Base (Superchain fees are a larger share of costs). Either way, net profitability exploded for rollups following the upgrade. If one excludes token incentives (ARB and OP are community-driven grants programs, no programmatic issuance), profitability went from ~20% on these chains to ~90%+. Base, for example, earned $19.5M in profit in Q2 2024 on $23.8M in revenues. In Q1 2024, off $27.3M in revenue, Base earned just $15.5M. Base paid $3.7M and $2.5M to Optimism in Superchain fees in Q2 and Q1, respectively. This dynamic had a major impact on Ethereum’s fee generation and was a big driver of its fall from $1.2B in Q1 to $500M in Q2. Because the fee market is non-linear, the impact to Ethereum was the loss of L2 fee revenue and that loss’ compounding effect on general gas prices on the chain for all users.
As for user experience, this was similarly a massive upgrade. We mentioned previously how competing L1s Solana and TON regularly saw transaction fees 1000x lower than that of Ethereum. While low, L2 average transaction fees have historically still been 50-100x higher than Solana and TON as well (e.g. $2 to transact vs. $0.003 or $0.02). Following EIP-4844, average transaction fees plummeted and are now on par with those of Solana and TON, falling to just a few cents or fractions of a cent. Fees at this level unlock consumer-level applications that historically were unfeasible due to high transaction costs.
Source: TokenTerminal. Spot when EIP-4844 was adopted by the L2s.
We also highlighted how user growth on those competing layer 1s has far surpassed that of Ethereum, primarily due to better experience resulting from the lower cost and speed to transact on those networks. That was only half the story, given that Ethereum has fully embraced its rollup-centric architecture. As such, it is important to consider ecosystem-wide user growth across all L2s, rather than just Ethereum mainnet, given that execution in this model (e.g. user transactions and applications) is supposed to take place on the L2s. If we take that more holistic perspective, it is clear that Ethereum, with its L2s, is still king. Adding only Arbitrum, Optimism and Base users to Ethereum’s count (purple line) the success of this model is clear.
Source: TokenTerminal
Since EIP-4844, Ethereum as an ecosystem can now compete with rival L1s in terms of execution and speed, and thus use cases, while still leveraging the massive security budget of the mainnet – a critical component that other chains cannot match, thus limiting upside potential around the types and value of applications or users willing to use those chains. It remains to be seen where fee levels will remain once user counts reach scale (e.g. 10s of millions). At that point, there is some risk that the cost of L1 settlement again reaches a point beyond feasibility. At that point, it will be important for Ethereum to implement additional upgrades to continue keeping those costs down, otherwise it is possible (perhaps likely) that L2s made the switch to using function-specific networks for data availability such as Celestia or EigenDA.
Collectively, these users are bridging massive values to the L2 networks. Just those three combined amounts for nearly $30B in value locked on the chains, up from ~$10B at this point last year and just $4B two years ago. Including several of the other top scaling networks, TVL is nearly $40B and peaked at $47B in June. If denominated in ETH, however, the growth has not slowed, having recently reached an all-time high of 13M ETH.
Source: growthepie.xyz
Conclusion
One of the biggest knocks on Ethereum historically has been the frequently absurd costs to transact on the network in times of high congestion. This lack of scalability has led to a proliferation of alternative layer 1 blockchains with more scalable architectures, such as Solana and TON. At the same time, it has also led Ethereum to evolve from a monothilic proof-of-work chain to a modular-first proof-of-stake chain with an exploding ecosystem of layer 2 scaling solutions to help spread user demand across execution-focused networks. This competition is one of the primary drivers of blockchain development and is pushing the industry forward. With EIP-4844 rolled out, and Solana’s new Firedancer Client still under development, the industry is reaching its ‘broadband’ moment in terms of capacity to support hyper-scale applications.
User growth, fee generation and network profitability have all grown immensely over the past several quarters. Ethereum is in a unique position as a layer 1 chain in not having to heavily incentivize activity as is the case with most others. At the same time, new networks launching can easily tap into Ethereum’s security budget and network effects while focusing entirely on improving the user experience (execution). If successful, as demonstrated by top L2s such as Arbitrum, Optimism and Base, profitability can be achieved far, far earlier than trying to bootstrap a new L1 from scratch. Fragmentation remains an issue and the overall upside in terms of network value is somewhat limited as an L2, but increasingly it is difficult for new networks to justify trying to compete against the established layer 1 ecosystems instead of building on Ethereum’s ever-growing modular stack.
He brings a wealth of knowledge from BlackRock, BGI, Goldman Sachs and more.