How much does it cost to run an L2?

Original Title: "Blockchain Economics: How much does it cost to run your own chain?"

Author: Sharanya Sahai, Hashed Emergent

Compiled by: 0x26, BlockBeats

Editor's Note: Galaxy Research recently published that "since the Cancun upgrade, the revenue from Layer 2 Ethereum mainnet protocols has been nearly 0." Ethereum is moving further down the path of scalability, but what is the actual cost of running an L2? Through this article, we can understand the true costs of "one-click chain" L2 projects.

In the past year, the number of new Layer 2 (L2) solutions has significantly increased due to technological advancements, focus on specific application scenarios, and strong community participation. While this development is encouraging, the main challenge remains how to scale these blockchains in a more cost-effective manner. Running application chains has become a key means to address this issue, as application chains can control the operational costs of the blockchain through various initiatives in a modular infrastructure stack.

Although L1 ------ Ethereum's specific initiatives have significantly reduced transaction costs on the blockchain, major Rollup and infrastructure providers are also actively promoting further scalability improvements to unlock use cases that are currently too costly to execute on-chain.

We can categorize and analyze these developments into three types: a) L1 initiatives, b) L2 initiatives, and c) modular infrastructure initiatives, all of which have made meaningful contributions to reducing the barriers to on-chain transactions.

First, Ethereum has undergone some upgrades, such as EIP 1559 and 4844, which have reduced costs and improved scalability.

Let's first look at the contributions of L1 initiatives to rationalizing Ethereum's on-chain transaction costs, such as EIP 1559 and EIP 4844 (Cancun upgrade). EIP 1559 introduced the concept of base fee + tip/prioritized fee, along with a dynamic pricing mechanism based on network congestion, providing users with a better mechanism to estimate costs and trade on the network according to their priorities and network congestion. EIP 4844 introduced a new transaction type for Ethereum through the concept of Blobs, allowing L2 to store data in the form of Blobs instead of expensive callData, significantly reducing costs when settling transactions on L1.

The implementation of Blobs has led to a significant decrease in transaction costs due to reduced storage costs per byte and increased block capacity, as Blobs do not compete for Gas with Ethereum transactions and are not permanently stored, being deleted from the blockchain after about 18 days.

Each Blob contains 4096 elements of 32 bytes, with a maximum of 16 Blobs per block, providing an additional maximum capacity of about 2 MB (4096 * 32 bytes * 16 Blobs per block). The current starting capacity is 0.8 MB, with a target of 3 Blobs per block, up to a maximum of 6 (after the implementation of EIP 4844). Considering the historical standard of 2-10KB of callData per block, EIP 4844 theoretically means a capacity increase of up to 384 times.

In practice, after the implementation of EIP 4844, many L2 fees have decreased by over 90%. However, relying solely on these upgrades is not enough for Ethereum to achieve greater scalability. With thousands of Rollups, the demand for storage space increases as large-scale applications emerge, and transaction costs may rise sharply.

As L2 moves execution off-chain to cut costs while maintaining security, industry initiatives such as open-source frameworks and revenue-sharing models are shaping the competitive landscape of the "L2 stack war."

The emergence of Rollups in the previous cycle aimed to significantly reduce on-chain operational costs by moving execution off the main chain while obtaining security from the main chain. While Op Rollups allow a single honest entity to submit "fraud proofs" and receive rewards for identifying misbehaving sequencers, ZK Rollups use zero-knowledge proofs to demonstrate that the L2 chain has been correctly updated.

Rollups perform the following tasks:

· Sorting: Organizing end-user transactions in order, batching them, and occasionally publishing these batches to L1

· Execution: Storing and executing operations and updating the state of the Rollup

· Proposing: Proposers regularly update the Rollup state root on L1, which is crucial for ensuring that the blockchain remains trustless and verifiable

· State root challenge: Submitting evidence of fraud for the state root and challenging the state root on L1 (only applicable to Op Rollups)

· Proof: Generating verification of state root updates from Rollup to L1 (only applicable to ZK Rollups)

They profit from transaction fees paid by users (sequencer revenue) and potential MEV (maximal extractable value), although MEV has not yet been extracted as part of their strategy. Their costs primarily come from L2 (operational costs) and L1 (data availability and settlement) costs. Organizations hoping to launch their own chains typically only consider doing so when they expect transaction fees to exceed the costs of such initiatives.

Base layer networks like Ethereum typically charge more for computation and storage because most nodes need to synchronize and verify the chain. However, in Rollups, even if only one honest entity can verify the chain, it is considered secure. Therefore, Rollups charge less for computation and storage but have higher costs for rolling up transactions to package and publish to L1, leading to L1 costs accounting for 98% of the L2 cost base before the launch of EIP 4844.

In addition to base layer optimizations, L2 is also actively promoting further cost reductions, which are referred to as L2 initiatives at the beginning of the article and can be categorized into two main types: industry alignment or company alignment.

Industry alignment initiatives include allowing new players to build their own chains through open-source L2 technology stacks (Rollup frameworks). This wave of initiatives was initially led by Op Rollups through the launch of OP Stack and Arbitrum Orbit, followed by other mature L2s like Polygon (Polygon CDK), ZK Sync (ZK Stack), and Starkware (Madara Stack), which also promoted large-scale applications by open-sourcing their proprietary technologies.

Company alignment initiatives involve these chains reducing costs through direct revenue/profit-sharing models or indirectly through the secondary effects of expanding their ecosystems, accumulating value for their tokens. Examples of such initiatives include Optimism's Superchain vision, Arbitrum's expansion plans, Polygon's aggregation layer, and ZK Sync's Elastic Chain. While the specifics of these projects may vary, they share a commonality of an interconnected network that provides enhanced interoperability, communication between multiple Rollups, and shared key infrastructure, such as shared data availability layers, shared cross-chain bridges, and aggregated proofs (only applicable to ZK chains), to further enhance capital efficiency—addressing the current issues of liquidity fragmentation and insufficient interoperability between Rollups in the Ethereum ecosystem. However, these stacks also allow individual chains to customize according to their needs in terms of block time, withdrawal periods, finality, token usage, Gas limits, etc., thus eliminating the high Gas costs and delays caused by other applications when operating on public chains.

Despite these independent ecosystems focusing on growth and application, we have begun to see some mature players like Optimism and Arbitrum gradually monetize.

Optimism charges participants wishing to be part of its Superchain a 2.5% fee on total sequencer revenue or 15% of sequencer profits (sequencer revenue - L1 settlement and data availability costs). Arbitrum charges participants using its stack to launch L2 a 10% fee on sequencer profits, while ZK Rollup stacks, including Polygon CDK and ZK Stack, are currently free to use, but as they develop and apply, they may incorporate sustainable economic models.

As all ecosystems compete to attract significant projects through unique strategies, the "L2 stack war" has officially begun. Optimism announced a $22 million bounty for Superchain builders, offering retrospective airdrops based on usage and participation metrics, while ZK Sync provided $22 million in ZK tokens to attract Lens to migrate from Polygon to its stack. Arbitrum offers its stack for free, provided that participants publish as L3 on Arbitrum (referring to using L2 as a settlement layer instead of Ethereum), as Arbitrum benefits from L3 activities, which will always pay settlement costs to Arbitrum throughout their lifecycle.

RaaS and alternative settlement and data availability solutions are redefining the cost structure of blockchains, and future innovations in modular infrastructure are expected to further reduce costs.

Despite the availability of these technology stacks, running a blockchain involves significant operational overhead, personnel, expertise, and resources. Developers looking to attract on-chain users do not want to be distracted by managing and maintaining chain infrastructure but prefer to focus on core business activities.

This issue has led to the emergence of RaaS providers, who collaborate with these developers to abstract the complexity of running chains using mature L2 frameworks/stacks. The services they provide include node operation, software updates, infrastructure management, and products such as sequencing, indexing, and analytics. RaaS providers adopt different market capture strategies, some aligning with specific L2 ecosystems, while others take a more framework-agnostic approach, providing integration across all ecosystems. Conduit and Nexus Network integrate with Op Rollups like Optimism and Arbitrum, while Truezk, Karnot, and Slush focus on ZK chains. On the other hand, Caldera, Zeeve, Alt Layer, and Gelato provide integration across both Op and ZK Rollups.

The typical business model of these providers includes a fixed fee plus a share of sequencer profits. The monthly subscription fee for running an Op Rollup generally ranges from $3,000 to $4,000, while the cost for running a ZK Rollup may exceed twice that, reaching $9,500 to $14,000, due to the immense computational intensity required to generate ZK proofs and the high costs of proof verification. Additionally, to align the incentives of RaaS providers and Rollups, a share of 3-5% of sequencer profits is often charged, allowing them to capture economic upsides as the attractiveness of these chains increases.

Caldera is exploring a different model, its Metalayer vision, which charges only a 2% share of sequencer profits without fixed costs, aiming to achieve interoperability among chains using Caldera, whether Op or ZK series.

It is important to note that the volatility of the industry and teams' efforts on these stacks, particularly ZK stacks, may further compress the subscription costs of RaaS providers. Moreover, due to the scarcity of strong consumer-grade Web3 businesses, large consumer-facing applications may be able to negotiate better economic sharing agreements with infrastructure providers, meaning initial pricing may not be standardized.

As mentioned earlier, the largest expenditure for Rollups is L1 costs, namely data availability and settlement costs. For a standard Rollup handling 100 million transactions, L1 costs can reach up to $25,000 per month, making L1 settlement feasible only for the largest and most frequently used chains. The demand for alternative settlement and data availability solutions has prompted specialized players to optimize costs and performance on these layers. Ethereum's data availability alternatives include Celestia, Near, and EigenDA, while the mature L2s discussed earlier aim to serve as settlement layers for Rollups, categorizing these chains as L3. Compared to Ethereum, these players have reduced the settlement and data availability costs of Rollups by orders of magnitude. The following chart provides a rough cost comparison, indicating how much savings could be achieved if Rollups published callData to Celestia instead of Ethereum. It is worth emphasizing that as transaction volumes increase, the savings gap grows exponentially.

In addition to data availability costs, there are also settlement costs, which involve Celestia publishing markers on Ethereum that point to relevant blocks on Celestia to ensure the ordering and integrity of the data published on Celestia.

The development of specialized players across modular infrastructure stacks, such as alternative data availability and RaaS providers, can be collectively referred to as modular infrastructure behavior. Other categories of innovations are further optimizing costs, including shared sequencers (Espresso, Astria, Radius), proof aggregation (Nebra, Electron), etc. These are currently in the early stages of development, and we expect costs to decline further as the industry matures.

Although the costs of on-chain operations have been significantly reduced, Web2 founders should conduct a thorough cost-benefit analysis before deciding to launch their own chain.

The total cost of running a chain depends on the specific usage requirements of each chain, but we can roughly estimate the costs of an average Op or ZK chain using alternative data availability solutions for processing 2 million transactions per month, as shown in the following chart.

Despite various optimizations at both the industry level and individual chain level, ZK Rollups still require a total monthly cost of $10,500 to $16,500, and $4,000 to $6,500 for Op Rollups. Additionally, once the chain becomes profitable, sequencer profits may need to be shared up to 20%.

The three categories of initiatives highlighted in this article will be key to driving industry adoption, with the ultimate goal of narrowing the cost and convenience gap between decentralized applications and Web2. Builders should carefully assess the cost-benefit analysis of running an independent chain versus building on existing chains based on their end-user needs, product priorities, performance metrics required by application scenarios, and existing market attractiveness.

We find that building solutions to reduce the cost and performance disparities between Web3 and Web2 infrastructure is necessary, as society's preference for using decentralized systems is insufficient to expand the application scope of Web3, and this challenge remains a critical bottleneck in driving large-scale blockchain adoption.