150 days after Decun: What will happen to Ethereum and L2s?

The introduction of EIP-4844 and blobdata has brought significant improvements to the operational economy and utilization of Ethereum rollups. However, this change has resulted in a partial shift in revenue from Ethereum L1 and ETH burning rates towards rollup operators, also known as sequencers.

8/24/202414 min read

Summary of Metrics After the Deployment of EIP-4844

Following the implementation of EIP-4844, Ethereum has undergone several notable changes related to blob revenue and operational costs for rollups:

  1. Blob Revenue and Fees:

    • A total of 2,225,958 blobs were purchased at an average price of $1.59/blob.

    • There were 1,104,315 Layer 1 transactions executed at an average cost of $5.22/transaction.

    • Total revenue generated from blobs amounted to 2,692.39 ETH and $9,318,794, with 89.45% (or 2,408.41 ETH) being burned and the remainder transferred to validators.

  2. Blob Usage and Costs:

    • Rollups purchased 285 GB of blob data and utilized approximately 76% of that capacity.

    • Blobs are of fixed size at 128 KB, with a maximum of 6 blobs per Ethereum block. Data in blobs automatically expires after two weeks.

    • Rollups incurred total costs of $3,549,430 for blobs, translating to $16,473 per GB of usage and $12,458 per GB purchased.

  3. Rollup Operating Costs Post-Dencun:

    • The average daily operating cost after the Dencun upgrade is $556.4 thousand, down from $1.07 million prior, when CALldata was used.

  4. Gross Margins:

    • Gross margins for rollups have significantly increased:

      • Optimistic Rollups rose from 22.65% to 92.3%.

      • ZK Rollups increased from 27.27% to 66.7%.

      • Despite a 42% decline in fee revenue after Dencun, gross profits improved due to reduced costs.

  5. L2 Activity and Transaction Failure Rates:

    • Post-Dencun, activity on L2 has increased but has also seen a higher rate of transaction failures, particularly due to active bot interference. However, the average user experiences only a slight increase in failure rates.

  6. Overview of Revenue and ETH Burned:

    • Revenue from Ethereum and the amount of ETH burned after Dencun have significantly decreased, with respective reductions of 69% and 84% compared to pre-upgrade averages.

This summary highlights the financial dynamics and operational shifts within the Ethereum ecosystem following the EIP-4844 upgrade, illustrating both the challenges and the potential for improved efficiency in rollup operations.

Summary

These changes reflect a shift in how Ethereum and rollups manage data and costs following recent upgrades.

Unlike calldata, which is stored on Ethereum's execution layer, blobdata is temporarily stored on the consensus layer and cannot be queried by smart contracts through the Ethereum Virtual Machine (EVM). This change helps offload computation, reducing the costs associated with rollup data storage and submission to Ethereum by decreasing dependence on calldata and the network's execution layer. Blobdata is submitted to Ethereum via type 3 transactions, where only references to the blobs are stored on the consensus layer, excluding the blobs themselves or the data contained within them.

  • The cost of storing rollup data in the calldata for type 2 transactions is calculated as follows

    (Base Fee + Gas Used for calldata) + (Priority Fee × Gas Used for calldata)

  • In contrast, the costs associated with type 3 transactions that contain blobdata on Ethereum are calculated as follows:

    (Base Fee for blob + Gas Used for blob) + (Base Fee + Gas Used for calldata) + (Priority Fee × Gas Used for calldata)

Despite the additional base fee for blobs factored into the total cost, type 3 transactions are generally cheaper than type 2 transactions because they use less calldata gas, and blob gas fees are typically lower than calldata gas fees.

Blobs have a separate fee market, independent of other types of Ethereum L1 transactions. This market has its own parameters for setting base fees, which are adjusted solely based on the demand for blobs in the previous block. Activities on Ethereum L1 (such as transactions on decentralized exchanges) do not impact blob costs. While L1 activity may affect the base fees of transactions containing blobs, it does not influence the base fees for blobs themselves.

The base fee for blobs in this separate market starts at 1 wei and increases as demand exceeds the target blob quantity per block until a maximum level is reached. In terms of gas, a single blob consumes 131,072 gas. With ETH priced at $3,000, the base fee for one blob could be as low as 0.00000000000131072 ETH (12 zeros) or $0.00000000039322 (nine zeros). This means that the initial stage of calculating fees for type 3 transactions may be nearly free.

This report focuses on analyzing the utilization of Ethereum blobs from the perspective of storage and submission across both the consensus and execution layers, as well as the impact of EIP-4844 on rollup economics. The report does not include an analysis of the eight other EIPs in the Dencun upgrade.

Blobs on Ethereum's Consensus Layer:

This section provides an overview of blob usage from the perspective of Ethereum's consensus layer, including the number of blobs purchased, the total costs incurred by participants, and the volume of blob capacity purchased and utilized.

The table below lists the top 10 rollups on Ethereum by the number of blobs purchased. In total, these rollups have acquired 1,939,657 blobs (equivalent to 248.2 GB of blob capacity, accounting for 87% of total blob demand) and spent approximately $2.1 million (or 603.6 ETH) within the 150 days following the activation of EIP-4844. Among these, Paradex, StarkNet, and Arbitrum utilized the majority of the blob capacity, with an average utilization rate of 95.23%. Base is the largest spender, investing $811,000, equivalent to 232.5 ETH, to purchase 619,516 blobs.

Consensus Layer Blobs Table:

To calculate the cost of a single blob in ETH, the following formula is used, where Blob Gas utilized is 131,072 gas:

(Base Fee for Blob * Blob Gas Used) / 1e18

This formula helps in understanding the expenditure related to blob purchases by rollups, reflecting their operational costs and strategic investing in blobdata for enhanced efficiency on the Ethereum network.

The cost of a blob can also be calculated in USD by multiplying the result above by the price of ETH at the time the blob is recorded in an Ethereum block. This corresponds to the ETH value at the moment the transactions carrying the blob are executed on-chain. As mentioned earlier in the report, the total cost of utilizing blobdata results from three components in the calculation of type 3 transaction fees. The cost of the blob is just the first part of this formula, while the other two components related to the costs of the transactions carrying the blobs will be discussed in detail later in the report.

As of May 28, 2024, the demand for blobs has remained stable at around 15,000 to 20,000 blobs per day, coinciding with the launch of Taiko, a new type of rollup. Taiko is characterized by its transactions not being confirmed until they are recorded on Ethereum L1. This requires Taiko to continuously purchase blobs to maintain chain operations, in contrast to optimistic or knowledge-free rollups, which allow for soft confirmation of transactions prior to their formal confirmation on L1. This design grants the sequencers of those rollups greater flexibility in acquiring blobs and managing the frequency of transactions on L1.

Despite being a new rollup, Taiko has quickly become the third-largest spender on blobs in the consensus layer and leads in spending for blob-carrying transactions in the execution layer. This highlights the growing importance of blob utilization in optimizing transaction processing and the evolving landscape of rollups on the Ethereum network.

Base and Taiko are currently the two largest daily purchasers of blobs, averaging a total of 8,667 blobs purchased each day, accounting for 47.4% of total blob demand since June 1, 2024. Base averages 5,094.5 blobs per day, while Taiko purchases 3,572.5 blobs per day during this period. Despite launching 75 days after blobs were introduced, Taiko remains one of the top three entities purchasing blobs since the deployment of Dencun.

From a data demand perspective, rollups have consistently purchased between 2 to 2.5 gigabytes of blob data each day since June 1, 2024. However, they have only utilized 1.5 to 2 gigabytes of actual data from the purchased blobs, indicating that only 71.91% of the capacity has been utilized during this period.

June 20 and 21, 2024, marked the highest utilization of blobdata, with capacities reaching 2.66 and 2.75 gigabytes, respectively. Notably, June 20 coincided with the Layer Zero airdrop event on Arbitrum, leading to a surge in activity on that platform. The period from March 27 to April 3, 2024, also recorded high demand for blob capacity, with an average utilization of approximately 2.7 gigabytes per day. During this timeframe, "Blobscriptions" gained popularity, artificially inflating demand for blobs. Blobscriptions allow users to embed arbitrary data into blobs, diverging from their primary purpose of storing rollup transaction data.

This data illustrates the evolving landscape of blob usage and highlights the growing significance of rollups in managing and utilizing blobdata effectively, as demand continues to fluctuate based on various events and new use cases.

In the 150 days following the launch of Dencun, an average of 21.1% of blob capacity went unused each day. The chart below illustrates the daily allocation of blob capacity, highlighting the difference between the portion utilized by rollups and the remaining unused capacity.

There were nine notable days when rollups’ blob spending surpassed $100,000 each day. These days coincided with the Blobscription frenzy at the end of March and early April, as well as the Layer Zero airdrop event on Arbitrum on June 20. During these nine days, total blob spending by rollups reached $3,542,579, with an average daily expenditure of $394,000, which is 815,714% higher than the average of other days. Remarkably, these nine days accounted for 99.8% of the total $3,549,430 spent on blobs since the deployment of Dencun. Excluding these exceptional days, the average daily expenditure of rollups was only $48.25.

Furthermore, on these days, the average number of blobs per Ethereum block was 2.9, which is 52% lower than the maximum limit of six blobs per block, 4% below the target of three blobs, yet 40% higher than the daily average of 2.06 blobs per block since the launch of EIP-4844. This indicates that costs surged when the demand for blobs consistently met or exceeded target levels.

Blob Costs per Gigabyte

Blob costs can be measured based on the blob space that rollups have purchased and utilized. During the peak of the Blobscription craze, the cost reached $503,000 per gigabyte of utilized capacity and $207,000 per gigabyte purchased.

This data underscores the volatility and dynamics of blob utilization, reflecting the impact of market events on pricing and capacity management within the Ethereum ecosystem.

Similarly, on the day of the Layer Zero airdrop on Arbitrum, the cost was $410,000 per gigabyte utilized and $330,000 per gigabyte purchased. It's important to note that the cost of usage is typically higher because rollups do not fully utilize the blob capacity they have purchased.

In total, rollups have spent $3.55 million and 1,020 ETH on blobs in the 150 days following the launch of Dencun, with the ETH being removed from circulation similarly to the transaction fees burned on Ethereum.

In summary, since the introduction of EIP-4844, companies have spent $12,458 for each gigabyte of blob space purchased and $16,473 for each gigabyte of blob space utilized. This data reflects the evolving economic landscape surrounding blob usage, highlighting the costs associated with managing and utilizing blobdata effectively within the Ethereum network.

Impact on Ethereum Validators and ETH Supply

The introduction of blobs has led to a decline in total revenue from Ethereum and the amount of ETH burned from rollup activities. The analysis below highlights three key points:

  1. Change in ETH Burned: Following the deployment of EIP-4844, the amount of ETH burned from base fees and blob fees of type 3 transactions has significantly decreased. The total ETH burned in the 150 days after EIP-4844 was 2,408 ETH, down from 3,286 ETH in the 150 days prior to EIP-4844. On average, the daily ETH burned from batch transaction commitments on Ethereum decreased from 15,052 ETH before Dencun.

  2. Direct Revenue for Validators: Revenue from priority fees paid to validators has shifted. Before Dencun, revenue from batch transaction commitments primarily came from calldata. After the implementation of EIP-4844, revenue from calldata sharply declined, totaling 39.69 ETH in the last 30 days compared to 13,163 ETH in the same period prior to Dencun.

  3. Total Revenue for Ethereum: The total revenue that Ethereum generated from rollup activities after EIP-4844 has also decreased compared to before Dencun. The pre-Dencun figures do not include revenue from zero-knowledge (ZK) proofs, as they only account for revenue from optimistic and ZK batch transaction commitments.

In summary, the deployment of EIP-4844 has led to a reduction in both the amount of ETH burned and the overall revenue of Ethereum from rollup activities, primarily due to the adoption of blobs as a replacement for calldata. This shift in how Ethereum processes transactions and collects fees suggests significant changes in economic incentives for both validators and the overall network.

The following chart compares the priority fees paid to validators before and after EIP-4844. In total, $974,876 has been paid to validators through priority fees from type 3 transactions containing blobdata, equivalent to the payment levels observed in May 2022 and June 2023 from batch calldata commitments before the use of blobs.

Prior to EIP-4844, batch calldata commitments generated an average of $1.196 million in priority fees over 150 days. Validators currently earn less from the priority fees of type 3 transactions containing blobdata compared to those from type 2 transactions with calldata.

Finally, we can compare the total revenue earned from blobs with the total revenue earned from batch calldata commitments.

The formula for calculating the total revenue earned from blobs is:

(Base Fee for Blob × Gas Used for Blob) + (Base Fee × Gas Used for Calldata) + (Priority Fee × Gas Used for Calldata)

The formula for calculating the total revenue earned from calldata confirmations is:

(Gas Used for Calldata) + (Priority Fee × Gas Used for Calldata)

Ethereum has generated a total revenue of $9,318,794 from blobdata and type 3 transaction fees, compared to an average total revenue of $29.92 million over 150 days from type 2 transactions utilizing calldata.

This analysis highlights the shifting revenue dynamics within the Ethereum network as it adapts to the implementation of EIP-4844. Even with the introduction of blobs, the overall revenue generated from traditional calldata transactions remains significantly higher, indicating potential areas for improvement as the ecosystem evolves.

Other Economic Indicators

The operating costs of rollups have significantly decreased following the deployment of EIP-4844, with notable exceptions during events such as the Blobscription frenzy, the Layer Zero airdrop, and the market volatility on August 5. On average, rollups have spent approximately $556,400 per day on operations, which includes costs associated with blobdata, submitting blobdata through type 3 transactions, batch calldata commitments, and zero-knowledge proofs (in the case of Scroll, Linea, and zkSync).

This figure marks a substantial reduction compared to the average daily operating costs of $1.9 million over the preceding 30 days, $1.27 million over the 90 days, and $1.07 million over the 150 days prior to the implementation of EIP-4844.

This reduction in operating costs signifies improved efficiency in managing rollup operations, reflecting the positive impact of blob utilization on overall cost structure. As rollups adapt to new economic conditions under EIP-4844, tracking these cost changes will be essential for understanding their sustainability and profitability moving forward.

The reduction in operating costs of rollups has led to a corresponding decline in their revenues, with notable exceptions on the Layer Zero airdrop day on Arbitrum and on August 5, 2024. On these two days, the total revenue of the six rollups reached $6.09 million, with Arbitrum accounting for $4.63 million (76%).

Since the implementation of EIP-4844, including the transition period to using blobdata, rollups have achieved an average daily revenue of $691,300. This figure is a decline from the average daily revenue of $2.33 million in the 30 days prior to the implementation, $1.46 million in the 90 days, and $1.2 million in the 150 days leading up to the effectiveness of EIP-4844.

This trend highlights the impact of cost reductions on revenue generation for rollups, indicating a potential challenge in achieving sustainable profitability in the new economic environment post-EIP-4844. While certain events have sparked temporary spikes in revenue, the overall downward trajectory emphasizes the need for rollups to adapt their strategies and explore new revenue streams to maintain viability in a competitive landscape.

The chart illustrates the net profit margins, calculated as revenue minus operating costs. Since the launch of EIP-4844, only Base and Scroll have encountered days of net loss:

  • Base is the only optimistic rollup that reported a day of net loss, suffering a loss of $175,000 on June 20, 2024. This loss was recovered, and the rollup continued to remain profitable in the following three days.

  • Scroll is the only ZK rollup that recorded negative profit margins on several days, including a loss of $25,000 on March 26, 2024, along with smaller losses on May 28, June 5, and June 6, 2024.

After the implementation of EIP-4844, rollups have achieved an average daily profit margin of $553,000. This figure is lower than the average daily margins of $685,000 in the 30 days prior, $389,000 in the 90 days, and $324,000 in the 150 days preceding the deployment of EIP-4844.

This analysis underscores the variability in profitability among different rollups, highlighting instances where specific rollups faced challenges in managing their costs effectively. Despite overall declines in average profit margins, the swift recovery of Base following its net loss illustrates the potential for resilience and adaptation within the rollup ecosystem post-EIP-4844.

Gross Margins Analysis

The gross margin percentage, expressed as a percentage, has seen a relative improvement following the implementation of EIP-4844, indicating that rollup companies are retaining more revenue than before. The chart breaks down the gross margins of optimistic and ZK rollup companies, highlighting the differences in how their utilization of blobs impacts each type’s profitability:

  • The average daily gross margin percentage for optimistic rollup companies has been 92.3% since March 13, 2024.

  • The average daily gross margin percentage for ZK rollup companies is 66.7% over the same period.

The additional costs associated with proofs for ZK rollup companies have reduced their gross margins compared to optimistic rollup companies.

Conclusion

The introduction of EIP-4844 and blobdata has brought significant improvements to the operational economics and utilization of Ethereum rollups. However, this change has resulted in a partial shift in revenue from Ethereum L1 and the rate of ETH burning towards rollup operators, also known as sequencers.

Before EIP-4844, the calldata model for data availability (DA) allowed Ethereum to retain up to 77% of the value generated by rollups. After the deployment of EIP-4844, Ethereum now retains only 8% of the value from optimistic rollups and 33% from ZK rollups, aggregating to (1 – Rollup Margin Percentage). This shift aligns with Ethereum’s objective to become an efficient DA layer, as rollups are now where users primarily conduct transactions and pay fees.

Despite the improvements in DA costs for Ethereum, the overall costs remain quite high, reaching $16,473 per gigabyte of blob data used and $12,458 per gigabyte purchased. High costs are primarily attributed to spikes in blob costs on certain days. When blob demand is low, costs become negligible. Excluding blob costs, type 3 transactions average only $5.22 per transaction. There are concerns that costs could escalate if blob demand remains high. A viable solution to reduce costs might be to increase the maximum number of blobs and the target per Ethereum block.