Adoption, Economics, and Risks of Ethereum Rollups

Public, permissionless blockchains have faced the challenge of limited transaction throughput and high fees during periods of network congestion. To address these constraints, the ecosystem has turned to Layer 2 (L2) scaling solutions such as state channels and rollups. The general idea of these scaling solutions is straightforward: move most of the computational workload off-chain, while still posting essential data and proofs back to the underlying Layer 1 (L1). The L1 then acts as a final “court,” ensuring that users can verify execution and challenge invalid activity if necessary.

Ethereum, currently the leading smart contract platform in terms of value secured, has focused its scaling strategy on so-called rollups. In short, rollups are a type of L2 implemented as a set of smart contracts on L1. Through periodic commitments and proofs, they can (in theory) inherit the base layer’s security guarantees. The details greatly depend on the specific design and assumptions, for example fraud/validity proofs, smart contract upgradeability, and system restrictions. (Schär F., 2024)

Over the past year, adoption of rollups has accelerated significantly. Measured in transaction count*, rollups now account for the majority of Ethereum-related activity. In September 2025, all rollups combined processed around 230 transactions per second (TPS), compared to approximately 20 TPS on Ethereum’s base layer. This corresponds to a scaling factor of 12.3× (L2 Beat - Activity).

Among the various rollups, a few stand out in terms of adoption: Arbitrum One, OP Mainnet, and Base. Within this group, Base has shown particularly strong growth. In fact, Base currently processes more transactions than the other major rollups combined, accounting for a substantial share of rollup activity in recent months. This rapid adoption raises several questions: How does Base achieve its level of throughput? What revenue streams does it generate? And how are transaction costs for users determined? To answer these questions, we take a closer look at Base’s technical and economic model.

Base Rollup: Throughput, Revenue, and Fees
Base’s rapid growth can be attributed to its design choices. By adopting shorter block times and higher gas limits than Ethereum L1, Base achieves substantially higher throughput. Blocks are produced every two seconds, with a current gas target of 75 million units per block. By comparison, Ethereum L1 produces a block roughly every twelve seconds with a gas target of 22.5 million units. These parameters imply nearly a 20× increase in throughput. Moreover, Base has demonstrated a willingness to adjust these parameters upward in response to demand (Base Engineering Blog).

Like most rollups, Base is operated by a centralized sequencer (sometimes called operator). The sequencer orders transactions, produces blocks, and submits commitments to Ethereum. Sequencers generate revenue primarily through transaction fees. On Base, these fees are collected at the sequencer’s address, which can be monitored on-chain. The fee system follows Ethereum’s EIP 1559model, but with a key difference: on L1, the base fee is burned, whereas on Base the sequencer retains it. The priority fee serves as a tip to incentivize inclusion and can also capture value through MEV (maximal extractable value). Thus, the sequencer captures both the base and priority components of L2 execution fees, while also controlling transaction ordering, which can create additional value through MEV strategies.

From the user’s perspective, the cost of a transaction on Base is determined by two components: the execution fee and the L1 data fee. The execution fee represents the cost of running the transaction within the rollup itself. It is computed in the same way as on Ethereum: the gas price (the sum of the base fee and priority fee) multiplied by the gas used. This directly measures the computational resources consumed on Base.

The second component, the L1 data fee, arises from the need to publish transaction data to Ethereum for security. Transactions executed on Base are bundled and then posted to Ethereum in compressed form (as blobs), ensuring that they can always be verified or challenged. The user pays this cost, which is computed as:

L1 Data Fee = L1 Gas Used × L1 Base Fee × L1 Base Fee Scalar / 10^6.

For example, a simple Ether transfer on Base is approximated with 1,600 units of L1 gas. The L1 base fee is read from the latest relevant Ethereum block, and the fee scalar is currently set to 2269. These parameters are determined by the operator and do not necessarily correspond to the sequencer’s actual costs. Operators can further optimize costs by adjusting the interval or timing at which data and proofs are submitted to L1. In most cases, the L1 data fee contributes only a small fraction of the user’s transaction fee total, but it becomes more significant for transactions with larger calldata, i.e., more contract execution instructions. The flexibility of parameters may create a gap between the nominal costs paid by users and the actual costs borne by the operator. Over time, the way rollup operators set and adjust such parameters could also become an important factor in competition, if users and developers are sensitive to small differences in transaction fees across platforms.

To summarize, Base achieves high throughput by raising gas limits and reducing block times, generates revenue through a centralized sequencer that captures both fees and MEV opportunities, and passes costs to users through a two-tiered fee model. While this design enables significant scaling, it also creates a system that is more opaque and dependent on operator discretion than Ethereum L1. These trade-offs set the stage for broader questions about competition, decentralization, and the long-term sustainability of rollup economics.

Outlook
While rollups have clearly improved Ethereum’s scalability, their current designs also raise questions. The economics of rollup sequencers are less visible, which makes ongoing monitoring and analysis essential in order to understand the real costs borne by users and the revenues captured by operators.

Another issue is competition among rollups. If one solution were to capture a dominant share of activity, it could gain significant market power. In such a scenario, the promise of cheaper transactions might give way to monopolistic pricing dynamics, particularly since users face switching costs when moving between ecosystems. The extent to which competition between rollups can prevent such outcomes remains an open question. The very success of rollups may fragment the ecosystem. Loss of composability - the ability for applications to interact seamlessly within the same execution environment - becomes a risk when liquidity and activity are spread across multiple rollups.

The role of sequencers presents additional challenges. Centralized sequencers not only capture transaction fees but also control transaction ordering. This opens the door to MEV, where sequencers can profit from reordering or inserting transactions.

Additionally, the pursuit of high throughput has trade-offs. Running a node for a rollup with very short block times and high gas limits demands substantial computational and bandwidth resources. Construction of validity proofs or participation in fraud-proof challenges is also costly. This raises the barrier to participation, increasing reliance on large servers and professional operators. Over time, this trend could reduce decentralization and shift trust assumptions back toward centralized or trusted entities, potentially undermining the core benefits of public, permissionless blockchains.

To summarize, rollups represent an important step forward in scaling blockchain systems, but they also introduce new economic, governance, and technical challenges. Overcoming the mentioned challenges will be crucial for realizing the scalable yet resilient decentralized infrastructure.

Footnote:
*Transaction counts are not an ideal measure of network activity. In some systems, even consensus-related votes may be recorded as transactions, inflating throughput figures. Moreover, a “transaction” on a smart contract platform can represent anything from a simple asset transfer to a complex multi-contract interaction. Still, for comparisons across Ethereum Virtual Machine (EVM) chains, transaction counts provide a reasonable benchmark.

About the Author
Article authored by Dario Thürkauf, PhD Candidate DLT / FintTech at Center for Innovative Finance, University of Basel