In the first part of Web3Caff Research’s 2026 Ethereum outlook, researcher ShirleyLi reviews Ethereum through the lens of almost twelve years of roadmap changes. Since Vitalik Buterin and his team formally introduced Ethereum to global users at an international conference in 2014, the network has moved from a niche experiment to one of the most influential base platforms in Web3. The report frames Ethereum as a system that must keep the network stable while continuing to define new technical directions for a growing ecosystem.
From early phases to the 2026 Strawmap draft
Between 2014 and 2016, Ethereum’s early roadmap took shape around four phases: Frontier, Homestead, Metropolis and Serenity. The first three are usually regarded as Ethereum 1.0, focused on basic functions and network stability. Serenity represented the longer-term goal of reworking the consensus mechanism and the underlying architecture in order to improve scalability and performance.
In 2020, Ethereum clarified the technical path for Serenity by formally committing to a transition to proof of stake and by introducing sharding logic. In 2022, a more complete medium- and long-term roadmap established Rollups as the core scaling path. Under that design, execution would increasingly move to Layer 2 networks, while the main chain would focus on security and data availability. In February 2026, the Ethereum Foundation released a “Strawmap” draft for the coming decade, with more specific optimization targets across the consensus layer, data layer and execution layer.
The article links these roadmap changes to the questions raised in the author’s late-2024 report, “Ethereum’s Future Road: Development Amid Controversy, Can the Ecosystem Giant Resist a Potential Crisis?” The central tension remains the same: Ethereum needs to preserve stability for a large ecosystem, while also changing enough to keep that ecosystem moving forward. Each roadmap is treated as a temporary balance among scalability, security, decentralization and ecosystem value distribution.
Foundation restructuring and tighter resource allocation
The report says that in early 2025, while the broader market environment improved and new narratives emerged, Ethereum’s own development pace appeared relatively slow to parts of the community. Critics argued that the Ethereum Foundation and core developers were lagging behind the industry in execution speed, market communication and ecosystem expansion. The Foundation then made a series of major internal changes.
In February 2025, Aya Miyaguchi, who had served as executive director since 2018, moved into the newly created role of President. Her responsibilities shifted away from daily operations and execution management toward external partnerships, institutional relations and cultural communication. Nethermind founder Tomasz Stańczak and Hsiao-Wei Wang became co-executive directors. Under the new structure, the Foundation streamlined its organization, cut 19 employees and moved its strategic focus back from Layer 2 to Layer 1. It also began placing more emphasis on external communication and transparency around technical direction, development priorities and resource use.
In June 2025, the Foundation reorganized its internal research and development system. The former “Protocol Research & Development,” or PR&D, was simplified to “Protocol.” Its short-term goals were to scale L1 performance, scale Blobs and improve user experience. The change marked a move from a primarily research-oriented structure toward engineering implementation and delivery. Earlier this year, the Protocol team further upgraded and clarified its work targets.
In February 2026, Tomasz Stańczak announced that he would step down as co-executive director. Bastian Aue and Hsiao-Wei Wang took over the role. During his tenure, Stańczak promoted exploration in privacy protection, quantum-computing security and the combination of AI and Ethereum. After stepping down, he said he would devote more energy to products and infrastructure related to AI and blockchain. The report also notes that the Foundation’s latest board includes Vitalik Buterin, Aya Miyaguchi, Patrick Storchenegger and Hsiao-Wei Wang, who are responsible for governance and strategic direction, while management and functional teams handle execution and operations.
The funding system also changed. In August 2025, the Foundation paused the open grant program that had operated since 2018, and in November relaunched the Ecosystem Support Program, or ESP. The allocation model shifted from passively receiving applications to actively guiding funding. The first funded areas covered cryptography, privacy, the application layer, security and community growth. The Foundation also decided to reduce annual spending from around 15% to 5% of funds, slowing the consumption of ETH reserves. In May this year, researchers Carl Beek and Julian Ma announced their departures, while former Foundation researcher Dankrad Feist publicly said the ecosystem needed a new organization more aligned with Ethereum’s economic interests to “save” Ethereum. Vitalik Buterin and Ethereum co-founder Joe Lubin responded that the disputes reflected the friction between Ethereum’s long-term technical building culture and the current commercialization process.
PoS staking, DVT and Blob capacity
Ethereum’s move to proof of stake removed the high-energy consensus model, but the 32 ETH staking threshold also raised the barrier for validators and created concerns about concentrated validation power. If the minimum stake for a single validator is lowered, the network then needs to handle the communication and coordination costs that come with a larger validator set, while also raising the cost of malicious behavior. Vitalik Buterin previously suggested that raising the participation threshold required for block finalization, for example from the current roughly two-thirds threshold signature level to 75% or higher, could strengthen security.
The Pectra upgrade was activated on Ethereum mainnet in May 2025. EIP-7251 raised the maximum effective balance of a validator from 32 ETH to 2048 ETH. The 32 ETH minimum to become a validator did not change. The proposal instead raised the amount of ETH that a single validator can have counted toward consensus weight, allowing one validator to represent more ETH directly in voting. Large stakers no longer need to split holdings across many validator nodes to receive corresponding incentives, which can reduce the situation in which one entity controls multiple validator nodes and lower communication and coordination overhead in the consensus process.
EIP-7002 optimized staking withdrawals by introducing a withdrawal path triggered from the execution layer. Under specific conditions, stakers can complete withdrawals without requiring the validator to actively sign. The report says this strengthens stakers’ control over assets, reduces the operational complexity of entering and exiting staking, and increases the flexibility of the PoS system. The Foundation also explored the use of Distributed Validator Technology, or DVT. DVT splits a single validator’s private key and signing capability across multiple nodes, reducing single points of failure. Since multi-node cooperation adds complexity, the Foundation has tried lighter implementations such as DVT-lite. Information disclosed by the Foundation in March 2026 showed that 72,000 ETH was staked through related mechanisms.
On data availability, EIP-4844 in the Dencun upgrade introduced Blob space as a lower-cost temporary data availability area. Layer 2 networks can submit batches of transaction data to Blobs, reducing the cost of publishing data onchain. The original design targeted 3 Blobs per block, with a maximum of 6. A floating fee mechanism raises the base fee when Blob use is above target and lowers it when use is below target. As Layer 2 demand for Blob space rose, actual Blob usage approached the target and repeatedly triggered fee increases.
To ease Blob capacity pressure, Pectra included two related proposals. EIP-7691 raised the target Blob count per block from 3 to 6 and the maximum from 6 to 9. EIP-7623 sought to increase the cost of Calldata, guiding Layer 2 networks toward greater Blob use and reducing pressure on mainnet data publishing. With the Fusaka upgrade in December 2025, Ethereum introduced additional mechanisms at the data availability layer. The report also states that on December 11, 2025, the Ethereum Foundation said Blob capacity per block had been raised to 15. In the longer technical path, Blobs are presented not as an isolated feature but as part of Ethereum’s move from execution sharding toward data sharding, with PeerDAS further improving the way Blob data is verified.
Rollups, fragmentation and Ethereum’s attempt to reconnect the stack
The report describes a structural change in the relationship between Rollups and Ethereum mainnet. Rollups reduce the execution load of the main chain by moving transaction execution to Layer 2, but they also change the fee distribution structure. More user transaction fees flow to Layer 2 networks, while mainnet mainly provides data publishing and settlement. Different Rollups have also developed as relatively independent ecosystems, increasing the complexity of cross-Rollup interaction.
Several approaches have emerged to address fragmentation. Interoperability protocols such as LayerZero and Hyperlane try to standardize message passing between Rollups. Shared sequencer projects such as Astria and Espresso start from the transaction-ordering layer and seek to provide a unified sequencing service for multiple Rollups. The report stresses that this “Rollup — third-party infrastructure — Rollup” communication model is still a non-native Ethereum extension path. It introduces additional trust assumptions and can create new security questions. Shared sequencers also participate in value capture, affecting the MEV distribution structure of Layer 1.
Different Rollup technology stacks are building their own ecosystems. Arbitrum is promoting Orbit chains through its Nitro / Orbit stack, with Plume Network and ApeChain named as examples. zkSync supports multi-chain expansion through the Elastic Chain architecture, with Abstract and ZERO Network as representative projects. Optimism has built the Superchain around OP Stack, with members including OP Mainnet, Base, BOB and Soneium. In the second half of February this year, the Base team announced that it would gradually move away from the Optimism OP Stack architecture and toward a unified Base Stack maintained by Base itself. Because Base was originally built with OP Stack, the report describes the process as a customized extension on top of an open-source framework rather than a complete split from Optimism. Base also said it would continue working with the Optimism ecosystem and follow relevant open-source norms.
The Ethereum Foundation has begun to address this issue more directly. While continuing to optimize Rollup data availability, it created a Platform Team under the Protocol framework to coordinate protocol research and ecosystem development. The team is intended to reshape the relationship between mainnet and Layer 2, moving them from a delicate alliance toward a more coordinated and mutually reinforcing platform.
On March 23 this year, the Foundation published a systematic explanation of how the roles of Layer 1 and Layer 2 are evolving. In the new framing, Layer 2 is no longer only for scaling. It also serves differentiated needs that Layer 1 does not provide, such as stronger privacy, lower latency and compliance adaptation for specific scenarios. Its growth is meant to create greater value for mainnet rather than compete with Layer 1. At the same time, Ethereum mainnet is positioned as the core settlement layer and liquidity layer for a multi-chain ecosystem.
Two paths are highlighted. The first is to strengthen the connection between Rollups and Ethereum mainnet. With Ethereum Foundation funding, Gnosis co-founder Friederike Ernst and Zisk founder Jordi Baylina recently announced the Ethereum Economic Zone, or EEZ. The plan attempts to build an L1<>L2 framework in which interactions between Rollups and mainnet, and among Rollups themselves, are included in a unified execution framework. The design aims for a stronger form of composability, including atomic operations across execution environments. The second path is Native Rollups, proposed in EIP-8079. A recent proof-of-concept prototype tries to expose Ethereum’s state transition function as an interface for Rollups, allowing them to submit transactions to mainnet for state computation under unified rules and reducing reliance on separate proof systems.
ePBS, Glamsterdam and faster confirmation
After Ethereum’s transition to PoS, the network randomly selects block proposers from validators staking ETH. A proposer is responsible for publishing a block, and can extract MEV by changing the order of transactions inside that block. To reduce a single validator’s control over transaction ordering and value extraction, Ethereum previously proposed proposer-builder separation, or PBS. Under PBS, specialized builders package transactions, while proposers only choose from multiple candidate blocks and submit the final block.
The report explains that this specialization improves efficiency but also introduces structural problems. As block-building capability concentrates among a small number of builders, their influence over transaction ordering and MEV allocation grows. In practice, PBS has not been written directly into the protocol. It has instead been implemented through third-party middleware such as MEV-Boost, which allows validators to outsource block building to third-party builders and pass information through relays. Although this improved block-building efficiency, it remained an offchain market that depends on trusted relays, carries centralization risk and lacks full onchain transparency and constraints around MEV distribution.
Glamsterdam is planned to restructure the responsibilities of different network participants. Its core directions include introducing parallel processing capability as preparation for faster transaction execution, separating the workflows of block creation and block verification to provide more time for data propagation and larger data loads, and adjusting the fee mechanism so that onchain data storage costs better reflect long-term resource use such as bandwidth and storage.
One of the most important planned changes is EIP-7732, or enshrined proposer-builder separation, ePBS. It writes PBS into the Ethereum protocol, makes builders formal participants, requires them to stake, and requires them to submit block commitments. It also decouples consensus verification from execution verification. During the key stage of block propagation, the network can prioritize consensus-layer validation and defer full execution-layer validation, including transaction execution and state updates, to a later stage. This reduces reliance on relays and gives nodes more time to propagate block data, supporting larger data loads, especially Blob writing, while maintaining network stability.
High-performance L1 competition, SSF and FCR
The report also places Ethereum’s Rollup-centric path beside other public-chain designs. Solana is described as improving overall processing capacity through parallel execution and local fee markets. Sui is described as using an object model and DAG architecture to achieve more efficient concurrent processing in specific scenarios. These features can serve use cases such as high-frequency trading and onchain games. High-performance Layer 1 networks can provide low-cost, high-throughput execution directly on a single chain, reducing user reliance on cross-Layer 2 operations. Compared with a Layer 1 plus Layer 2 system, ordinary public chains can also offer a more integrated development environment and execution model, while Ethereum developers must handle cross-chain communication and liquidity fragmentation between Layer 1 and different Layer 2 networks.
Ethereum’s long-term answer includes single-slot finality, or SSF. In Ethereum, validators reach consensus roughly every 12 seconds on the transactions and order that can be included in a block. This time window is called a Slot. Every 32 Slots, or about 6.4 minutes, form an Epoch. Under the current mechanism, a block needs at least two-thirds of validators to complete two rounds of voting, across two Epochs, before it is considered finalized and no longer reversible. A block therefore usually takes about 2 Epochs, or 12 to 15 minutes, to reach finality. SSF aims to compress the finality logic that currently spans two Epochs into a single Slot.
The difficulty is that simply reducing the validator count or raising hardware requirements would weaken decentralization. The key bottleneck is optimizing verification and communication so that nodes can process more signatures and validation work in the same time window. Ideas include a “super committee” mechanism that randomly selects a subset of validators in a Slot to vote and confirm quickly, or adjustments to validator participation and weight distribution. These approaches also add technical complexity in collecting, aggregating and verifying large-scale signatures within a very short period.
Before SSF is fully implemented, the Foundation has proposed an interim optimization called the Fast Confirmation Rule, or FCR. Its goal is to reduce confirmation times for deposits from Ethereum L1 to L2s and centralized exchanges from several minutes to about 13 seconds. It does not change the existing finality mechanism. Instead, it uses the first round of voting earlier to judge safety. When the first round has already reached a high proportion, such as clearly above two-thirds, the system can treat the block as basically confirmed. FCR is not the same as true finality. Its safety depends on most validators behaving honestly and on stable, low-latency network communication. If the network is congested or attacked, a quickly confirmed block can still be reorganized under the original rules.
The report closes with a disclaimer that its information is for reference only and does not constitute a forecast, investment advice, proposal or offer. It states that the views reflect the author’s personal opinions as of the stated date, are not the position of Web3Caff Research, and that readers should comply with the laws and regulations of their own jurisdictions.