Layer 2 blockchain list: 7 scaling networks driving adoption

The harder question is whether lower fees, faster execution, and increasingly specialized architectures are sufficient to retain capital and users when the aggregate Layer 2 market has contracted so sharply.

In late July 2026, total value locked across Ethereum’s Layer 2 ecosystem fell to approximately $5 billion, down from an early-year peak of $48 billion. That is a decline of more than $43 billion. The contraction does not invalidate the scaling thesis, but it changes how we should evaluate the top Layer 2 networks: liquidity concentration, proof systems, sequencer design, interoperability, and application-specific utility now matter more than headline throughput.

This Layer 2 blockchain list focuses on seven active ecosystems with distinct architectural positions: Base, Arbitrum One, OP Mainnet, Mantle, Starknet, ZKsync Era, and Stacks. They are not interchangeable deployments of the same model. Some are optimistic rollups, some use zero-knowledge proofs, one is Bitcoin-native, and several are moving toward more independent infrastructure.

The 2026 TVL contraction changes the scaling debate

A Layer 2 network usually compresses transaction execution away from Ethereum and submits data, commitments, or proofs back to the base layer. The design reduces the cost of each transaction while preserving some connection to Ethereum’s settlement and security assumptions.

That mechanism remains valuable. However, the economic performance of a network depends on more than throughput. A chain can process transactions cheaply and still struggle if its applications cannot attract durable liquidity, if its bridge architecture creates excessive fragmentation, or if its token emissions subsidize activity that disappears when incentives decline.

The 2026 TVL contraction therefore deserves a more precise reading.

First, the decline indicates that capital is no longer distributed broadly across Ethereum’s scaling ecosystem. Optimistic rollups—Base, Arbitrum, and OP Mainnet—held approximately $4.8 billion of the remaining $5 billion in late July, or 96% of total Ethereum Layer 2 TVL. This is an unusually concentrated outcome for a market that previously presented itself as a field of competing rollup architectures.

Second, TVL is not a complete measure of network usage. It captures assets deposited into protocols, bridges, and applications, but it does not directly measure transaction count, developer activity, fee revenue, active addresses, or the quality of settlement guarantees. Nevertheless, a collapse of this magnitude cannot be dismissed as a cosmetic metric change. It reflects a substantial reduction in capital committed to Layer 2 environments.

Third, the destination of the capital that exited remains unclear. No major protocol team has published a definitive post-mortem identifying where the withdrawn liquidity went. Consequently, we should avoid treating the contraction as proof that Ethereum scaling has failed, while also avoiding the opposite mistake of assuming that all departing capital will return.

The strongest Layer 2 networks in 2026 are not necessarily those with the lowest fees; they are the networks that can convert technical capacity into persistent economic activity.

Optimistic rollup dominance: Base, Arbitrum, and OP Mainnet

Optimistic rollups still define the practical center of the Ethereum Layer 2 market. Their basic model assumes that submitted state transitions are valid unless challenged during a dispute period. Fraud-proof systems then provide a mechanism for contesting an invalid claim.

This approach has historically offered a comparatively accessible path for Ethereum-compatible applications. Developers can deploy Solidity-based contracts with limited changes, users can interact through familiar wallets, and liquidity can move through an established EVM environment.

The three leading optimistic rollups occupy different positions despite their shared category.

Base: distribution, simplicity, and increasing independence

Base has become one of the most visible Ethereum scaling ecosystems because it combines EVM compatibility with a broad distribution channel through Coinbase. That relationship gives Base access to users and infrastructure that many technically similar networks cannot easily reproduce.

Its architectural direction changed on February 18, 2026, when Base announced that it was migrating away from the OP Stack toward a self-managed “base/base” codebase. The stated purpose was to consolidate the sequencer, client, and proof systems under a more directly controlled architecture.

This is a significant shift in protocol governance and engineering responsibility. A network operating on a shared stack can benefit from common tooling, security research, and upgrades developed across an ecosystem. However, it also inherits design constraints and dependency relationships. A self-managed codebase potentially gives Base greater control over execution, sequencing, and proof-system development, although the final mainnet launch date for that architecture remains unknown.

Base’s early-2026 TVL was reported at $10.7 billion, a figure that illustrates the scale of capital it had accumulated before the broader contraction. We should not interpret that earlier level as the current state of the network, but it does demonstrate how quickly liquidity can concentrate around an ecosystem with strong user distribution and a recognizable application environment.

The key question for Base is whether its next architectural phase improves the network’s long-term control without weakening composability with Ethereum and the wider OP Stack ecosystem.

Arbitrum One: execution flexibility and fraud-proof progress

Arbitrum One remains one of the most technically mature optimistic rollups. Its position is supported by deep DeFi integration, a large developer base, and a multi-layer architecture that separates execution concerns from the settlement relationship with Ethereum.

Two developments are especially important.

Arbitrum Stylus allows developers to build contracts using languages such as Rust and C++ in addition to Solidity, while compiling them for execution within the Arbitrum environment. The upgrade is associated with performance gains of up to 70x for certain workloads. That figure should be understood as workload-dependent rather than as a universal improvement to every transaction, but the architectural implication is clear: Arbitrum is broadening the computational model available to application developers.

This matters because EVM compatibility alone does not guarantee efficient execution for every type of application. Gaming logic, order-book exchanges, high-frequency state updates, and cryptographic operations may benefit from languages and runtimes with different performance characteristics. Stylus gives developers more room to optimize those workloads without abandoning Ethereum settlement.

Arbitrum’s BoLD system, or Bounded Liquidity Delay, is designed to support permissionless fraud proofs. Permissionless dispute participation is an important decentralization milestone because security should not depend indefinitely on a small, curated group of entities willing to challenge incorrect state claims.

Arbitrum’s early-2026 TVL was reported at $16.8 billion, the largest figure among the networks listed in the available market snapshot. Its long-term position will depend less on preserving a historical TVL lead than on maintaining application depth, reducing operational dependencies, and converting technical upgrades into reliable user and developer experiences.

OP Mainnet: shared infrastructure and the limits of standardization

OP Mainnet is the reference ecosystem for the OP Stack, a modular framework used by multiple chains. Its importance therefore extends beyond the activity occurring directly on OP Mainnet. The network represents a model in which many deployments can share components for execution, governance, interoperability, and upgrades.

The advantage is scale through standardization. A common stack lowers the barrier for launching an Ethereum-compatible chain and can support a broader network of interoperable environments. Developers also gain access to familiar tooling, while infrastructure providers can reuse operational systems across deployments.

The trade-off is architectural dependence. If many chains rely on common components, a change in the underlying stack can have consequences across the entire ecosystem. The migration announced by Base demonstrates this tension: a major deployment may eventually seek more control over its codebase precisely because shared infrastructure imposes constraints on differentiation.

OP Mainnet’s early-2026 TVL was reported at $8 billion. As with Base and Arbitrum, that figure belongs to the period before the late-July market contraction. Its relevance is primarily comparative: the three optimistic rollups together accounted for most of the capital still held in Ethereum Layer 2 networks after the decline.

Mantle and Stacks: scaling beyond the standard Ethereum rollup template

The next group illustrates how Layer 2 ecosystems are differentiating through modularity and alternative settlement relationships.

Mantle: modular architecture with a DeFi center of gravity

Mantle Network uses a modular Layer 2 architecture and ZK proofs, while positioning itself as a major environment for decentralized finance. In March 2026, Mantle crossed $1 billion in DeFi TVL and reached a stablecoin market capitalization of $980 million.

These figures indicate a more specific form of adoption than general wallet activity. DeFi liquidity requires bridges, lending markets, collateral systems, stablecoins, and protocols capable of supporting continuous capital movement. A network that reaches this scale has established at least a meaningful financial application layer, even if the durability of that liquidity still depends on incentives and market conditions.

The distinction between “modular” and “monolithic” architecture is useful here. A modular network separates functions such as execution, data availability, settlement, and proof generation across specialized components. This can improve flexibility and allow the network to optimize each layer independently. The cost is greater system complexity: more interfaces must remain secure, correctly synchronized, and economically viable.

Mantle’s challenge is therefore not merely to demonstrate ZK proof capability. It must show that modularity improves the user-facing experience and creates sustainable fee generation rather than adding infrastructure complexity that remains invisible to users but expensive to maintain.

Stacks: Bitcoin-native DeFi and a different settlement thesis

Stacks occupies a different position from Ethereum-focused rollups. It is a Bitcoin Layer 2 network designed to bring programmable applications and DeFi activity into the Bitcoin ecosystem. Its growth is closely tied to sBTC, a mechanism intended to represent Bitcoin within Stacks applications.

Stacks reached $545 million in sBTC TVL during the first quarter of 2026 following its Nakamoto upgrade. The figure is important because it measures the extent to which Bitcoin liquidity is being used in an application environment rather than held only as a passive asset.

The network’s scaling thesis is consequently not “Ethereum, but cheaper.” It is “Bitcoin as a settlement and collateral base for programmable activity.” That distinction creates both an opportunity and a constraint. Bitcoin offers a large and culturally significant asset base, but the surrounding application infrastructure is less mature than Ethereum’s in several areas. Bridges, custody assumptions, finality models, and liquidity routing all require careful examination.

If we look at Stacks through a tokenomics and protocol-design lens, sBTC adoption is more informative than generic transaction growth. The system must attract Bitcoin liquidity while preserving a credible relationship between the represented asset and the underlying network. The value proposition depends on that relationship remaining understandable and robust as application complexity increases.

Bitcoin-native scaling is not a variant of Ethereum scaling; it is a separate settlement strategy with its own bridge, custody, and liquidity risks.

ZKsync Era and Starknet: proof systems become product features

Zero-knowledge rollups compress transaction data and use validity proofs to demonstrate that a batch was executed correctly. Unlike optimistic systems, they do not rely on a challenge period to establish validity, although the practical security model still depends on implementation quality, proving infrastructure, upgrade controls, and data availability.

The category is not uniform. ZKsync Era and Starknet use different execution environments and development models, which affects compatibility, tooling, and application migration.

ZKsync Era: ecosystem consolidation after ZKsync Lite

Matter Labs is shutting down ZKsync Lite in 2026 and directing users toward ZKsync Era. This is a migration event, but it is also an architectural consolidation.

Legacy networks can remain operationally expensive when they preserve older execution models, bridge systems, and tooling that no longer align with the primary protocol roadmap. Moving users and assets into ZKsync Era can reduce fragmentation, but migrations introduce their own risks. Users must understand destination contracts, bridge procedures, withdrawal timing, and application compatibility. Developers may also need to adapt contracts when the execution environment or supported features differ from the legacy chain.

For ZKsync Era, the central strategic question is whether consolidation produces a stronger active ecosystem or merely shifts assets between related networks. A healthy migration should improve application density, liquidity routing, developer support, and protocol maintainability.

The broader lesson is that a Layer 2 blockchain list should include operational status, not only technical branding. A network may have a recognizable name and a well-known proof system, but its relevance depends on whether users are being directed toward it as an active destination.

Starknet: custom execution, staking, and privacy expansion

Starknet is built around validity proofs and a custom execution environment rather than direct EVM equivalence. This architecture gives the protocol greater freedom to optimize its virtual machine and programming model, but it also creates a steeper migration path for developers accustomed to Solidity and Ethereum’s standard tooling.

Its 2026 upgrades move beyond raw scalability. Starknet launched liquid staking on July 29, creating a mechanism through which users may seek staking exposure without fully immobilizing their assets. Liquid staking can improve capital efficiency, although it also introduces additional smart-contract, liquidity, and redemption risks.

Starknet also introduced the STRK20 privacy standard for shielded transfers in mid-2026. Privacy functionality changes the design requirements of a network because balances, transfer validity, compliance expectations, and application composability must be considered together. Shielded transfers can protect transaction information, but they also require clear wallet support, auditing, and carefully designed interfaces so that users understand what is private and what remains observable.

The protocol’s long-term value will depend on whether its custom architecture supports applications that benefit from proof-oriented execution, rather than merely asking developers to accept a less familiar environment in exchange for theoretical scalability.

Network sunsets are part of the Layer 2 lifecycle

The Polygon zkEVM Mainnet Beta sequencer was officially sunset on July 3, 2026. Block production stopped, and Agglayer Bridge withdrawals were halted. It should therefore not be treated as an active network in a current list of leading Layer 2 ecosystems.

This event matters because it exposes a part of blockchain infrastructure that market narratives often omit: networks have lifecycles. Launches receive attention, but closures, migrations, bridge deprecations, and sequencer transitions determine whether users can continue to access their assets safely.

A network sunset can occur for several reasons: insufficient activity, an architectural pivot, duplicated infrastructure, unsustainable operational costs, or a strategic decision to concentrate resources elsewhere. None of these explanations should be inferred without a formal post-mortem. What can be stated is that deployment status is a core analytical variable.

For users and developers, the practical implications are direct:

1. Confirm whether the sequencer is still producing blocks. A branded network may remain visible in wallets and documentation after its active infrastructure has been retired.

2. Separate bridge availability from application availability. A bridge may support deposits while withdrawals are delayed, restricted, or routed through a migration contract.

3. Check the destination architecture during migrations. ZKsync Lite’s transition toward ZKsync Era demonstrates that a successor network may not provide identical execution behavior or contract compatibility.

4. Treat sequencer design as part of the trust model. Permissioned sequencing, permissionless fraud proofs, proof-generation controls, and emergency upgrade rights all affect how a network behaves during failure or transition.

5. Evaluate state bloat and archival requirements. A chain that accumulates large amounts of state can become more difficult and expensive for independent operators to maintain, regardless of its nominal transaction throughput.

What separates active Layer 2 ecosystems from temporary liquidity venues

The current top Layer 2 networks 2026 landscape should be assessed through several interacting dimensions rather than a single ranking.

Throughput is the visible metric, but it is only the first layer of the analysis. We also need to examine whether the chain can sustain execution under load, whether transaction ordering is transparent, and whether the cost of proving or publishing data scales efficiently.

State bloat is equally important. Every application and account adds data that validators, sequencers, provers, or full nodes may eventually need to process. A network that reduces transaction fees by externalizing operational costs can appear efficient in the short term while becoming harder to validate over time.

The emission curve also matters. Layer 2 tokens and ecosystem incentives can attract liquidity rapidly, but emissions create a recurring liability. If rewards exceed fee revenue and genuine application demand, the network may produce temporary TVL rather than durable economic activity. A healthier model links incentives to useful services, developer retention, security participation, or measurable application revenue.

A practical blockchain scalability comparison should therefore include:

• Settlement assumptions: Whether the network settles directly to Ethereum, Bitcoin, or another base layer, and what finality means operationally.
• Proof mechanism: Whether the system uses fraud proofs, validity proofs, or a hybrid model, including how disputes and upgrades are managed.
• Sequencer structure: Whether sequencing is centralized, distributed, permissionless, or governed by a restricted set of operators.
• Data availability: How transaction data is published, retained, and made available for independent verification.
• Execution compatibility: Whether developers can use standard EVM tooling or must adopt a custom language and virtual machine.
• Liquidity quality: Whether TVL reflects productive lending, trading, and collateral activity or primarily short-term incentive deposits.
• Interoperability: Whether assets can move across ecosystems without creating excessive bridge, messaging, or trust assumptions.
• Token sustainability: Whether emissions support security and adoption without overwhelming fee generation or diluting long-term participants.

These dimensions explain why the seven networks in this list should not be reduced to a simple “best” and “worst” hierarchy. Base and OP Mainnet emphasize ecosystem distribution and shared infrastructure. Arbitrum is pushing execution flexibility and fraud-proof decentralization. Mantle is building around modularity and DeFi liquidity. Starknet and ZKsync Era represent different approaches to ZK-based scaling. Stacks pursues Bitcoin-native programmability rather than Ethereum compatibility.

The long-term implication: scaling must become economically self-supporting

The 2026 contraction has removed an easy assumption from Layer 2 analysis: that every credible scaling network will benefit automatically from Ethereum growth. Capital is more selective, network architectures are converging in some areas and diverging in others, and users are increasingly exposed to migration risk when protocols sunset or restructure.

The most durable networks will need to demonstrate three forms of efficiency simultaneously.

They must improve execution efficiency through higher throughput and lower per-transaction costs. They must improve capital efficiency by supporting applications that use liquidity productively rather than merely locking it. Finally, they must improve protocol efficiency by reducing dependency on centralized operators, unmanaged state growth, fragile bridges, and inflationary incentives.

If we look at the current Layer 2 blockchain list through that framework, there is no single winning architecture. Optimistic rollups retain most of the remaining Ethereum Layer 2 value because they combine maturity, EVM compatibility, and established application liquidity. ZK networks continue to develop stronger proof-oriented infrastructure, but their competitive position depends on proving costs, developer usability, and reliable execution environments. Bitcoin-native systems such as Stacks are expanding the definition of what a Layer 2 can be by targeting a different settlement base and liquidity source.

Consequently, the next stage of Layer 2 adoption will be determined less by launch announcements and more by operational durability. The networks that survive the contraction will be those that turn technical capacity into persistent usage, maintain credible security assumptions, and manage their emission curves and state growth well enough to remain useful after incentives fade.