Ethereum Layer 3s: Application-Specific Chains on Top of L2s

Ethereum's scaling roadmap has produced dozens of Layer 2 rollups, each posting compressed transaction data back to the L1 for security. But a growing number of projects are now building one level higher: Layer 3 chains that settle to an L2 rather than directly to Ethereum itself. The thesis is straightforward: certain applications need their own execution environment with custom gas tokens, fee structures, or throughput guarantees that general-purpose L2s cannot provide.

This article examines the Layer 3 landscape: the technical stacks enabling it, the use cases driving adoption, the economic model underpinning the hierarchy, and the fragmentation risk that comes with proliferating chains.

Why Not Just Deploy on an L2?

An application can deploy smart contracts on any existing L2 and inherit its throughput, fee market, and user base. For most DeFi protocols, NFT marketplaces, and token contracts, that is the right choice. But certain categories of applications hit limits that shared L2 execution cannot solve.

• Gaming applications need sub-cent transaction fees and high throughput without competing for block space with DeFi arbitrage bots and MEV searchers
• Enterprise deployments may require permissioned execution environments where only authorized participants can submit transactions, while still anchoring settlement to a public chain
• DeFi protocols with custom fee structures (rebates, gas subsidies, or application-specific token fees) need control over the fee market itself, not just the contract logic
• Privacy-focused applications may need custom precompiles or execution environments that support zero-knowledge computations natively

In each case, the application does not merely need cheaper transactions. It needs sovereignty over the execution environment while retaining a connection to Ethereum's security guarantees. That is the appchain thesis, and L3s are Ethereum's answer to it.

Vitalik's framing: Ethereum co-founder Vitalik Buterin has argued that L3s should not be viewed as "scaling on top of scaling." Stacking the same rollup compression twice does not yield multiplicative gains. Instead, L3s make sense when each layer serves a different purpose: L2 for general scaling, L3 for customized functionality such as privacy, application-specific execution, or weakly-trusted data availability via Validiums.

The L3 Technology Stacks

Four major frameworks currently support launching application-specific chains that settle to an Ethereum L2. Each takes a different approach to proof systems, interoperability, and data availability.

Arbitrum Orbit

Arbitrum Orbit is the most widely adopted L3 framework. Orbit chains can be configured as L2s settling directly to Ethereum or as L3s settling to Arbitrum One or Arbitrum Nova. They come in three flavors: Rollups (posting full data to the parent chain), AnyTrust (using a Data Availability Committee for cheaper storage), and custom configurations. Orbit chains run the Arbitrum Nitro execution engine and support the EVM natively, making migration from existing Ethereum or Arbitrum deployments straightforward.

OP Stack and the Superchain

The OP Stack powers the Optimism Superchain and supports L3 deployments that settle to any OP Chain. An L3 on the OP Stack submits transactions to its parent L2 rather than to Ethereum, reducing the fixed overhead of posting data commitments and output roots. Rollup-as-a-Service providers like Conduit have offered L3 deployment on Base since early 2024, using Celestia for data availability to further reduce costs.

ZK Stack and the Elastic Chain

Matter Labs' ZK Stack is an open-source framework for launching ZK-powered chains. Originally branded "Hyperchains," the architecture was renamed to the Elastic Chain alongside the ZKsync 3.0 upgrade in mid-2024. The v24 upgrade restructured ZKsync from a single rollup into a network of interconnected ZK chains. A chain built on ZK Stack can operate as an L3 by settling its validity proofs to ZKsync Era (itself an L2), with options for proof aggregation across multiple chains. The key differentiator is the use of ZK proofs rather than fraud proofs, which eliminates the seven-day challenge window that optimistic rollups require.

Polygon CDK and AggLayer

Polygon CDK allows teams to deploy ZK-powered chains that connect to the AggLayer, a shared proof-aggregation and settlement layer. As of mid-2025, CDK supports both its native Erigon-based stack and an OP Stack configuration, with the AggLayer accepting proofs from EVM and non-EVM chains alike. While CDK chains typically operate as L2s settling directly to Ethereum, the AggLayer architecture could support deeper nesting in the future.

Comparing the L3 Frameworks

Use Cases Driving L3 Adoption

Gaming has emerged as the primary driver of L3 deployments. But the appchain model extends to DeFi, enterprise applications, and social platforms as well.

Gaming: Dedicated Throughput

Gaming is the category most naturally suited to L3 deployment. On-chain games generate high volumes of low-value transactions: player movements, item crafts, reward distributions. These transactions are economically unviable on Ethereum L1 and still expensive on congested L2s. Xai is the most prominent example: an Arbitrum Orbit L3 using AnyTrust that launched with direct support from Offchain Labs. Xai's model lets the Xai Foundation subsidize gas fees for games in its ecosystem, removing the fee barrier for players entirely.

DeFi: Custom Fee Markets

DeFi protocols with specialized execution needs can benefit from L3 deployment. Reya Network, built on Arbitrum Orbit, operates as a purpose-built L3 for perpetual futures trading. By controlling the execution environment, it can optimize for order matching, implement custom fee rebate structures, and ensure that MEV extraction does not degrade the experience for traders. Similarly, Orbs Network operates as a DeFi middleware layer, running decentralized backends for protocols across multiple L1s and L2s.

Social and Community Chains

Degen Chain demonstrated a different model: a community-driven L3 built on Arbitrum Orbit, settling to Base. Launched in early 2024 with help from Syndicate, Degen Chain uses the DEGEN token as its native gas token and targets tipping, community rewards, and social applications from the Farcaster ecosystem. The chain illustrates how L3 deployment costs have fallen low enough that even a memecoin community can launch and operate its own chain.

Enterprise: Permissioned Execution

Enterprise applications often require controlled access: only vetted participants can submit transactions, but the settlement and audit trail should be publicly verifiable. An L3 can restrict its sequencer to accept transactions only from authorized addresses while posting state roots to a public L2. This gives compliance teams the permissioning they need without sacrificing the transparency and finality guarantees of Ethereum settlement.

The Economic Model: Settlement Fees in Layers

The L1-L2-L3 hierarchy creates a layered fee structure. Understanding where costs accrue is critical for evaluating whether an L3 deployment makes economic sense.

How fees flow

An L3 pays its parent L2 for two things: execution (including sequencer fees) and data availability. The L2, in turn, pays Ethereum L1 for its own settlement and data posting. After EIP-4844 introduced blob transactions, the cost of posting data to Ethereum dropped significantly, which compressed L2 margins and made L3 settlement to an L2 only marginally cheaper than settling directly to L1.

This is why data availability is the most important cost variable. Coinbase's analysis estimates that DA accounts for over 95% of an L2's total costs when using Ethereum L1 for data. By using an alternative DA layer like Celestia or a Data Availability Committee, an L3 can reduce this cost by orders of magnitude, at the tradeoff of weaker security assumptions.

The cost hierarchy

The real savings for L3s do not come from an extra layer of compression. They come from the freedom to choose weaker (but cheaper) data availability while still anchoring to Ethereum's security through the L2 settlement layer.

The Fragmentation Problem

Every new L3 creates another isolated execution environment with its own liquidity pool, user base, and bridging requirements. This is the central tension of the appchain thesis: sovereignty and customization come at the cost of composability.

On a shared L2 like Arbitrum One, any contract can call any other contract atomically within a single transaction. The moment an application moves to its own L3, atomic composability with the rest of the L2 ecosystem breaks. Users must bridge assets between chains, introducing latency, cost, and bridge security risk. Capital that was concentrated on the L2 gets spread across dozens of smaller chains, reducing depth in every pool.

The tooling burden: Each L3 requires its own block explorer, indexer, bridge interface, wallet integration, and RPC endpoints. For developers, this multiplies the surface area for bugs and integration failures. For users, it means managing assets across an ever-growing number of networks. Shared sequencers and cross-chain messaging protocols like LayerZero standardize communication but do not restore the atomic composability that a shared execution environment provides.

Proposed mitigations include shared sequencers (Espresso, Astria), which coordinate transaction ordering across chains, and proof aggregation layers (AggLayer, Elastic Chain), which batch settlement proofs to reduce L1 costs. But these are coordination mechanisms, not liquidity solutions. A DEX pool on Chain A still cannot atomically access liquidity on Chain B, regardless of how efficiently their proofs are batched.

L3s vs. Cosmos Appchains and Polkadot Parachains

The appchain concept is not new to Ethereum. Cosmos and Polkadot pioneered application-specific chains years before L3s emerged.

Cosmos zones are sovereign chains that connect to the Cosmos Hub via IBC (Inter-Blockchain Communication). Each zone runs its own consensus, virtual machine, and governance. The tradeoff: full sovereignty means each zone must bootstrap its own validator set and security budget. Cosmos deliberately minimizes the importance of the hub, prioritizing chain sovereignty over shared security.

Polkadot takes the opposite approach. Parachains connect to the Relay Chain and inherit its security directly. Each parachain can customize its consensus, VM, and governance, but it does not need to recruit its own validators. The tradeoff: parachain slots are limited and historically required winning an auction to secure.

Ethereum L3s sit between these models. Like Polkadot parachains, they inherit security from a parent chain (the L2, which inherits from L1). But like Cosmos zones, they can launch permissionlessly without auctions or slot limits. The cost of this hybrid position is that L3s are further removed from the base layer's security: an L3's settlement guarantee depends on both the L2 and L1 functioning correctly, adding a layer of trust assumptions.

The Rollup-as-a-Service Layer

Deploying an L3 no longer requires deep infrastructure expertise. Rollup-as-a-Service (RaaS) providers have commoditized the process, offering no-code or low-code deployment flows for Arbitrum Orbit, OP Stack, and ZK Stack chains.

Caldera, Conduit, and Gelato are the leading providers. Each offers similar capabilities: choose a framework, pick a DA layer, configure gas tokens, and deploy a testnet or mainnet chain. The competition has driven deployment timelines down to weeks and costs down to monthly subscription fees, making L3 launches accessible to teams that would never have considered running their own chain infrastructure.

This ease of deployment is a double-edged sword. Low barriers to entry accelerate experimentation, but they also contribute to chain proliferation and the liquidity fragmentation problem described above. The RaaS providers compete primarily on price, deployment speed, and managed services, which means the decision to launch an L3 is increasingly driven by how cheap it is rather than whether the application truly needs its own chain.

The L1-L2-L3 Hierarchy

The relationship between the three layers follows a clear pattern of delegated security and specialization.

At the base, Ethereum L1 provides consensus, finality, and the ultimate data availability guarantee. It is the most secure and most expensive layer. L2 rollups batch transactions and post compressed data (or validity proofs) to L1, inheriting its security while offering cheaper execution. L3 chains repeat this pattern one level up: they batch their own transactions and settle to an L2, inheriting the L2's security (which itself inherits from L1). Each additional layer trades some security guarantee for greater customization and lower cost.

This creates a trust chain: an L3 user trusts that the L3 sequencer is honest (or that fraud/validity proofs work), that the L2 is functioning correctly, and that Ethereum L1 provides final settlement. If any link breaks, the guarantees weaken. This is why Vitalik Buterin has cautioned against viewing L3s as purely additive: each layer introduces its own failure modes and trust assumptions.

What L3s Mean for the Broader Scaling Debate

The L3 proliferation on Ethereum highlights a fundamental tension in blockchain scaling: how much complexity is acceptable in pursuit of performance? The L2 rollup fee war has already compressed margins and pushed rollups to compete on throughput, developer tooling, and ecosystem incentives. L3s add another competitive dimension, but they also add another layer of abstraction that users, wallets, and applications must navigate.

As of late 2026, L2BEAT tracks over 70 active Ethereum rollups with combined TVL above $45 billion. The number of L3 chains is growing, led by gaming and social applications, but most struggle to sustain meaningful transaction volume once initial hype fades. The chains that persist tend to be those with genuine throughput requirements that cannot be met on shared L2 infrastructure.

A Different Approach: Scaling Without Nested Chains

Not all ecosystems have adopted the layered-chain model. Bitcoin's scaling approach has taken a fundamentally different path, favoring Layer 2 protocols that extend the base layer without creating nested chain hierarchies.

Spark, for example, enables instant Bitcoin transfers using a statechain-based architecture with FROST threshold signatures. Users get self-custodial, near-instant transfers without deploying a separate chain, running a sequencer, or managing bridge infrastructure. There is no L3 on top of an L2 on top of an L1: the protocol connects directly to Bitcoin's base layer for settlement while handling transfers off-chain.

This reflects a philosophical difference. Ethereum's modular scaling thesis says applications should be able to customize their execution environment at every level. Bitcoin's approach says the base layer should remain simple, and scaling protocols should minimize the number of trust assumptions and intermediate layers between the user and final settlement.

For developers exploring Bitcoin-native scaling, the Spark SDK documentation provides integration guides for building wallets and payment applications on top of the protocol. And for a broader look at how these scaling approaches compare, see our research on the Bitcoin Layer 2 landscape.

What Comes Next

The L3 landscape is still in its early stages. Several developments will shape whether appchains become a permanent part of Ethereum's architecture or a transitional phase.

• Shared sequencers (Espresso, Astria) could reduce fragmentation by enabling atomic cross-chain transactions, but the technology is still maturing and adoption remains limited
• Proof aggregation (AggLayer, Elastic Chain) could lower settlement costs by batching proofs from many chains into a single L1 submission, making L3 economics more sustainable
• Chain abstraction tools that hide the underlying chain from users could resolve the UX problem, letting wallets route transactions to the cheapest available execution environment without the user knowing which L3 they are on
• The ongoing drop in blob fees after proto-danksharding may eliminate the cost advantage of L3s over L2s, making the case for a third layer purely about customization rather than economics

Whether Ethereum's scaling future looks like dozens of interconnected appchains or a handful of dominant L2s with most applications deployed directly remains an open question. The answer likely depends less on technical capability and more on whether the fragmentation problem can be solved without sacrificing the sovereignty that makes appchains valuable in the first place.

This article is for educational purposes only. It does not constitute financial or investment advice. Bitcoin and Layer 2 protocols involve technical and financial risk. Always do your own research and understand the tradeoffs before using any protocol.