Blockchain Layers Explained: Layer 0, Layer 1, Layer 2, Layer 3 — The Complete Guide

The full blockchain layer stack

Before diving into each layer individually, it helps to see how they connect. The diagram below shows the conceptual architecture: Layer 0 protocols provide the communication backbone that connects independent Layer 1 chains. Layer 2 rollups sit on top of L1s and handle execution. Layer 3 application-specific chains sit on top of L2s. Data availability layers run alongside this stack, providing the storage and publication infrastructure that rollups depend on to verify state.

This is a conceptual map, not a wiring diagram. Real-world deployments mix and match layers depending on the application. A DeFi protocol might run on Arbitrum (L2), which settles on Ethereum (L1), uses EIP-4844 blobs for data availability, and is accessible from Cosmos chains via LayerZero (L0). A gaming application might run on an Arbitrum Orbit chain (L3), which settles on Arbitrum One (L2), which settles on Ethereum (L1). Every additional layer adds throughput and reduces cost — but also adds trust assumptions that need to be understood.

Layer 0: the interoperability layer

Layer 0 is the infrastructure that connects independent blockchains and enables them to communicate. Without L0, each blockchain is an island — it can run its own applications and settle its own transactions, but it cannot natively talk to another chain. Layer 0 protocols solve this with cross-chain messaging, shared security models, and interoperability frameworks.

There are two distinct architectural approaches at Layer 0. The first is the shared-security relay chain model, where a central chain provides consensus and security to connected application-specific chains. The second is the messaging protocol model, where independent chains maintain their own consensus but use a standardized protocol to pass messages and verify state across chain boundaries.

Relay chain model: Polkadot and Cosmos

Polkadot (DOT, approximately $2.0B market cap) uses a relay chain architecture. The relay chain handles consensus, security, and cross-chain message passing. Connected chains, called parachains, lease slots on the relay chain and inherit its security. As of early 2026, Polkadot supported 65 parachains using Nominated Proof of Stake (NPoS) consensus.

Cosmos (ATOM, approximately $1.07B market cap) takes a fundamentally different approach. Cosmos chains are sovereign — each runs its own consensus via the Tendermint (now CometBFT) engine. The Inter-Blockchain Communication (IBC) protocol handles cross-chain communication. Over 100 IBC-connected chains operated across the Cosmos ecosystem as of early 2026, each with independent security but standardized interoperability.

The philosophical difference matters. Polkadot provides shared security from the relay chain down — a parachain inherits the relay chain’s validator set. Cosmos provides sovereignty first and interoperability second — each chain manages its own validator set and chooses to connect via IBC. Both are valid engineering responses to the problem of cross-chain coordination.

Messaging protocols: LayerZero and Chainlink CCIP

LayerZero (ZRO, approximately $465M market cap) is an omnichain messaging protocol that connects over 70 networks. Rather than running its own chain, LayerZero provides a standardized communication layer that existing blockchains can plug into. It has secured over $33.6B in cross-chain value through its relayer and oracle verification system.

Chainlink CCIP (Cross-Chain Interoperability Protocol) extends Chainlink’s oracle network into cross-chain messaging. CCIP connects over 60 chains and, uniquely among L0 protocols, bridges into traditional finance through SWIFT integration with approximately 11,000 banks. LINK, the Chainlink token, carries a market cap above $9B. CCIP’s key differentiator is that it leverages Chainlink’s existing decentralized oracle network for cross-chain verification, adding institutional-grade reliability.

Avalanche (AVAX, approximately $4.3B market cap) sits at the intersection of L0 and L1. Its subnet architecture allows anyone to launch an independent blockchain that shares the Avalanche consensus engine. Over 75 subnets operated as of early 2026. Each subnet can define its own validator set and virtual machine, making Avalanche function as both a base settlement layer (its C-Chain for EVM-compatible DeFi) and an interoperability platform (subnet-to-subnet communication).

The blockchain trilemma

The blockchain trilemma, articulated by Vitalik Buterin, is the foundational engineering constraint behind the entire multi-layer stack. It states that a single blockchain can optimize for at most two of three properties simultaneously: decentralization, security, and scalability.

Bitcoin optimizes for decentralization and security. With over 19,000 reachable nodes running the same consensus rules and Proof of Work providing thermodynamic finality, Bitcoin is the most censorship-resistant and hardened network in operation. The tradeoff is throughput: 5 to 7 transactions per second.

Solana optimizes for scalability and security. Its combination of Proof of History (PoH) timestamping with Tower BFT consensus achieves 600 to 700 real-world TPS, with theoretical capacity above 65,000 TPS. The tradeoff is that validator hardware requirements are substantial — running a Solana validator requires high-end server-grade hardware, which limits the number of independent validators and the degree of decentralization compared to Ethereum or Bitcoin.

Ethereum optimizes for decentralization and security at L1 while explicitly delegating scalability to Layer 2 rollups. Ethereum’s base chain processes 15 to 30 TPS, but its L2 ecosystem (Arbitrum, Base, Optimism, zkSync) collectively processes thousands of TPS while inheriting Ethereum’s security. This is a deliberate architectural choice — the multi-layer stack is the scalability solution.

Layer 1: the base settlement layer

Layer 1 is the base blockchain — the chain that runs its own consensus mechanism, executes transactions, and finalizes state independently. Every L2, L3, and cross-chain bridge ultimately depends on the security and finality of its parent L1. The L1 is where final settlement happens and where the economic security budget (staked value or mining hashrate) sits.

Consensus mechanisms explained

Consensus is how a distributed network of validators agrees on the correct state of the blockchain. Different mechanisms make different tradeoffs between energy cost, speed, decentralization, and security guarantees.

Major Layer 1 blockchains

The L1 landscape in 2026 spans battle-tested incumbents and newer high-throughput chains. The table below captures key metrics for the major L1s.

L1 throughput comparison

The TPS numbers above represent observed real-world throughput, not theoretical maximums. Most newer L1s advertise theoretical TPS in the hundreds of thousands — those numbers represent capacity under ideal conditions with full validator participation and optimized transaction types. Real-world usage rarely approaches those ceilings. The gap between theoretical and observed TPS is itself a useful data point: it tells you how much headroom a chain has before congestion becomes a constraint.

Layer 2: the execution layer

Layer 2 is where the blockchain scalability problem gets solved in practice. L2s process transactions off-chain — in their own execution environment — and periodically post compressed proofs or state data back to the parent L1 for final settlement. The result is dramatically higher throughput and lower fees while inheriting the security of the base chain.

The L2 ecosystem on Ethereum has grown into a multi-billion-dollar infrastructure layer. As of early 2026, Ethereum L2s collectively held more TVL than most standalone L1 chains.

Types of Layer 2 solutions

Not all L2s work the same way. The five main categories differ in how they prove correctness, where they store data, and what tradeoffs they accept.

Optimistic Rollups

Optimistic Rollups assume all transactions are valid by default. They execute transactions off-chain, batch the results, and post the batch to L1. If any observer believes a batch contains an invalid transaction, they can submit a fraud proof during a challenge window (typically 7 days). If the fraud proof succeeds, the invalid batch is reverted and the sequencer is penalized.

The advantage is simplicity — Optimistic Rollups are easier to build and are EVM-equivalent by default, meaning existing Ethereum smart contracts can deploy without modification. The disadvantage is the 7-day challenge window, which introduces a withdrawal delay when bridging assets back to L1. Arbitrum and Optimism are the dominant Optimistic Rollup implementations.

ZK Rollups (Zero-Knowledge Rollups)

ZK Rollups generate a cryptographic validity proof for every batch. This proof mathematically demonstrates that all transactions in the batch were executed correctly, without revealing the underlying transaction data. The L1 smart contract verifies the proof (a fast computation) rather than re-executing every transaction.

The advantage is instant finality — once the proof is verified on L1, the batch is final with no challenge period. The disadvantage is that generating ZK proofs is computationally expensive, and achieving full EVM compatibility with ZK circuits has been technically challenging. zkSync Era, Starknet, Polygon zkEVM, Scroll, and Linea are the major ZK Rollup implementations.

Other L2 approaches

• Validiums — like ZK Rollups but store transaction data off-chain (often on a separate DA layer) rather than posting it to L1. Lower cost, but the security assumption shifts to the DA provider. StarkEx powers several Validium deployments.
• State Channels — two parties open a channel, transact off-chain unlimited times, and settle the final state on L1. Extremely efficient for repeated interactions between the same parties. The Lightning Network on Bitcoin is the best-known state channel implementation.
• Plasma — an earlier L2 design that creates child chains anchored to L1. Plasma was largely superseded by rollups because it struggled with general-purpose smart contract execution, but some implementations remain active.

How rollups work: 5-step process

EIP-4844: Proto-Danksharding

One of the most important Ethereum upgrades for the L2 ecosystem was EIP-4844, also known as Proto-Danksharding, which went live with the Dencun upgrade in March 2024. EIP-4844 introduced blob transactions — a new transaction type specifically designed for rollup data that is stored temporarily (approximately 18 days) rather than permanently on-chain.

The impact was dramatic. Before EIP-4844, rollups posted data as calldata to Ethereum L1, which was expensive because calldata is stored permanently by all Ethereum nodes. Blob transactions reduced L2 data costs by 10 to 100 times, dropping the average transaction fee on rollups like Base and Arbitrum from several cents to fractions of a cent. This single upgrade did more for L2 adoption than any other technical change in Ethereum’s recent history.

Major Layer 2 networks

L2 fees versus L1 fees

The entire point of L2s is cost reduction without sacrificing security. The table below illustrates the fee difference for a standard token transfer on L1 versus major L2 networks.

Layer 3: application-specific chains

Layer 3 takes the multi-layer concept one step further. Where L2s provide general-purpose execution on top of L1, L3s are application-specific chains built on top of L2s for specialized use cases. The logic is straightforward: if an application has unique requirements (custom gas tokens, ultra-low latency for gaming, specific privacy features, dedicated throughput for a single protocol), it can run its own chain on top of an existing L2 rather than competing for block space with every other application on that L2.

L3 examples

Xai (approximately $219M market cap) is a gaming-focused L3 built on Arbitrum Orbit. It provides a dedicated execution environment for blockchain games, with transaction fees subsidized by the network so players do not need to hold crypto to play. Xai settles on Arbitrum One (L2), which settles on Ethereum (L1) — a three-layer stack where each layer adds specialization.

Orbs (approximately $40M market cap) operates as a DeFi execution layer that sits on top of existing L1 and L2 networks. Rather than running its own settlement, Orbs provides backend computation for DeFi protocols, handling complex execution logic (automated strategies, limit orders, dollar-cost averaging) while the settlement remains on the parent chain.

dYdX is technically its own Cosmos-based appchain (dYdX Chain, launched in late 2023), which makes it architecturally a sovereign L1 within the Cosmos ecosystem rather than a true L3. However, its design — a single-purpose chain optimized entirely for perpetual trading with a custom order book — embodies the L3 philosophy of application-specific chain design.

L3 frameworks

L3 proliferation has been enabled by standardized rollup frameworks that allow teams to deploy custom chains without building the infrastructure from scratch.

The OP Stack deserves particular attention. With approximately 67% market share of all L2s by framework, it has become the de facto standard for launching new rollups. Base (Coinbase), Mantle, and over 30 other chains run on the OP Stack, all contributing sequencer revenue back to the Optimism Collective. The Superchain vision aims to make all OP Stack chains interoperable by default, turning what was a collection of independent L2s into a unified multi-chain network.

Sidechains vs. Layer 2s: the security inheritance question

One of the most common points of confusion in blockchain architecture is the difference between a sidechain and a true Layer 2. The distinction is fundamental and comes down to a single question: where does the chain’s security come from?

Polygon PoS is the most prominent example of a sidechain that is frequently mislabeled as an L2. Polygon PoS runs its own Proof of Stake consensus with its own validator set. It checkpoints to Ethereum periodically, but Ethereum does not verify Polygon PoS state transitions — it only stores the checkpoint hashes. If Polygon PoS validators colluded, Ethereum could not independently detect or prevent invalid state transitions. This is structurally different from Arbitrum, where any single honest observer can submit a fraud proof to Ethereum and trigger a rollback of an invalid batch.

Polygon has acknowledged this distinction and has been transitioning toward ZK-based security (Polygon zkEVM, Polygon CDK) that would give its ecosystem true L2-grade security inheritance. The label matters because it determines the trust model: on a true L2, you trust Ethereum’s validator set; on a sidechain, you trust the sidechain’s own validators.

Modular vs. monolithic: how blockchains are being rebuilt

Traditional blockchains are monolithic — a single chain handles all four core functions: execution (processing transactions), settlement (finalizing state), consensus (agreeing on transaction ordering), and data availability (storing and publishing transaction data). Bitcoin, Solana, and BNB Chain are monolithic architectures where one chain does everything.

The modular approach separates these functions across specialized layers. Ethereum post-EIP-4844 is the most prominent example of a blockchain moving toward modular design: Ethereum L1 handles settlement and consensus, L2 rollups handle execution, and blob transactions (or dedicated DA layers like Celestia) handle data availability.

The modular thesis argues that specialization produces better results than one chain trying to do everything. A dedicated DA layer can optimize entirely for throughput and storage cost. A dedicated execution layer (rollup) can optimize for fast computation and low latency. The settlement layer only needs to verify proofs and maintain the canonical state root, which is a much lighter workload than executing every transaction.

Neither approach is objectively superior. Monolithic chains like Solana demonstrate that a single well-optimized chain can achieve impressive throughput without modular separation. The tradeoff is that monolithic designs tend to require higher hardware requirements for validators, which concentrates the validator set. Modular designs add complexity (more layers to understand, more bridge trust assumptions) but allow each component to scale independently.

Data availability: the layer most people overlook

Data availability (DA) is the least visible but arguably most critical component of the modular blockchain stack. The question DA answers is deceptively simple: can anyone who wants to verify the blockchain’s state actually access the transaction data they need to do so?

For rollups, DA is not optional. When a rollup posts a batch to L1, validators and users need access to the underlying transaction data to verify that the new state root is correct. Without DA, you have to trust the rollup sequencer blindly — which defeats the purpose of inheriting L1 security.

DA solutions compared

Celestia (TIA, approximately $345–457M market cap) pioneered the dedicated DA layer concept and holds roughly 50% of the DA market. Its key innovation is Data Availability Sampling (DAS), which allows light nodes to verify data availability by downloading only small random chunks of a block rather than the entire thing. This makes verification lightweight and scalable — as the network grows, each individual node actually needs to download less data.

EigenDA (backed by EIGEN, approximately $1.3B market cap) takes a different approach by leveraging EigenLayer’s restaking mechanism. Ethereum validators who have staked ETH can opt into securing EigenDA as well, providing DA throughput of up to 100 MB/s backed by Ethereum’s own economic security. This makes EigenDA one of the highest-throughput DA solutions available without requiring a separate security model.

The DA market is still young and evolving rapidly. The key question for any rollup choosing a DA solution is the tradeoff between cost, throughput, and trust assumptions. Posting data to Ethereum L1 (via blobs) provides the strongest security guarantee but has limited throughput. Dedicated DA layers provide orders-of-magnitude more throughput and lower cost but introduce a separate security assumption that the DA layer’s own validators are honest.

Bridges: connecting the layers (and their risks)

Cross-chain bridges are the infrastructure that moves assets between different blockchains and layers. They are the practical implementation of Layer 0 interoperability — without bridges, each chain is an isolated island of liquidity. As of early 2026, cross-chain bridges collectively held approximately $21.94B in TVL.

Bridges are also the single largest attack vector in the multi-chain stack. Over $606M was stolen in bridge exploits in April 2026 alone. The reason bridges are disproportionately targeted is structural: they hold large pools of locked assets (the value backing wrapped tokens on the destination chain), and the verification of cross-chain state is inherently more complex than on-chain verification within a single network.

The bottom line

The blockchain layer stack — L0 for interoperability, L1 for base settlement and consensus, L2 for execution, L3 for application-specific chains, with data availability as a cross-cutting concern — exists because no single blockchain can maximize decentralization, security, and scalability simultaneously. Every layer in the stack is an engineering response to the blockchain trilemma.

The practical implication for anyone navigating this space: know which layers you are using and what trust assumptions each layer carries. When you transact on Arbitrum, you are trusting Ethereum’s validator set for final settlement, Arbitrum’s sequencer for transaction ordering (with fraud proofs as a backstop), and EIP-4844 blobs for data availability. When you bridge to a Cosmos chain, you are adding the bridge’s trust model and the destination chain’s independent validator set. Each hop adds throughput and reduces cost — and each hop adds a trust assumption.

None of the architectures described above are inherently superior to the others. Monolithic chains, modular stacks, shared-security relay chains, sovereign interconnected ecosystems, and application-specific L3s all represent different engineering tradeoffs optimized for different use cases. The multi-layer stack is not a hierarchy of quality — it is a toolkit. The right combination depends on the application, the security requirements, the throughput needs, and the acceptable trust model.

Related reading

Top 25 Blockchains Ranked

Comprehensive ranking by market cap, TVL, throughput, and ecosystem activity.

Ethereum Glamsterdam Fork Explained

What changed in Ethereum’s latest upgrade and how it affects L2 costs and validator economics.

Whale Activity on Ethereum Layer 2s

How whale wallets are moving across Arbitrum, Base, Optimism, and zkSync.

Cross-Chain Bridge Whale Activity

Where the largest bridge flows are heading and which protocols carry the most whale volume.

What Is a Crypto Whale?

Definition, thresholds, wallet types, and why whale behavior matters for on-chain analysis.

Top 200 Cryptocurrencies

Complete guide covering the top 200 crypto assets by market cap with sector breakdowns.

Live whale feed → Token whale tracker → Whale wallet leaderboard → Research hub →