Layer 2 networks are often seen as Ethereum’s answer to high fees and slow speeds. While decentralized, they’re supposed to be faster, cheaper, and just as secure. But are they really as independent as they seem? From big-name rollups like Optimism and Arbitrum to corporate-backed chains like Base, many of these solutions rely heavily on Ethereum and are still controlled by small teams. In this article by SwapSpace CEO Andrew Wind, we’ll dig into what Layer 2s promise, how they work, and why truly decentralized scaling might not be as real as it sounds.

Layer 2 (L2) solutions were introduced to solve Ethereum’s biggest pain points: scalability and affordability. As Ethereum mainnet transactions became expensive (median fees peaking at $11.77 and average fees surpassing $23.43 during the 2021 DeFi boom), L2s promised relief by executing transactions off-chain while settling data back on-chain.

Rollups, the most prominent L2 architecture, come in two main types: optimistic rollups (like Optimism and Arbitrum) and zk-rollups (like zkSync and Starknet). Both aim to reduce fees dramatically, often up to 10 to 100 times, while preserving Ethereum’s security guarantees. For instance, as of early 2025, users on Base can send a transaction for under $0.05, compared to $1–$5 on Ethereum L1.

L2s also unlock higher throughput. While Ethereum handles ~15 TPS (transactions per second), rollups can scale this to hundreds or even thousands of TPS. This scalability is crucial for mainstream use cases like gaming, social media, and payments.

More importantly, L2s have been marketed not just as technical upgrades, but also as ideological advancements: decentralized systems that don’t rely on trusted intermediaries. The dream is modular scaling — many independent L2s atop a credibly neutral base layer. In theory, these L2s would compete, interoperate, and evolve organically. However, that dream depends on how decentralized and self-sovereign these L2s are.

While Layer 2s are designed to offload transaction processing from Ethereum, they’re still deeply dependent on Layer 1 in several critical ways. This reliance shapes their performance, security, and even their long-term viability.

L2s don’t have their own consensus mechanisms, they rely on Ethereum to finalize transactions. Optimistic rollups like Optimism assume transactions are valid unless challenged. Fraud proofs are posted to Ethereum, and disputes are resolved there. ZK-rollups (e.g., zkSync, Starknet) generate cryptographic proofs off-chain and post them on-chain for verification. In both models, Ethereum acts as the final judge. If Ethereum is compromised, censored, or experiences downtime, L2s are directly affected.

Layer 2s must publish transaction data to Ethereum to preserve transparency and trust. This is expensive. For example, posting data to Ethereum currently costs $0.20 — $0.50 per KB, depending on gas prices. Moreover, a single rollup like Arbitrum spends millions annually just on call data fees to Ethereum. Thus, L2s are economically tied to Ethereum’s fee market. If L1 fees spike, the cost of using L2s increases too.

L2 assets exist as representations of tokens on Ethereum. Moving assets between layers requires bridges; for instance, the official Arbitrum bridge withdrawals back to L1 take 7 days due to fraud-proof challenge periods. Besides, most bridges are secured by multisigs or centralized actors, creating potential points of failure.

Since Ethereum provides critical infrastructure, Layer 2s often rely on centralized sequencers to avoid delays and complexity. This tight coupling makes L2s more like extensions of Ethereum than fully sovereign chains. Thus, while L2s offer scalability, they’re still bound to Ethereum’s rules, costs, and security, raising questions about how “independent” they truly are.

Despite being marketed as decentralized, most Layer 2s today rely heavily on centralized actors for critical operations, especially when it comes to sequencing transactions and governing the protocol itself. These central control points raise serious concerns about censorship, reliability, and user sovereignty.

A sequencer is the entity that orders transactions on a rollup. Currently, almost every major L2 uses a single, centralized sequencer controlled by the development team or foundation.

Example! Optimism, Arbitrum, zkSync, and Base all use centralized sequencers today. Thus, the team running the sequencer can censor transactions, reorder them, or go offline, creating a potential single point of failure.

While some L2s have launched governance tokens, real power remains centralized: ARB, the Arbitrum governance token, is held mostly by insiders. In its first DAO vote, the foundation pushed through a $1 billion treasury allocation, sparking backlash and raising questions about how democratic the process really is. OP, Optimism’s token, is also largely held by early backers, and its “Citizens’ House” governance layer is still in its early stages.

Most major upgrades and protocol changes are proposed and approved by the core teams or closely aligned DAOs.

Many rollups have raised tens to hundreds of millions from VCs. For instance, Matter Labs, the company behind zkSync, has raised a total of $458 million across multiple funding rounds. This funding often comes with board seats, influence, and privileged access to token allocations.

Together, centralized sequencing and insider-heavy governance mean that many L2s function less like decentralized networks and more like Web2 startups running on blockchain rails.

To understand how centralized or decentralized today’s Layer 2s are, let’s look at three of the biggest players: Optimism, Arbitrum, and Base.

Optimism runs on a single sequencer controlled by the Optimism Foundation. This centralized control allows for efficient block production, but also means users must trust the Foundation not to censor or manipulate transactions.

• While Optimism has a governance token (OP), real power still lies with the Foundation and its “Token House”.

• The planned “Fault Proof” system, which would allow for permissionless fraud proofs, has not yet been launched as of Q2 2025.

• Optimism is also leading the Superchain initiative, aiming to unify multiple chains (including Base) under shared infrastructure, raising further questions about the consolidation of power.

Arbitrum launched the ARB token and a DAO to govern the protocol. While this suggests decentralization, early governance was rocky.

• In 2023, the Arbitrum Foundation allocated $1B worth of ARB before community approval, a move that angered users.

• A few whales still dominate voting power, and the sequencer remains centralized under Offchain Labs.
  No concrete timeline exists for decentralizing the sequencer.

Base is built by Coinbase using Optimism’s tech stack. Unlike other rollups, it is entirely operated and governed by Coinbase.

• No governance token exists.

• All upgrades, fee settings, and censorship decisions are made by a centralized entity.

• While Base is technically part of the Ethereum ecosystem, its control structure is fully corporate.

Thus, despite differing in structure and branding, most current L2s are still far from fully decentralized or independent.

Important! The BASE token listed on sites like CoinMarketCap belongs to an unrelated project called Base Protocol, which aims to track the total crypto market cap. Coinbase’s Base Layer 2 network does not have a native token, and the team has publicly stated they have no plans to launch one.

For Layer 2s to live up to their decentralized ideals, they need to shift away from today’s centralized foundations. The good news? That shift is already starting. Projects like Espresso, Astria, and Radius are building shared, decentralized sequencers that any rollup can use, reducing censorship risk and reliance on a single operator. While still early, these systems aim to make L2s more neutral and resilient.

On the governance side, some rollups are experimenting with more transparent token allocations, quadratic voting, and minimal governance models that reduce human intervention. Arbitrum’s DAO, despite some missteps, is one of the more advanced efforts here.

There’s also a growing call for transparency standards, audits, and reports that show how decentralized an L2 really is. These would cover key areas like how the sequencer operates, how secure and trust-minimized the bridges are, and who holds decision-making power. Decentralization isn’t a one-time feature, it’s an ongoing process. And if users and developers demand it, L2s can move beyond speed and scale to real independence.

Layer 2s have brought real scalability to Ethereum, but at the cost of decentralization in many cases. Centralized sequencers, opaque governance, and economic dependencies on Layer 1 raise important questions about their independence.

The path forward isn’t just about faster transactions, it’s about building trustless systems that truly align with crypto’s core values. If decentralized scaling is the goal, it must be actively prioritized, not simply assumed to exist by default.