All articles
IntermediateInfrastructure

Layer 2s and Bridges: Scaling Dreams, Security Nightmares

L2s make Ethereum cheaper. Bridges make chains talk. They also account for billions in hacks. Here's what you need to know.

May 21, 2025
5 min read

Dive Deeper with AI

Click → prompt copied → paste in AI chat

Ethereum gas fees hit $100+ for a simple swap in 2021.

The solution? Layer 2s. Cheaper, faster, built on Ethereum security.

The problem? To move between chains, you need bridges.

Bridges have lost over $2.5 billion to hacks.

This is the scaling trilemma in practice.


What are Layer 2s?

L2s are chains that run on top of Ethereum.

You do transactions on the L2 (cheap, fast). The L2 posts proofs to Ethereum (expensive, secure).

It's like doing 1000 transactions, then writing one summary to the main chain.

Types:

Optimistic rollups (Arbitrum, Optimism) - Assume transactions are valid. Fraudulent ones can be challenged.

ZK rollups (zkSync, StarkNet) - Use math proofs to verify transactions. No trust needed.

Validiums - Like ZK but data stored off-chain. Even cheaper, different security.

All of them: "Trust Ethereum, but faster."


The bridge problem

You have ETH on Ethereum. You want to use it on Arbitrum.

You can't just... send it. Different chains don't talk to each other natively.

Enter bridges.

You send ETH to a bridge contract on Ethereum. The bridge mints "wrapped ETH" on Arbitrum. Now you can use it.

To go back, you burn the wrapped ETH on Arbitrum. The bridge releases your real ETH on Ethereum.

Simple concept. Massive attack surface.


Why bridges get hacked

Bridges are essentially giant pools of money connecting two systems.

Attack vectors:

Smart contract bugs. Bridge code handles billions. One bug = drained.

Validator compromise. Many bridges use validators to verify cross-chain messages. Compromise enough validators = fake messages = drained.

Private key theft. Some bridges use multisigs. Get enough keys = drained.

Logic errors. Different chains have different rules. Edge cases can be exploited.

The fundamental problem: bridges are trusted intermediaries in a trustless ecosystem.


The hall of shame

Ronin Bridge - $625M. Axie Infinity's bridge. 5 of 9 validators compromised. North Korea suspected.

Wormhole - $320M. Bug in the Solana contract allowed minting unbacked tokens.

Nomad - $190M. Copy-paste hack. Once the exploit was public, hundreds of people drained the bridge.

Harmony Bridge - $100M. 2 of 5 multisig keys compromised.

These aren't edge cases. They're the normal state of bridge security.


Canonical vs external bridges

Canonical bridges are official. Arbitrum's bridge, Optimism's bridge.

They're usually safer because:

  • Run by the L2 team
  • Use L1 security for finality
  • Have withdrawal delays (7 days for optimistic rollups)

That 7-day delay? It exists so fraud can be proven. Annoying but secure.

External bridges are third-party. They promise faster withdrawals, multi-chain support.

They're riskier because:

  • Different security models
  • Often use validator sets
  • Smaller teams, less audited

When people say "I'll just use a faster bridge" - they're trading security for convenience.


L2 security assumptions

Not all L2s are equal.

Fully inheriting Ethereum security:

  • Ethereum validates everything
  • If Ethereum is secure, L2 is secure

Training wheels:

  • Most L2s have admin keys that can override things
  • "Upgradable" contracts mean trust the team
  • Emergency stops, manual interventions possible

Check L2Beat.com. They rate L2s by actual security level.

Many popular L2s are still "Stage 0" - significant training wheels.

"It's an L2" doesn't mean "it's as secure as Ethereum."


The fragmentation problem

There are now dozens of L2s.

Arbitrum, Optimism, Base, zkSync, Starknet, Linea, Scroll, Zora, Mode, Blast, Manta, Mantle...

Each one:

  • Has its own liquidity
  • Has its own bridges
  • Has its own apps
  • Has its own tokens

Using DeFi across L2s means bridging constantly. Bridging means risk.

The "Ethereum ecosystem" is fragmenting into isolated islands connected by sketchy bridges.

Some call this composability. Others call it a mess.


How to minimize risk

Use canonical bridges when possible. Yes, the 7-day wait sucks. Security is worth it.

Minimize bridge size. Don't bridge your life savings. Bridge what you need.

Check L2Beat. Understand what security you're actually getting.

Diversify L2s. Don't keep everything on one L2.

Be patient with new bridges. Battle-tested > new and fast.

Watch for bridge audits. Are they audited? By whom? Multiple times?


The future

Several attempts to fix this:

Native rollups. L2s built into the Ethereum protocol itself. No external bridges needed.

Shared sequencers. Multiple L2s sharing infrastructure for better composability.

Intent-based bridges. Instead of lock-and-mint, market makers fulfill cross-chain intents.

Restaking security. EigenLayer-style shared security for bridges.

These might help. Or they might introduce new attack vectors.

Bridge security is an active area of research. Translation: it's not solved.


The honest take

L2s are necessary. Ethereum can't scale alone. Gas fees prove it.

But the current state:

  • L2 security often has training wheels
  • Bridges are single points of failure
  • The ecosystem is fragmented
  • Users don't understand the tradeoffs

Using L2s is probably fine for normal activity. Just understand:

  • Your security model changed
  • Your money can be stuck if things go wrong
  • Bridges are additional risk

The dream: seamless scaling with Ethereum security.

The reality: it's a work in progress, and billions have been lost along the way.

Use L2s. Just don't pretend they're the same as L1.


Next: Vaults and escrows - smart contract patterns for holding your money.

Liked this article? Follow me!

@t0tty3
#layer2#bridges#scaling#ethereum

Dive Deeper with AI

Click → prompt copied → paste in AI chat