What Is a Crypto Bridge? How Cross-Chain Transfers Actually Work

A crypto bridge is the piece of infrastructure that lets value leave the chain it was issued on. Every time you move USDC from Ethereum to Arbitrum, or swap a token that only exists on one network for one that only exists on another, something has to keep track of two separate ledgers and make sure both agree. That is the bridge's job, and it is a harder job than it sounds.

Bridges also have the worst security record in crypto. Understanding how they work, and where the trust actually sits, is the difference between using one sensibly and treating it as a neutral pipe.

Why blockchains cannot simply talk to each other

Each blockchain is a self-contained record. A node on Ethereum has no way to read Arbitrum's state, and a node on Arbitrum cannot verify what Ethereum believes. They are separate computers running separate rules, connected only by the internet.

That isolation is deliberate. It is what lets a chain enforce its own rules without asking permission. It also means value cannot move across chains the way a bank transfer moves between accounts. There is no shared database and no central clearing house that both chains recognise.

A bridge is the workaround: an outside system that observes one chain, takes custody of something there, and then triggers an equivalent action on the other chain. Trust moves from the protocol to whoever runs that outside system.

What a crypto bridge actually does

Most bridges fall into two patterns.

The first is lock-and-mint. You send tokens to a contract on the source chain, where they are locked. The bridge then confirms that deposit and mints a representation of the token on the destination chain. To go back, the representation is burned and the original is released from the lock. This is how most wrapped assets work.

The second is burn-and-mint, used when the issuer itself controls supply on both chains. Circle's Cross-Chain Transfer Protocol, for example, burns USDC on the source chain and mints native USDC on the destination, so no third party holds a claim on your coins. Some bridges instead run liquidity pools on both sides: you deposit into a pool on one chain and withdraw from a pool on the other, with market makers rebalancing in between.

A cross-chain bridge can use any of these designs, and many combine them. The important detail for a user is not the mechanism but the answer to one question: what exactly do you hold after the transfer?

A wrapped token is a claim, not the coin

When a lock-and-mint bridge gives you wrapped BTC or wrapped ETH, you are not holding bitcoin or ether. You are holding a token that says the bridge owes you bitcoin or ether. Its value depends entirely on the bridge's reserve still being there and still being redeemable.

That distinction feels academic until a bridge is drained. Then the wrapped token trades at a discount or stops trading at all, because the claim behind it no longer has backing. Holding a cross-chain representation is a counterparty position, even when the token looks like a familiar ticker in your wallet.

Native issuance avoids this. If an asset is issued directly on a chain by its issuer, there is no third-party claim involved. That is one reason stablecoin transfers through issuer-run channels are less fragile than the same transfer through a lock-and-mint bridge.

Canonical bridges versus third-party bridges

Layer 2 networks, which settle back to Ethereum, ship with a bridge maintained by the network itself. These are usually called canonical bridges. Deposits are typically quick. Withdrawals take longer, because the rollup has to prove its state to Ethereum first: optimistic rollups wait out a challenge period of about seven days, while zero-knowledge rollups use validity proofs and usually finish in hours.

A third-party cross-chain bridge competes on speed and convenience. It can move assets between chains that have no direct relationship, support tokens that were never issued natively elsewhere, and skip the wait. The trade-off is that you are trusting their contracts, their validators and their operational security instead of the destination chain's own settlement guarantees.

Neither option is universally safer. A canonical bridge carries the risk of the underlying rollup; a fast third-party bridge carries the risk of a smaller team holding a large pool of assets.

Where the risk actually sits

A crypto bridge concentrates money and trust in one contract, and that is precisely what attackers look for. Bridge exploits are not one bug, though. They are several distinct failure modes that happen to end the same way.

Smart contract flaws are the most familiar. A bridge contract holds a pool of locked assets, which makes it a single high-value target, and a logic error in deposit accounting or signature checking can let an attacker withdraw funds they never deposited.

Key and validator compromise is the second. Many bridges verify transfers with a multisig or a small validator set. If enough signers are phished or their infrastructure is breached, the bridge will happily authorise a fraudulent withdrawal. The Ronin bridge, which lost roughly $624 million in March 2022, was compromised through validator keys rather than a contract bug.

Message verification is the third, and the least intuitive. A bridge has to accept proof that something happened on another chain. If that proof can be forged or replayed, the bridge can be tricked into releasing funds. The Wormhole exploit in February 2022, about $326 million, abused a signature verification flaw; Nomad lost around $190 million in August 2022 after a flawed initialisation let users submit copied transactions.

The pattern is not subtle. Chainalysis estimated that bridge exploits accounted for roughly 69% of all crypto stolen in 2022, a year that also included the roughly $611 million Poly Network incident, most of which was later returned. Bridges concentrate money and trust in one place, which is exactly what attackers look for.

Bridges and messaging protocols are not the same thing

A newer generation of interoperability products, including LayerZero and Chainlink's CCIP, is often described as bridging. Strictly, these are messaging layers: they pass data and instructions between chains, and asset transfers are one application built on top.

That architecture can be more flexible, since a single message format can carry a token transfer, a governance vote or a price update. It does not automatically make the movement of assets safer. The same questions apply: who verifies the message, how many parties must collude to forge one, and what happens if the verification layer is wrong?

Peg risk, liquidity risk and the quiet failures

Not every bridge failure is a hack. Some are quiet.

If a bridge's reserve is drained slowly, or if redemptions outpace the assets available, the wrapped token can drift below the value of what it represents. That is a peg break, and it can happen without any contract being exploited. A related problem is liquidity: a bridge may quote a rate that looks fine for a small transfer and punish a large one, because the destination pool is thin.

There is also the question of what happens when a bridge shuts down. Teams wind down, chains get abandoned, and a wrapped asset whose issuer has stopped operating has no obvious route back to the original coin. The token may still exist in your wallet. That does not mean it is redeemable.

How to move assets across chains without unnecessary risk

A few habits cover most of the practical risk.

Use the destination chain's canonical bridge when one exists and the wait is acceptable, especially for large amounts. If you need speed or a route the canonical bridge does not offer, prefer a bridge with a long operating history, published audits and a design where the trust assumptions are stated plainly rather than buried.

Send a small test transfer first, confirm it arrives and can be withdrawn, then move the rest. Check the contract address against the bridge's own documentation rather than a link from a search result or a chat message. Compare the quoted fee and the destination amount, not just the headline rate.

Finally, treat withdrawal time as part of the cost. A seven-day wait is not a bug in an optimistic rollup; it is the mechanism that lets the rollup be challenged. Plan around it instead of discovering it after the deposit.

What a bridge cannot do for you

Bridging does not make an asset safer, generate yield, or remove the risks of the chain you are moving to. It does not avoid tax reporting or identity checks, and it does not make a thin token liquid on the other side.

It also does not protect you from the most common loss of all: approving a malicious contract, or following instructions from someone impersonating support. No legitimate bridge asks for your seed phrase, and none needs it to complete a transfer.

The useful mental model is simple. A bridge is a custodian with a technical interface. Sometimes the custodian is the destination chain itself, and sometimes it is a company holding a very large pool of locked coins. Knowing which one you are dealing with, before you sign anything, is most of the work.