Cross-chain bridges: the real cost (and how to pick one that won't lose your funds)
Bridges have been the single largest source of DeFi exploits. Here is the security model, the cost structure, and how to choose.
25/04/2026 · 10 min de leitura · Bridges · Cross-chain · Security
What bridges do
A cross-chain bridge moves value from one blockchain to another. ETH on Ethereum becomes ETH on Arbitrum, or USDC on Solana becomes USDC on Polygon. The bridge is the mechanism that makes this transfer.
Bridges are necessary because chains can't talk to each other directly. There is no global consensus across Ethereum, Solana, Bitcoin, and others. A bridge is a trust system that locks tokens on chain A and mints equivalent tokens on chain B.
Total value bridged in any given month exceeds $10B. The infrastructure is critical and used by every cross-chain trader.
Why bridges keep getting hacked
Bridges hold massive amounts of value (the assets locked on chain A) and have complex smart contract logic to verify cross-chain messages. Both are attack vectors.
Ronin: $625M in 2022. Wormhole: $325M in 2022. Nomad: $190M in 2022. Multichain: $130M in 2023. Orbit: $80M in 2024. The pattern: smart contract bug or compromised validator key.
Aggregate losses from bridge hacks since 2020 exceed $3B. This is the highest-loss category in DeFi history.
Bridge security models
Native bridges (canonical bridges). The bridge built by the chain itself, e.g., Arbitrum Bridge, Optimism Bridge, Base Bridge. These rely on the chain's own security and have the lowest trust requirements. Slowest (7 days for withdrawal from Optimistic Rollups).
Multi-sig bridges. A group of validators sign cross-chain messages. Trust assumption: the multi-sig isn't compromised. Faster than native bridges but historically the most exploited.
Light client bridges (IBC, CCIP). Each chain runs a light client of the other. Highest security but most complex. Currently limited to specific chain pairs (Cosmos IBC).
MPC bridges. Threshold cryptography distributes signing across nodes. Better than multi-sig but still has trust assumptions on the participants.
How to pick a bridge
For large amounts (>$10k), use canonical bridges. The 7-day delay on optimistic rollup withdrawals is annoying but the security model is the cleanest.
For smaller amounts and faster transfers, use established aggregators: LI.FI, Squid, Across, Stargate. They route through multiple underlying bridges and pick the most efficient.
Avoid unknown bridges. "This bridge is faster and cheaper" is often code for "this bridge hasn't been audited and rugged yet." Stick to bridges with at least 12 months of operation and meaningful TVL.
Cost structure
Gas fees on the source and destination chains. For Ethereum mainnet origin, this is the dominant cost ($5-30 typical).
Bridge fees. Usually 0.05% to 0.30% of the amount bridged, sometimes flat fees. Across has the lowest fees among major aggregators in 2026.
Liquidity provider fees on aggregators. Bridges that use intent-based execution charge a LP fee for the relayer. This is built into the quote you see.
Slippage. Bridges that route through DEXs along the way have slippage. Always check the displayed slippage parameter.
Time considerations
Canonical withdrawals: 7 days from Optimistic Rollups (Arbitrum, Optimism, Base), few minutes from ZK Rollups (ZKsync, Scroll).
Multi-sig and MPC bridges: usually 30 seconds to 5 minutes. Fast enough for trading.
Cross-chain swaps: 1-10 minutes depending on the route. The bridge time plus the destination chain confirmation time.
What to do when a bridge fails
If your transaction is stuck, check the explorer on both sides. The bridge UI might be lagging but the chain status is canonical.
If funds are sent but not received after the typical wait, file a support ticket. Most reputable bridges have refund processes for stuck transactions.
If the bridge is hacked, you're usually out of luck. Major bridges sometimes reimburse users from treasury (Wormhole was made whole by Jump Trading) but it's the exception.
Bottom line
Bridges are necessary infrastructure but historically the most dangerous part of DeFi. Use canonical bridges when you can afford the wait, established aggregators when you need speed, and never use a no-name bridge to move material funds.
If you're trading on OFFCODE and want to move funds to or from another chain, the convert product (powered by LI.FI in the background) handles bridging with audited routes. For larger amounts, the canonical bridge is still cleaner.