Crypto bridges move value between chains, but the word "bridge" hides at least nine distinct mechanisms with very different trust assumptions. A burn-mint bridge that destroys USDC on Ethereum and reissues it on Base is not the same thing as a liquidity-pool bridge that swaps your USDC for a pool's USDC on the other side. This glossary defines every term you need to read a bridge whitepaper, compare two routers, or debug a stuck transfer in 2026.
​

Lock-mint is the original bridge design. The source chain locks your native asset in a smart contract, and the destination chain mints a wrapped IOU representing the locked deposit. WBTC, the largest wrapped asset by market cap, is the canonical example: BTC sits in BitGo custody, WBTC circulates on Ethereum.
​

The wrapped token is only as solvent as the lock contract is secure. If the lock is drained, the wrapped supply on the other side becomes unbacked. Most billion-dollar bridge hacks, including Ronin and Wormhole, targeted lock-mint vaults.
​

Burn-mint replaces the wrapped IOU with native issuance. The source chain burns your token, an attestation is signed, and the destination chain mints a fresh native token. There is no locked vault to drain because supply migrates rather than duplicates.
​

Circle's CCTP (Cross-Chain Transfer Protocol) is the reference implementation. Per Circle's CCTP documentation, V2 supports fast transfers in under 20 seconds on supported chains, with Circle's attestation service signing burn events on the source before the destination mint. Because USDC stays native on every chain, burn-mint avoids the wrapped-USDC fragmentation that plagued earlier bridges.
​

Liquidity-pool bridges (sometimes called "pool-based" or "AMM bridges") never mint anything. They hold inventory of the same asset on both chains. You deposit on the source, the bridge pays out from its destination-side pool, and rebalancers later equalize the books.
​

Stargate and Across both use this model for USDC and ETH. The user-visible advantage is speed: settlement is bounded by how fast the relayer can fill, not by canonical messaging finality. The cost is pool depth. Large transfers hit slippage curves or get rejected outright when the destination pool runs dry.
​

Optimistic bridges assume relayer claims are honest unless challenged inside a dispute window. The relayer fronts the user's funds on the destination immediately, then submits a claim on the source. A challenger has a window (typically 30 minutes to 2 hours) to dispute with fraud proof.
​

Across uses this model. Per Across documentation, relayers are bonded and a fraud-proof bond is slashed if a false claim posts. The end-user experience feels instant because the relayer eats the latency. The security assumption is that at least one honest watcher monitors the dispute window.
​

ZK bridges replace trusted attestations with cryptographic proofs. The source chain generates a zero-knowledge proof that a deposit or burn happened, the destination chain verifies the proof onchain, and no committee signs anything.
​

Polyhedra, zkBridge, and Succinct ship this design. ZK proofs are the strongest known security model because they reduce trust to the underlying cryptography. The tradeoff in 2026 is still cost: proof generation is compute-heavy, and verification gas on the destination chain remains 3 to 10x more expensive than a multisig attestation.
​

A DVN (Decentralized Verifier Network) is LayerZero's primitive for configurable security. Per LayerZero DVN documentation, every cross-chain message can require N-of-M attestations from a developer-chosen set of DVNs. A high-value transfer might require Google Cloud, Polyhedra, and LayerZero Labs to all sign. A low-value message might require only one.
​

The DVN model decouples bridge security from any single validator set. Apps pick their own risk profile rather than inheriting whatever the bridge ships by default.
​

A token bridge moves a specific asset. A message-passing layer moves arbitrary data and lets apps build whatever they want on top, including token bridges. LayerZero, Hyperlane, Wormhole, and Axelar are all message-passing layers. CCTP, Stargate, and Across are bridges built on top of (or alongside) message-passing rails.
​

The distinction matters for developers. If you only need to move USDC, a token bridge is fine. If you want cross-chain governance votes, NFT teleportation, or composable lending, you need a message-passing layer.
​

Intent-based bridges flip the model: instead of the user specifying the route, they declare an outcome ("I want 1000 USDC on Base, paying with USDC on Ethereum"). A network of solvers competes to fulfill the intent, and the user signs the winning quote.
​

Across, UniswapX, and CoW Swap all use intent architecture for cross-chain. The user gets the best execution without picking a route, and solvers absorb routing complexity. The model is now the dominant UX pattern for stablecoin bridging in 2026.
​

An atomic swap exchanges tokens across chains without any bridge in the middle, using hash-timelock contracts (HTLCs). Either both sides complete or neither does. Atomic swaps were the original trustless cross-chain primitive, used heavily for early BTC/LTC trading.
​

They remain niche in 2026 because they require both counterparties to be online and pre-fund the swap. Modern intent bridges deliver the same atomicity guarantee with better UX, so HTLC volume has collapsed to under 1% of cross-chain flow.
​

A native asset is issued canonically on its chain. USDC on Ethereum issued by Circle is native. A synthetic asset is a wrapped IOU: USDC.e on Avalanche is synthetic USDC that was bridged via lock-mint before Circle launched native USDC on Avalanche.
​

Synthetic assets fragment liquidity and create depeg risk if the bridge fails. The 2026 industry direction, led by Circle's CCTP rollout, is to make every major stablecoin native on every major chain and let burn-mint handle migration.
​

The right mechanism depends on the asset, the route, and your tolerance for trust assumptions. The table below summarizes the tradeoffs.

For stablecoins, burn-mint plus intent solving has become the default stack. CCTP V2 handles the canonical migration and intent solvers handle execution. For ETH and long-tail assets, optimistic and DVN-secured liquidity pools dominate. Lock-mint survives only for assets whose issuer cannot or will not ship a burn-mint version. ZK is gaining share where security premium outweighs gas cost, especially for high-value institutional flows.
​

Mechanism definitions cross-referenced against Circle CCTP documentation (developers.circle.com/stablecoins/cctp-getting-started), LayerZero DVN documentation (docs.layerzero.network/v2/concepts/modular-security/security-stack-dvns), Across documentation (docs.across.to), and Stargate documentation. Volume context from DeFiLlama bridge dashboard. No depeg or hack figures cited without primary sources.
​