Key Highlights
-
LayerZero V2 routes messages across more than 40 blockchains through a single EndpointV2 contract deployed at 0x1a44076050125825900e736c501f859c50fE728c on every EVM chain
-
An OFT transfer typically needs 200,000 gas on the destination chain, and underquoting the executor gas limit is the most common cause of stuck packets
-
LayerZero Labs raised $135 million in March 2022 at a $1 billion valuation from a16z crypto, Sequoia, and FTX Ventures
-
The ZRO token launched on June 20, 2024 with a 1 billion supply, and protocol fees can be paid in ZRO or in the source chain native token
-
Default DVN sets on most routes pair the LayerZero Labs verifier with a second operator such as Google Cloud, and integrators can raise that threshold through setConfig
Learning how to bridge assets with LayerZero starts with a correction: you do not call a bridge contract. You send a packet through one endpoint contract, pay a fee quote that funds destination gas, and rely on independent verifiers to attest delivery. The transfer itself is the easy part. The quote, the gas limit, and the revert path are where integrations break.
Most guides stop at “approve, then send.” That advice gets users across a testnet and nowhere else. This walkthrough covers the SDK setup, the fee math, and the fallback wiring that production teams actually ship.
What LayerZero Actually Does When You Bridge Assets
LayerZero is a messaging layer, not a liquidity pool. There is no wrapped asset parked in a vault and no bonded liquidity provider on the far side. Value moves because the token contract on the source chain burns or locks, and the token contract on the destination chain mints or unlocks after a verified packet arrives.
Every EVM chain runs the same EndpointV2 contract at 0x1a44076050125825900e736c501f859c50fE728c. A packet carries a nonce, the source endpoint ID, the sender address, the receiver address, the payload, and an options field that tells the executor how much gas to spend on delivery.
Three roles do the work. The DVN verifies that the source transaction happened. The executor delivers the payload and pays destination gas from the fee you supplied. The MessageLib, ULN302 by default, stores the configuration that defines which DVNs must attest before a packet is considered valid.
That separation is the whole design. You are not trusting a multisig to hold your tokens. You are trusting a named set of verifiers, and you can inspect and change that set.
How to Bridge Assets with LayerZero: SDK Setup
Install the packages under the @layerzerolabs scope: the definitions package for endpoint IDs, the v2 utilities package for options encoding, and the OFT EVM package if you are shipping your own token. The LayerZero CLI scaffold generates a working OApp with the config files already wired.
Use endpoint IDs, never chain IDs. Ethereum mainnet is 30101, Arbitrum is 30110, Optimism is 30111, Polygon is 30109, Base is 30184, and BNB Chain is 30102. Passing a chain ID where an endpoint ID belongs is the second most common integration bug after gas underquoting.
If your token already exists on one chain, deploy an OFTAdapter there and a plain OFT on every new chain. Deploying a standard OFT on the origin chain mints supply that was never locked, and that mistake is unrecoverable once liquidity exists.
Register the peer addresses in both directions. A route that is configured on the source but not the destination will quote successfully and fail on delivery.
Quoting and Paying Gas Without Overpaying
Call quote() before every send. It returns a MessagingFee struct with two fields, nativeFee and lzTokenFee. Set msg.value to nativeFee plus any native drop you requested, and pass a refundAddress that is a real wallet rather than the zero address.
The options bytes are where integrations quietly lose money. Options.newOptions().addExecutorLzReceiveOption(200000, 0) caps destination execution at 200,000 gas. A straightforward OFT mint with a fee sweep fits inside that. A compose call that triggers a swap on the far side does not, and it will revert after you have already paid.
Overquoting is cheap because the surplus returns to the refund address. Underquoting is not, because the fee is spent whether or not the payload executes.
If you pay fees in ZRO rather than native gas, check that the destination chain accepts the ZRO payment path before you ship. Not every route supports it.
Configuring Fallbacks So Messages Do Not Get Stuck
When lzReceive reverts on the destination, the executor does not drop the packet. EndpointV2 stores the payload and emits a MessageFailed event. Anyone can retry delivery by calling lzReceive again with the same GUID and payload, or clear the stored message if the receiving app exposes a recovery function.
Build that recovery path into your own contract. Emit a custom event inside _lzReceive, persist the GUID, and expose a permissionless retry function that calls endpoint.lzReceive with the stored arguments. Without it, a single revert strands user funds until you deploy a patch.
Ordered delivery is the trap most teams miss. EndpointV2 enforces nonce order per channel by default, so one stuck packet blocks every message behind it. Setting unordered execution on the receiver removes that head of line blocking and lets later transfers land while you debug the failed one.
For OFTs, add a pause switch and a credit back path on the destination. A mint that reverts because of a supply cap should not require a governance vote to unwind.
Hardening the Route: DVNs and Security Defaults
The default configuration on most routes uses two DVNs. You can require more. Calling setConfig with a ULN config that names three or four independent operators raises the per message cost by a few dollars and removes single operator trust from the equation.
Set block confirmations deliberately. Defaults are conservative on Ethereum and thin on cheap L2s, where reorg depth is smaller but not zero. If you are moving size, pay for the confirmations.
Two DVNs run by two companies is better than one. It is not the same as two independent implementations, and you should say so plainly in your docs rather than implying a stronger guarantee than you have.
Review the config after every chain upgrade. A DVN that gets removed from a route without your knowledge changes your security model silently.
Frequently Asked Questions
How does bridging with LayerZero actually work?
You do not call a separate bridge contract. You send a packet through the single EndpointV2 contract, pay a fee quote that funds destination gas, and rely on independent verifiers to attest delivery.
Why do LayerZero transfers get stuck?
The most common cause is underquoting the executor gas limit. An OFT transfer usually needs about 200,000 gas on the destination chain, so a quote that is too low can leave the packet stuck.
What is the LayerZero endpoint contract address?
LayerZero V2 uses one EndpointV2 contract at 0x1a44076050125825900e736c501f859c50fE728c on every EVM chain. That single endpoint is what routes messages across more than 40 blockchains.
Can I pay LayerZero protocol fees in ZRO?
Yes, protocol fees can be paid in ZRO or in the source chain native token. The ZRO token launched on June 20, 2024 with a 1 billion supply.
The TCB View
TCB is bullish on LayerZero as infrastructure and cautious on the way most teams integrate it. The problem is that configurability only helps operators who actually configure it, and the majority of integrations we have reviewed ship with default DVNs, default confirmations, and no retry function at all.
We see the winners as teams that treat the quote and the revert path as product surface: they show users a real fee, they store the GUID, and they expose a retry button. The losers are teams that copy a tutorial, hardcode 80,000 gas, and discover ordered nonce blocking during their first incident. Users pay for that lesson with stranded funds.
Watch the DVN count on the routes you depend on. If a major integrator moves from two verifiers to four, that is a signal the market is pricing security properly, and it will show up first in average fee per packet rather than in any announcement.

