Notice: Editorial analysis, not financial advice. Data verified on August 23, 2026, against the LayerZero public metadata registry, DefiLlama, and notices published by the protocols themselves; the schedule originates from the LayerZero blog post dated July 24, 2026, as updated on August 21. Total Value Locked figures fluctuate daily. CleanSky does not receive commissions or referral payments from any of the cited protocols.

LayerZero withdrew its five-chain verifier at the end of July 2026 and has twenty-seven others on the schedule before October: thirty-two networks whose contracts remain exactly where they were. LayerZero — the infrastructure that transports messages between blockchains and upon which the Stargate bridge operates — ceased its verification and delivery services on Botanix on July 30 and on Moonriver, Moonbeam, Nexera, and Canto on July 31. Nothing was deleted: the endpoint — the contract that sends and receives LayerZero messages on each chain — remains deployed, the executor does too, and anyone holding USDC.e — the receipt a bridge issues on the destination chain while the real USDC remains locked on the source chain — still sees it in their wallet at market price. What is missing is a signature: that of the verifier certifying that the return message is legitimate. This article reconstructs the dependency chain of a bridged asset — who issues the receipt, where the collateral stays, who verifies, who delivers — and answers the question that decides everything else: when the service is turned off, is there any action the holder can execute on their own?

What exactly did LayerZero shut down and on what dates?

The notice was published on July 24, 2026, on the LayerZero blog under the title Support Update and was updated on August 21. The update matters because the list grew: the social media message on July 24 named five chains for "the next 30 days," while the current version includes thirty-two spread across four waves, with an express warning that chains and dates are subject to change. The tables are included in the post as images rather than text, a detail that explains why the schedule circulated in fragments: the report spread by the Chinese outlet PANews on July 25 mentioned twenty chains, those from the original version.

What is ceasing are two services that operate off-chain: the DVN (Decentralized Verifier Network, the network of verifiers that certify a message actually left the source chain) and the Executor (which delivers that already verified message to the destination). LayerZero justifies the withdrawal based on the minimal activity of these networks. The affected Stargate assets are USDC.e, wETH, and Hydra USDT, and the recommended destinations are Ethereum, Arbitrum, Base, and BNB Smart Chain (BSC).

The warning in the notice, in an abbreviated translation: "failure to act before support is withdrawn will result in losing access to your funds." This is the full schedule with the status of each network in the LayerZero public metadata registry:

Cut-off DateChainsNo.Registry Status (Aug 23, 2026)
July 30, 2026Botanix1DEPRECATED
July 31, 2026Moonriver, Moonbeam, Nexera, Canto4DEPRECATED (all four)
August 28, 2026EDU Chain, Meter, Shimmer, Cyber, Silicon, Sophon, Bitlayer, DFK Chain, Arbitrum Nova, DOS Chain, Cronos zkEVM, Degen, Skale Europa, Superposition, Shrapnel1514 active · DOS Chain already DEPRECATED
September 30, 2026Aurora, Taiko, BounceBit, Japan Open Chain, LightLink, Otherworld Space, Viction, Anime, XPLA, Merlin, Gnosis, Zora12All twelve active
Total326 with label applied · 26 pending

The five from July already appear with chainStatus: DEPRECATED in the query from August 23, 2026; DOS Chain was decommissioned on its own before entering the schedule. The next date is August 28, four days after this publication.

The notice carries a nuance that changes who needs to worry: of the five July chains, only Botanix held Stargate assets. The notice table marks "n/a" in the Stargate column for Moonriver, Moonbeam, Nexera, and Canto, and the LayerZero message on X specifies that "some" chains will cease to be supported by Stargate Hydra — the Stargate mode that issues receipts on chains without their own pool. In those four, what is left without a verifier are the OApps and third-party OFTs that used the LayerZero Labs DVN and Executor by default: each application with its own balance and no public aggregate figure.

Why does a bridged token stop being redeemable when the service is turned off?

Imagine a store credit card for a retail chain. The day the only branch authorized to redeem it closes, the card remains in your wallet and the money remains in the company's account; what is missing is the counter that converts one into the other.

Botanix provides the specific case: it was the only one of the five July chains with Stargate Hydra assets — USDC.e and wETH — and its DVN ceased operations on July 30, 2026. When sending USDC from Ethereum to Botanix via Hydra, the original USDC did not travel: it remained locked in the Stargate pool on the source chain, and an equivalent amount of USDC.e was issued on Botanix — a different token, with another contract and another address. USDC.e and USDC are separate assets, just like wETH and ETH: the first of each pair is a receipt issued against collateral held elsewhere.

Exiting works in reverse: you burn the USDC.e, and that burn generates a message that must reach the collateral chain so the pool can release the native USDC. That message is transported by LayerZero, and its journey has two gatekeepers: the DVN observes the burn and signs that it occurred, and the Executor picks up the signed message and executes the delivery at the destination by paying the gas. Withdrawing both services from a chain is like removing both gatekeepers from that segment: the burn can be executed — the contract is still there — but the message proving it is not signed, and without a signature, there is no release of collateral.

The analogy ends here: at the store, there is a company to claim against; on an orphaned chain, no claim is possible because the system did not fail. There simply ceased to be anyone operating the part that does not live on the blockchain.

Where does the USDC.e collateral go when LayerZero withdraws the DVN and Executor?

Distinguishing between deployed code and a service with a payroll behind it decides everything else:

ComponentWhere it livesWho maintains itWhat happens if they disappear
Collateral (Native USDC)Stargate pool on source chainContract, no operatorRemains intact and locked
Receipt (USDC.e)Orphaned chainContract, no operatorStays in your wallet, transferable within that chain
Endpoint and messaging librariesOrphaned chainContract, no operatorRemain deployed and functional
DVN (Verification)Off-chainOperator (company or third party)The message is never certified
Executor (Delivery)Off-chainOperator, replaceableDelivery can be done manually

The first three rows are deployed code, and code does not turn off as long as validators produce blocks. The last two are services that someone runs on a server, paying for infrastructure and gas: the reversibility of a bridged asset depends on a business continuing to find it profitable to service that route.

Hence the counterintuitive part: the holder sees nothing. No transaction in their history, no notification; the explorer shows the balance normally and the wallet values it at the price of the underlying asset. The only signal is in a blog post they have no reason to read.

Checking the registry on August 23, 2026 — three weeks after the cut-off — Canto and Nexera retain the full set of Version 2 contracts: endpoint, executor contract, send and receive libraries, and endpoint viewer. It is the same inventory presented by any active chain: the DEPRECATED label lives in the catalog, not on the blockchain. Moonbeam (identifier 30126) also lists its own, but there the coincidence is deceptive: the network entered maintenance mode at 00:00 UTC on August 1, 2026, and rejects user transactions even though it continues to produce blocks.

Among these contracts appears deadDVN, a filler verifier that blocks messages from a route whose default configuration is inactive, present in 169 of the 184 mainnet deployments with Version 2 — including fully active chains — so its presence says nothing about the service status.

Can the user withdraw their funds on their own without the LayerZero DVN?

The question that decides whether this is convenience or custody is whether there is a redemption path the holder can trigger alone. LayerZero documentation answers this in two halves:

Delivery is rescueable. The Executor is, according to the documentation itself, a permissionless service that anyone can run: once the message is verified, any third party — including a user with technical knowledge and gas on both chains — can execute it by calling the contract. If only the Executor were turned off, this would be a technical nuisance.

Verification is not. The message only reaches the delivery phase when the DVNs configured for that route have signed it. If that DVN ceases to operate — as happened on Botanix on July 30, 2026 — the message remains unverified and there is no documented fallback mechanism. Changing which DVN monitors a route is the responsibility of the application owner or their delegate, which in Stargate is the protocol and never the holder.

The verdict is neither destruction of funds nor a simple delay. The asset exists, the collateral exists, and the return path can be reopened if someone reconfigures the route or operates a verifier for it. What remains is no unilateral action for the holder, and that is the operational criterion of custody: if moving your money requires a third party to decide something, the custody was not yours. The thesis survives with nuance: it is not destruction of value, but the silent conversion of a self-custodied balance into a dependent one.

What is the difference between a bridge being hacked and being abandoned?

Bridge security literature is built almost entirely on one threat model: someone forces a false verification — forges a signature, compromises a validator set, or exploits the verification contract — and the destination chain accepts as true a deposit that never occurred. Ronin, Wormhole, and Nomad — the big three of 2022 — are variants of over-verifying. In abandonment, no one verifies, and the consequences change across all dimensions:

DimensionAttacked BridgeAbandoned Bridge
What failsVerification accepts something falseThere is no verification of any kind
TimingInstant and without warningAnnounced weeks in advance
VisibilityMaximum: headlines, threads, forensicsMinimum: a blog post
Who losesLiquidity providers on the attacked sideReceipt holders on the orphaned chain
CompensationPossible (treasury, rescue, insurance)None planned: no breach occurred
Useful early signalAudits, signer concentrationDrop in activity, withdrawal of incentives

The visibility asymmetry has a perverse effect: an attack generates an incident that everyone studies, while abandonment generates a blog post that no one links to. For the attack model, see bridge architecture after the hacks, why the bridge is the weakest link in DeFi, the 2026 bridge comparison, and the showdown between LayerZero, Wormhole, and Axelar. For code that remains standing after its operator leaves, see the case of Aztec Connect.

How many chains has LayerZero marked as deprecated and how much Stargate value remains exposed?

The registry allows for measuring the full phenomenon, not just the July episode. As of August 23, 2026, it contains 480 entries, of which 211 are production networks: 173 active, 3 private, and 35 with the DEPRECATED label. One in six production networks in the catalog of the largest cross-chain messaging network is already listed as a retired service. Of those 211, 184 have Version 2 deployed, which is the basis for the contract counts in the previous section.

The list of 35 includes, in addition to the five from July, names that once had their own ecosystems and funding rounds: Boba, Swell, zkEVM, zkLink, Rarible Chain, Sanko, Bahamut, PGN, or Tenet. Also Plume, for a different reason: its entry was decommissioned when the network relaunched under the key plumephoenix, which is listed as active. For the others, there is no common event; there is a continuous regime of pruning that has occurred without a news cycle.

There is no public data series for the value trapped in the five July chains: DefiLlama has never published Stargate TVL for any of them, and as of August 23, 2026, the protocol is listed with a presence on 24 chains, with no trace of those five even in the history. What can be measured is what lies ahead. Three of the twelve chains with a September 30 date have a published Stargate v2 series, and one concentrates the bulk: Gnosis, which appears in the notice table with the express instruction for S*USDC and S*ETH liquidity providers to withdraw their positions.

ChainCut-off DateStargate v2 (Aug 23, 2026)All-Time HighDate of High
GnosisSep 30, 2026$1,675,069$5.46 millionNov 10, 2025
AuroraSep 30, 2026$382,344$2,724,169May 9, 2025
LightLinkSep 30, 2026$101,613$891,265Oct 7, 2025
Total Exposed$2,159,026

That is $2.15 million in Stargate liquidity across three networks with an announced expiration date and thirty-seven days remaining at the time of this publication. The highs explain the withdrawal: Gnosis retains 31% of the liquidity it once had, Aurora 14%, and LightLink 11%. The liquidity left before the service did, and the cut-off catches the tail end that stayed inside.

The figure is a floor: Taiko, EDU Chain, and Superposition also appear in the notice table with Hydra assets pending redemption, and DefiLlama does not publish a Stargate series for any of them. For the August 28 wave, only the TVL of the entire network is known, serving as an order of magnitude: Bitlayer $448,030, DFK Chain $366,493, Arbitrum Nova $360,292, and Cyber $2,703, according to DefiLlama as of August 23, 2026. This is an ecosystem ceiling and does not measure the funds at risk of losing their route.

What had Botanix warned before LayerZero withdrew its verifier?

The first chain on the schedule has a prior history that resets the holder's clock, and it is the most instructive case of the thirty-two: it shows the two layers falling in a cascade.

DateEventAffected Layer
July 1, 2025Botanix launches its mainnet, a Bitcoin L2Chain
June 10, 2026Announces shutdown with a public post-mortemChain
July 9, 2026Final withdrawal deadline: the federation — the group of signers custodying deposited Bitcoin — sweeps what remainsChain
July 24, 2026LayerZero announces withdrawal of DVN and ExecutorMessaging
July 30, 2026Botanix moves to DEPRECATED in the registryMessaging

The chain died twenty-one days before the messaging provider withdrew its service. Anyone monitoring only LayerZero notices received their alert after the real deadline had passed by more than two weeks, because the clock that mattered was Botanix's: after July 9, any assets remaining on the network would be swept by the federation. The effective window lasted twenty-nine days, from June 10 to July 9.

The shutdown numbers explain the arithmetic of these decisions. Botanix raised $11.5 million in public rounds — $3 million pre-seed in 2023 and $8.5 million seed in May 2024 — although the CoinDesk closing piece raises the total figure to $14.4 million. It processed about 25 million transactions and attracted around 200,000 wallets without token incentives or airdrops, and closed with $119,500 in TVL according to DefiLlama. The post-mortem argued that the technology worked — 100% availability, zero security incidents — and that what was missing was demand: users preferred wrapped Bitcoin in large ecosystems over native Bitcoin infrastructure.

Neither layer did anything anomalous: a company closed a product without demand giving twenty-nine days' notice, and an infrastructure provider stopped paying for servers for a route without traffic giving thirty. The loss mode resulting from adding them together does not appear in the documentation of either.

How to detect in time that a chain you use is losing support?

Five checks to detect that a chain is losing messaging support before the formal notice, in this order.

  1. Separate native assets from representations. An asset is native if the contract issuing it is the canonical one for that network — Circle's USDC on Ethereum, the chain's own ETH. It is a representation if its value depends on collateral custodied elsewhere, and the clue is usually in the name: the .e suffix, the w prefix, or the words bridged or wrapped in the explorer. Every representation carries a service dependency; native assets do not. The what is a bridge guide elaborates on this distinction.
  2. Watch activity, not price. Signs of abandonment are operational and appear months before the notice: TVL falling steadily for quarters — Aurora at 14% of its Stargate liquidity peak, LightLink at 11% — incentives not being renewed, large protocols stopping the deployment of new versions, and the network client without updates in the repository. A chain that has lost 85% of its liquidity in a year and a half is a candidate for this schedule rather than a cheap opportunity.
  3. Identify where notices are published. They appear on the infrastructure protocol's blog and social media accounts, not in the application you used to bridge nor in your wallet: no channel connects "the messaging provider withdrew this route" with "you have $3,000 in a token on that route." With positions on small networks, the chain's blog is ahead of the provider's blog.
  4. Exit in the correct order. Redeem the representation for the native asset before moving anything else; a frequent mistake is bridging the bridged token to another small chain, which chains one dependency upon another. The safe route is toward a network with deep liquidity and several messaging providers competing for it, even if the gas that day is worse.
  5. Apply a size rule. If the TVL of an entire chain fits comfortably within your position, the asymmetry is no longer in your favor: there will be no exit liquidity when you need it, nor a critical mass of affected users to push for keeping the route alive. Cyber, with a cut-off on August 28, reported $2,703 across the entire network on August 23, 2026, according to DefiLlama: any four-figure position there is a substantial part of the ecosystem.

What can I do if I left funds on a chain that LayerZero no longer supports?

Distinguishing which of the two layers fell determines if there is anything left to do.

If the chain is still producing blocks and only messaging was withdrawn — the clean cases are Canto and Nexera, with no announced closure — the balance remains in your wallet and is still transferable within that network. It is worth checking if any local market provides a counterparty for that asset and if another messaging provider maintains routes to the outside: the withdrawal affects a specific provider and does not describe a property of the network. The default LayerZero Labs DVN stops operating; the registry still lists eight third-party DVNs on Canto and five on Nexera — Nethermind, Horizen, Canary, or Nansen among them — whose activity is not publicly verifiable. We have not performed a test transaction. It is also worth asking in the support channels of the protocol that issued the receipt, the only actor with the capacity to reconfigure the route.

If the entire chain was shut down — Botanix, Moonbeam, and Moonriver — the margin changes with each case. In Botanix, with the federation sweep on July 9, there is no procedure: the asset ceased to have a network where it could exist, and the collateral remains locked in the source pool with no one to certify the receipt burn. Moonbeam and Moonriver closed as chains on July 31, 2026, with a 1:1 migration of GLMR and MOVR to Base, and Moonbeam has been in maintenance mode since 00:00 UTC on August 1: it rejects user transactions even though it continues to produce blocks. The standard window closed on July 31 at 23:59 UTC, and those who arrived late are referred to an email helpdesk, with case-by-case review and no public claims portal.

What does not work in any case: there is no claim, because no commitment was breached; no DeFi insurance covers infrastructure withdrawal, as policies are written against contract failures and attacks; and collective pressure does not arrive when the aggregate amount of the affected group is below the cost of operating the solution.

The practical consequence fits into a question applicable to any position before opening it: if tomorrow the team maintaining this route decides it's not worth it, can I exit on my own? For a native asset on a network with independent validators, the answer is yes. For a bridged representation, it depends on someone continuing to sign messages: a service provided as long as it remains profitable.

The July precedent leaves a pattern with verifiable dates: 35 of 211 production networks decommissioned in the registry as of August 23, 2026, fifteen more chains on August 28, twelve on September 30, and $2,159,026 in Stargate liquidity still sitting on three of them. The notice arrives with weeks to spare and the mechanism is predictable; the only requirement to avoid being stuck inside is knowing that the reversibility of a bridged token is not written on the blockchain, but in a company's operations budget.

Sources and links: LayerZero — Support Update (July 24, 2026, updated Aug 21) · LayerZero on X — five-chain notice · PANews — July 25 schedule reading · LayerZero Public Metadata Registry · LayerZero V2 Documentation (DVN and Executor) · Stargate — Hydra Documentation · DefiLlama — Stargate v2 by chain · DefiLlama — Chain Rankings · Crypto Adventure — Announcement Coverage · AMBCrypto — Botanix Shutdown · Crypto Briefing — July 9 Withdrawal Deadline · CryptoSlate — Moonbeam Halts User Transactions · Moonbeam — Relaunch on Base and 1:1 Migration

Related Articles: Bridge Architecture After the 2026 Hacks. LayerZero, Wormhole, and Axelar Compared. Aztec Connect: The Risk of Code Without an Operator. What is a Blockchain Bridge. Monitor your positions on CleanSky — our portfolio tracking tells you which network each balance lives on, which is the first piece of data you need to know which are representations and which are native assets.