Two numbers were published. One denominator was not.
Arbitrum's monitoring reports that $2.5B has moved through Mayan, of which $2.3B is stablecoin. That is a 92% stablecoin share. General-purpose bridges typically run 40–70%, because they must also carry ETH, SOL, and long-tail assets across the speculative boundary.
A 92% skew is not a rounding artifact. It is an architecture fingerprint.
But before the fingerprint means anything, the metadata has to survive scrutiny. There is no stated time window. No methodology — no Dune dashboard reference, no indexer, no deduplication rule. No second source. DefiLlama, Wormhole, and Circle are all absent from the citation chain. The publisher is the ecosystem itself, monitoring its own inflow.
So the honest framing is this: the $2.5B is probably real on-chain; the narrative wrapped around it is constructed. This piece decomposes the second part.
Mayan is not a liquidity pool bridge. It is an intent-based settlement layer. The mechanics matter because they determine what the flow number can and cannot tell you.
A user signs an intent: deliver USDC on Arbitrum, I will pay from Solana. Solvers — professional market makers with pre-positioned inventory on the destination chain — bid in a Dutch auction for the right to fill. The winning solver fronts the capital, the user receives funds in seconds, and the solver is later repaid on the source chain. Settlement of the repayment leg runs over a message layer: either Wormhole's 19-guardian validator set, or Circle's CCTP, which burns USDC on the source and mints it natively on the destination after attestation.
The asset path is the trust surface. A lock-and-mint bridge carries bridge-contract risk — the wrapped asset is only as good as the vault behind it. CCTP carries no wrapped-asset risk at all, but collapses trust to a single permissioned issuer: Circle, and its blacklist function.
This is where the 92% becomes diagnostic. If the dominant payload is USDC, and CCTP exists on both ends, rational solvers route through CCTP. Zero slippage, no wrapper, near-zero fee. The 92% share is therefore consistent with one conclusion: this channel is a treasury plumbing route, not an asset-migration route.
The disclosure lands in a sideways market, which is precisely when ecosystem data gets repurposed for positioning. In chop, attention-starved L2s publish cumulative inflow figures because cumulative figures only go up. Interval figures can decline. That asymmetry is the whole reason the time window is missing.
Gross inflow is not net stock, and the disclosure never distinguishes them. $2.3B entered Arbitrum through Mayan. The only figure that matters for ecosystem health is the delta in Arbitrum's stablecoin supply over the same window. If that delta is materially smaller than $2.3B, the money is in transit, not in residence — passing through to a yield venue, a CEX withdrawal, or a fresh TGE, then leaving. A channel can process enormous gross volume while contributing almost nothing to the destination's balance sheet. This is the single largest logical gap in the release, and it is not an oversight; it is the difference between a headline and a finding.
The causality is almost certainly inverted. The more plausible chain is: Arbitrum offers a yield or issuance opportunity, capital wants in, capital needs a cheap route, Mayan supplies the route. Mayan is the effect, not the cause. I ran into the same inversion repeatedly during the 0x Protocol work in 2017 — relayers were credited with creating order flow that actually followed wherever the deepest book already sat. Routing infrastructure does not generate demand. It captures it, taxes it at a few basis points, and loses it the moment a cheaper path appears.
Flow and revenue decouple almost completely under the intent model. The protocol's take is the auction spread — typically low single-digit basis points after solver competition. At 3 bps, $2.5B gross yields roughly $750k in protocol-level value, most of which is competed away to solvers and users. The $2.5B is a throughput metric wearing a growth metric's clothing. Volume is not commercial scale.
Bridge volume is among the most gameable statistics in Web3. Round-tripping is trivial and historically rampant: one dollar crossing five times reads as five dollars. Without a disclosed address-deduplication rule and without a stated incentive window, the number must be discounted by an unknown factor. My ERC-721A gas analysis in 2021 taught me the same lesson in a different register — a metric that is cheap to produce will be produced cheaply, and the published figure will reflect the production cost, not the underlying demand.
Half the volume arrives in tickets of $100k or more. That reads as institutional validation. Operationally it means the opposite: wholesale flow is the least loyal capital in the system. It is rate-elastic to the basis point, and its half-life after an incentive ends is measured in weeks. Retail flow leaves a residue — wallets, habits, positions. Treasury flow leaves a timestamp.
Value capture in bridging sits at the bottom of the DeFi stack. A bridge sells transit. Switching cost is zero, price is fully observable, and the 92% stablecoin segment is the most transparent price market in the entire industry — CCTP has effectively set the clearing price near zero. There is no liquidity moat here, because solvers hold the inventory and can reallocate it in a block. Compare that to an AMM, where TVL is sticky by construction.
The 92% is precisely the segment being commoditized by its own upstream. Circle's CCTP turns stablecoin transfer into a public good. Every expansion of native USDC minting erodes the fee surface of the aggregation layer sitting on top of it. The unintended consequences of CCTP's success are not felt by Circle; they are felt by the routers. An intent aggregator whose payload is 92% USDC is, structurally, a UX wrapper on a permissioned issuer's API.
A single direction of flow carries a hidden concentration risk. If the dominant route is Solana to Arbitrum, Mayan's fate is bound to Solana's capital outflow rhythm rather than to Arbitrum's own attractiveness. Half the tickets at $100k or more also implies the channel may be dominated by a handful of addresses. A bridge whose volume is concentrated in a few treasury desks is one withdrawal away from a flat line, and the disclosure provides no address-level distribution.
The competitive set is crowded and converging. Across runs the same intent model. deBridge and Stargate compete on the same stablecoin corridor. LayerZero-based routes and native CCTP direct integration bracket the market from both sides. Differentiation in this layer is measured in latency and basis points, not in architecture. A protocol that competes only on price competes with everyone, permanently.
| Disclosure item | Status | Severity | |---|---|---| | Time window | absent | high | | Cumulative vs interval | unstated | high | | Methodology / dedup | undisclosed | high | | Independent cross-check | none | high | | Gross vs net | not distinguished | high | | Publisher | ecosystem self-monitoring | medium |
The contrarian reading is not that Mayan is failing. It is that the reader is being handed a metric engineered to produce a specific belief, and the unintended consequences of that choice land on the reader's model, not on the publisher's reputation.
Here is the security blind spot that the flow number quietly obscures. No audit status is disclosed. No contract upgradeability, no open-source confirmation, no solver admission standard, no slashing or insurance mechanism. For a bridge clearing six-figure tickets, that is not a footnote. The intent model's characteristic failure is not a reentrancy bug; it is a solver that fills, then fails to settle, leaving funds stranded between two chains with no canonical owner. That failure mode has no clean rollback.
And the deeper structural point: when 92% of the settlement asset is USDC, the protocol's censorship resistance is delegated to Circle's freeze list. Calling that a decentralized bridge is a category error. The decentralization lives in the auction; the finality lives in a permissioned attester set. Two different trust domains, marketed as one.
Watch two numbers over the next two quarters, and ignore the cumulative headline. The non-stablecoin share of Mayan's Arbitrum volume — if it stays near 8%, the business is a CCTP front-end and will be priced as one. And unincentivized retention: gross flow in a period with no points program, no grant, no TGE. My forecast is that the intent aggregation layer's value window is transitional, not durable. It exists because CCTP coverage is still incomplete and solver inventory is still scarce. Both conditions are temporary. The unintended consequences of publishing a cumulative number are that the number outlives the conditions that produced it. The question worth holding is not how much money crossed the bridge. It is what remains on the other side once the crossing stops paying.
