MoneyGram ran a Stellar-based settlement pilot from 2021 through 2023, then ended it without a public post-mortem. In 2025 the same company returned to the same problem โ cross-border payout float โ with a different vendor, Rain, and a Visa-branded card issued to users in Colombia. The press cycle filed it under "crypto adoption." That label is wrong by exactly one accounting layer.
Colombia received just over $10 billion in remittances in 2023 and approached $12 billion in 2024, the fourth-largest inflow in Latin America. The binding constraint in that corridor is not blockchain throughput. It is pre-funding: the idle dollars that MoneyGram must park in nostro accounts at Colombian correspondent banks so that an agent can pay out in minutes rather than days. Idle capital earns nothing and costs something. Stablecoins are not a payment rail in this design. They are a cheaper liability to hold against a payout obligation that already exists.
That distinction determines how the entire product should be audited.
Context: three vendors, one balance sheet
MoneyGram is a money transmission company with roughly 350,000 agent locations, a US listing, FinCEN supervision, and money transmitter licenses in most states. It is not a crypto company and has never been priced as one. Its margin is the spread between what it charges to move value and what it costs to settle that value at the destination.
The 2021 Stellar partnership was supposed to close that spread. The Stellar Development Foundation put money into the relationship through its enterprise fund, and MoneyGram was to settle USDC through Stellar rails. It wound down quietly โ no breach, no scandal, just a cost comparison that did not clear the bar. That history matters, because the new card is a second attempt at the same problem with the same underlying asset and a different chain abstraction. A company that abandons one settlement rail and adopts another within twenty-four months is telling you the rail was never the constraint.
Rain is the less visible party. Rain is a card program manager โ a crypto-native firm sitting between a sponsor bank's BIN, the Visa acceptance network, and a stablecoin reserve. Program managers do the unglamorous work: card issuance, KYC orchestration, dispute handling, treasury, and the reconciliation that lets a Visa authorization clear against a dollar balance. That is where the technical risk of this product lives, and it is disclosed nowhere in the announcement.
Visa itself has settled in USDC against acquirer obligations since 2023, first on Ethereum and later on Solana. Visa's role here is not innovation. It is plumbing that already exists and already has a published fee schedule.
Colombia is a rational first market for three reasons. Remittance volume is large and effectively dollarized at the household level. The central bank's Bre-B instant payment system is in phased rollout, which means the incumbent slow rails are about to be repriced whether or not stablecoins participate. And the supervisor has not blocked banks from servicing crypto-adjacent flows. Choosing Colombia is not a statement about the technology. It is a statement about which regulator will let the experiment run while it is still small enough to fail quietly. The same logic explains why Hong Kong drafted a virtual asset licensing regime calibrated to capture institutional flow that would otherwise route through Singapore. Jurisdictions compete for license revenue and clearing volume. They do not compete for ideology.
Competition in this corridor is also more crowded than the announcement implies. Western Union operates the same agent network model. Wise prices on mid-market rates with a transparent fee. Stellar still processes corridor payments through anchor networks. The stablecoin card does not enter an empty market; it enters a market where the incumbents have already optimized the visible fee and are still carrying the invisible one. The innovation is not the card. The innovation is the removal of a pre-funding line item that no competitor has yet been willing to disclose.
Core: where the chain actually sits
Strip the marketing and the Colombian card resolves into four layers.
Layer one โ authorization. A user taps a Visa card in Bogota. Visa authorizes against a balance held by the program manager. Nothing here is on-chain. Latency is measured in milliseconds and governed by ISO 8583 message standards that predate Bitcoin by four decades.
Layer two โ funding. Rain converts a user's stablecoin balance, or receives a fiat wire, to satisfy the authorization. If the user holds USDC, the card is a spend interface over a stablecoin reserve. If the user holds pesos, the stablecoin is invisible and exists only on Rain's side of the ledger.
Layer three โ treasury. This is the only layer where a blockchain plausibly appears. MoneyGram's Colombian payout obligation can be extinguished by moving stablecoins between its own accounts rather than by maintaining a fully pre-funded peso position. The chain function is reconciliation and transfer between entities that already trust each other contractually. That is a private settlement problem dressed in public infrastructure.
Layer four โ redemption. Stablecoins convert back to pesos for the agent, the ATM, or the merchant. A licensed bank or exchanger absorbs the reserve and the regulatory obligation attached to it.
The performance claim implied by the product โ faster, cheaper payout โ is therefore a claim about layer three. It is testable. The relevant metric is the float reduction ratio: how many dollars of pre-funding does MoneyGram retire per dollar of stablecoin reserve it holds? At 1:1, the product is a marketing exercise. Materially above 1:1, the product is real, and the savings accrue to MoneyGram's treasury rather than to the sender. The absence of that ratio from the launch materials is the single most informative fact about the launch.
Placing this against a liquidity-cycle framework is useful. I use a three-axis matrix โ funding cost, settlement latency, and reserve transparency โ and score payment products against it. MoneyGram scores well on the first axis if stablecoin reserves replace dollar nostro balances, neutral on the second because Visa authorization latency dominates the user experience regardless of the settlement rail, and unknown on the third. An unknown score on reserve transparency is not a neutral result. It is the axis that determines whether the product survives a stress event. The matrix is not decoration. It is the only way to compare a stablecoin card against a correspondent bank without relying on either party's marketing.
Three further technical points deserve scrutiny.
First, reserve transparency. A card that redeems against a stablecoin reserve inherits that reserve's risk. Attestation is not audit. A monthly letter confirming a point-in-time balance does not disclose composition, maturity profile, or counterparty concentration. In a payment product, the relevant failure mode is not a depeg lasting three days; it is a redemption gate lasting three hours inside a settlement window. Exit strategies are written in ice, not in hope, and a card program's exit strategy is its reserve attestation standard.
Second, the yield question. If Rain holds float in stablecoin form, that float earns. Where the yield goes determines who subsidizes whom. Money market funds paid roughly five percent on short Treasury exposure through 2023 and 2024. On-chain lending markets price deposits off a utilization curve that has no structural relationship to the funding markets that actually set short rates โ those models are administrative parameters wearing the costume of market discovery. A card program funding reserves through an on-chain pool is taking basis risk it does not need. The defensible structure is a Treasury-collateralized reserve with a disclosed yield split.
Third, settlement layer choice. Rollups have spent two years pitching themselves as the venue for exactly this kind of low-value, high-frequency flow. The economics do not currently support it. Post-Dencun blob space cleared the fee floor for a while, but blob demand scales with rollup count rather than payment volume, and blobs per block are capped. Blob saturation is a scheduling problem, not a capacity problem, and when it arrives the marginal cost of each batch lands back on the users of that batch. A remittance corridor cannot absorb fee variance measured in multiples. Payment settlement will stay on rails with contractually predictable pricing, which is why Visa settled in USDC on Solana โ deterministic fees โ and not on the cheapest rollup of the month.
Contrarian: adoption is the wrong numerator
The standard reading is that stablecoins are being adopted by traditional finance. Reverse the arrow.
Stablecoin supply correlated with crypto conditions for a decade because the dominant demand was speculative: traders parking collateral, exchanges holding working capital, DeFi protocols accumulating liquidity. That regime is ending. The demand now arriving is transactional โ remittance float, merchant settlement, card reserves, corporate treasury. Transactional demand correlates with trade volumes, rate differentials, and corridor costs. It does not correlate with Bitcoin's price.
Two consequences follow. First, stablecoin issuance is becoming a short-duration government bond business with a card attached, and its growth will track the T-bill market rather than the crypto cycle. Second, the risk changes shape. Speculative demand exits quickly and loudly. Transactional demand exits quietly, and only when reserves are questioned. A payment stablecoin does not fail at the peg. It fails at the redemption window.
Takeaway: what to watch, and in what order
Three signals, in sequence. One: whether Rain publishes a reserve attestation with named custodians and a maturity schedule. Two: whether MoneyGram's next quarterly filing discloses a reduction in settlement assets, which is the only honest measure of the product's value. Three: whether the program expands to a second corridor under a different regulator, because a single-market launch is a pilot with a press release, and a two-market launch is a strategy.
The card is not the story. The story is that a listed remittance company has decided its largest cost center is idle capital, and it is now willing to hold that capital in token form. Whether that decision survives its first redemption stress test is a question nobody in the press cycle has asked.