On September 29, 2025, Coinbase's Chief Policy Officer — not a product lead, not an engineer — stood up to announce Sivar, a stablecoin remittance app for El Salvador. Two numbers should stop you cold before the marketing copy does.
The first is 25,000. That is the pre-registration count. El Salvador has roughly 1.6 million people who depend on remittances. Pre-registration is not activation, and activation is not retention. Even at full conversion, the product reaches about 1.6% of the population it claims to serve.
The second number is 1,000. That is the count of physical cash redemption points Sivar will lean on for off-ramping. MoneyGram, which quietly launched its own stablecoin settlement in El Salvador months earlier, operates through a retail network of roughly 500,000 locations across 200-plus countries. The gap is not a rounding error. It is a 500-fold structural disadvantage in the exact layer that decides whether a remittance product survives.
Check the source code, not the hype. Here the source code is a stack of mature components bolted together. The interesting part is who captures the value.
El Salvador's remittance corridor runs about $9 billion a year, with roughly 92% of that originating in the United States. This is the market Sivar is chasing, and it is a real one — not a narrative confection. Families in San Salvador and Santa Ana genuinely depend on dollars wired from Los Angeles, Houston, and Washington.
Sivar's technical stack, assembled from public statements, runs as follows. Settlement on Base, the Coinbase-operated Ethereum Layer 2 built on the OP Stack. A dollar stablecoin — almost certainly USDC given the Coinbase and Circle infrastructure involved, though the reporting never confirms the ticker. Debit-card funding on the US side. An in-app non-custodial wallet. Off-ramp through 1,000-plus physical cash points. The developer is a company called Modveon.
The announcement lands in a specific regulatory moment. US stablecoin legislation has passed, institutional adoption is accelerating, and the post-GENIUS framework has turned stablecoin payments into a crowded narrative. El Salvador, meanwhile, has been quietly rewriting its own crypto posture — not toward Bitcoin maximalism, but away from it. That is the context the coverage missed. This is not a crypto-payment victory lap. It is a country pivoting under external constraint.
Start with the settlement layer, because the marketing implies decentralization that does not exist. Base currently runs a single sequencer, operated by Coinbase. That means ordering, censorship resistance, and uptime for every Sivar transaction route through one corporate actor. This is not a bug specific to Sivar — it is the structural state of most "Layer 2" scaling today. But when you are moving household remittances, "single sequencer" is not an academic footnote. It is a single point of failure and a single point of control simultaneously.
Then there is the custody claim. Sivar markets a non-custodial wallet: users hold their own keys. Technically true. Operationally misleading. The on-ramp depends entirely on Coinbase's card infrastructure; the off-ramp depends on Sivar's partner cash network. Funds move freely on-chain, but the path to convert them back to spendable cash is gated at two ends by single entities. This is formally non-custodial and substantively single-point-dependent — the user holds the key but does not hold the exit.
I spent 200 hours in 2024 reviewing custody solutions for the Bitcoin ETF applicants, and I flagged a flaw in Fireblocks' multi-party computation implementation that exposed a sliver of assets to single-point failure. That memo was ignored internally; I published an anonymized version. The lesson carries directly here: the interesting risk in "trusted" infrastructure is never the cryptography. It is the operational chokepoint wrapped around it.
Now the unit economics, which the coverage treated as settled. The headline is a $2 flat fee. The problem is that card-network and processing fees on the US funding side typically run in the low single digits as a percentage of transaction value. On a $100 card top-up, the visible fee alone can approach or exceed the $2 charged. Add FX spread, cash-point commissions, compliance overhead, and fraud losses, and the cost stack is undisclosed at every layer. A flat $2 fee is not always cheaper for small transfers. Below roughly $33, it represents a rate above 6%. The genuine advantage window is mid-sized transfers — $100 to $500 — where a flat cost beats a percentage. The claim that Sivar targets a market where percentage fees erode small payments does not survive the arithmetic, because legacy operators often use flat structures too.
Run the market math and the narrative deflates further. A $9 billion corridor, assuming an average transfer of $200 to $300, implies roughly 30 to 45 million transactions a year. At 100% market share and $2 per transfer, that caps annual revenue near $60 to $90 million. At a realistic 10 to 20% share, it falls to $6 to $18 million. For Coinbase, booking billions annually, that is noise. The strategic return is not in the P&L.
The funding source for all of this is worth stating plainly. El Salvador fully dollarized in 2001. Adding stablecoin rails deepens that dollarization. The interest earned on the reserves backing USDC accrues to Circle; the sequencing fees accrue to Coinbase; the float accrues to the issuers. The Salvadoran user receives a cheaper settlement channel and nothing else. The value capture is almost entirely externalized — the country is a demand-side user, not a stakeholder.
Three technical tracks are now competing in the same corridor. Coinbase plus Base plus a dollar stablecoin, which is Sivar. Stellar plus Circle's USDC, which is the MoneyGram path. And Bitcoin's Lightning Network plus Tether's USDT, which is the Tether path. The technical differences between them are marginal. The decisive variable is the density of the cash network and the cost of compliance — not throughput, not finality, not consensus design.
Then the part nobody is auditing: the 1,000 cash points. Off-ramping physical dollars requires a network of agents, and that network is a de facto money-services business. Under US law, money services businesses must register and run AML programs. Under Salvadoran law, the obligations are less clear. The reporting discloses nothing about who operates these points, what KYC they perform, or how sanctions screening is handled. Regulations are lagging, not absent — and in a corridor moving $9 billion a year, that gap is not a technicality. It is the single most under-priced risk in the entire structure.
I have seen this pattern before. At 25, I led a compliance audit of NovaChain, a privacy-focused L1, and documented 45 specific instances where its ZK-rollup implementation failed NYDFS capital reserve requirements. The eventual fine was $2.4 million. The findings were not exotic. They were mundane gaps in reserve attestation and reporting that nobody had bothered to check because the technology narrative was more exciting than the compliance file.
Now the sovereignty question, which the bullish coverage treats as background noise. El Salvador is operating under a $1.4 billion IMF program. That program has already produced concrete policy reversals: the removal of Bitcoin's mandatory legal-tender status, voluntary private-sector acceptance, dollar tax payment, and withdrawal from the Chivo wallet. The IMF has stated that no public resources were used to accumulate Bitcoin. Translation: the government's active accumulation channel is effectively closed.
The 7,789 BTC on the sovereign books — roughly $780 million at a $100,000 price assumption — is no longer an accumulating reserve. It is a frozen position. If Bitcoin falls sharply, the mark-to-market loss hits the sovereign balance sheet directly, and the government has committed to not buying more to average down. Past performance predicts future panic, and this position has no averaging mechanism left.
Finally, the governance vacuum. The Bitcoin Office, the IMF, Coinbase, Tether, MoneyGram — every counterparty has a name, a paper trail, and a background. The one entity that actually operates Sivar, Modveon, has none. No founders. No funding history. No prior delivery record. No audit disclosure, no open-source status, no contract upgrade mechanism. For a product carrying national-scale remittance flows, publishing a company name and nothing else is a significant information asymmetry, not a footnote. A non-custodial label reduces the operator's fiduciary duty; it does not reduce its compliance duty.
Here is what the bulls got right, and it deserves a fair hearing. The structural trend is real. Stablecoin remittance is not a fad; it is a cost curve about to move systemically downward. The most important signal in this entire story is not Sivar. It is that MoneyGram — a legacy remittance giant — proactively rebuilt its settlement rails on Stellar with Circle's USDC. When the incumbents embrace the technology rather than fight it, the direction is settled.
Coinbase's motivation is also rational, just not the one being advertised. A sub-$100 million annual revenue opportunity is immaterial to a company booking billions. But a real, sovereign-scale national deployment is enormously valuable political capital in the middle of US stablecoin policy debates. That is why a policy officer, not a business lead, fronted the announcement. The ROI here is measured in influence, not income — which means the project can be sustained on political logic and, just as easily, wound down when that logic fades.
And El Salvador is being more rational than it gets credit for. By hosting Tether's relocated headquarters while simultaneously running a Coinbase and Base deployment, the country is deliberately keeping two competing stablecoin paths alive rather than being locked into one counterparty. That is not naivety. That is hedging. The narrative frame has shifted from "Bitcoin is the future of money" to "Bitcoin is a strategic reserve and stablecoins are the money people actually use" — and that shift is happening faster than the market's mental model of El Salvador has updated.
So read this as what it is: an infrastructure-shift experiment documented in a country that has lost the room to say no. The forward question is not whether Sivar succeeds. It is whether anyone will publish the 90-day active-to-registered ratio. If that number comes in under 20%, the model does not replicate. Watch three signals and nothing else: the compliance arrangements around the 1,000 cash points, whether the sovereign BTC position resumes accumulation under IMF constraint, and whether Washington tightens cross-border stablecoin flows. Liquidity vanishes; insolvency remains — and in a corridor this small, the first thing to vanish is the narrative.