I keep a spreadsheet of conference agendas. It is an odd hobby for someone with a doctorate in cryptography, but it started in the summer of 2017, when I spent three months manually auditing fifteen early-stage ICO contracts for a Seattle meetup group and discovered something that had nothing to do with Solidity: the documents people circulate tell you far more about what they want you to believe than about what they have actually built. Three of those contracts had reentrancy holes deep enough to drain them. Every one of them had a whitepaper describing a global settlement network. The gap between the paper and the byte was never a lie, exactly. It was a projection โ an honest-enough description of an intention, dressed in the grammar of a fact.
So when Money20/20's agenda dropped with six stablecoin sessions across two days, I did what I always do. I opened a blank sheet and started highlighting titles.
Two of them stopped me.
The first: "Your Stablecoin Pilot Will Be Ready for Production After This Session." The second: "Stablecoins Without the BS."
Read those again, slowly, the way you would read a function signature before you read the body. The first title is not a declaration that the industry has entered production. It is an admission that a large number of institutions are still sitting in pilot, and that they require a scheduled hour of hand-holding to cross the threshold. If the production era had genuinely arrived, nobody would book a room to explain how to enter it. The second title is stranger still โ it is the industry's own infrastructure vendors telling a paying audience that the surrounding conversation is padded with nonsense.
That is the whole story, and it arrived in the marketing copy rather than in the panels. The narrative has shifted from whether stablecoins work to how they scale. That shift is real, and it matters. But a shift in narrative is not a shift in fundamentals, and the article that framed this agenda as proof of a completed transition committed a specific, identifiable reasoning error: it treated the density of a conference schedule as evidence about the maturity of a settlement system.
I have spent enough time listening to the silence between market cycles to know that the loudest signal in a room is usually the one nobody is scheduled to discuss. So let me walk through what is actually on that agenda, what is conspicuously absent from it, and what the arithmetic says once you supply the denominators that the promotional copy leaves out.
The Room, and Why the Room Matters
Money20/20 is not a developer conference. It is a commercial one โ the annual clearing house where card networks, acquiring banks, processors, remittance operators, and the compliance vendors who serve them negotiate the shape of the next eighteen months. When a topic gets six dedicated sessions in two days, that reflects a sponsorship and exhibitor calculus: which subject sells floor space, which subject fills a keynote theater, which subject the attendees' bosses have asked them to come home with an answer about. It is a demand signal from the buy side of the payments industry. That is genuinely informative, and I do not want to dismiss it.
The roster assembled around those sessions is the more useful artifact. Visa Direct, described as running in commercial production. Mastercard, which acquired BVNK rather than building the capability internally. Western Union, the incumbent remittance network, appearing as a participant rather than as a disruptor. VersaBank on the banking side. Fireblocks on custody and movement. The Atlantic Council supplying the policy framing. Monad Foundation representing a high-performance Layer 1. Rain on card issuance. Stablecoin Blueprint and Velocity rounding out the vendor tier. And a panel built around a number โ one and a half trillion dollars โ with a title that implicitly interrogates whether that number means anything.
The claimed figures floating through the agenda material are these: four hundred billion dollars in stablecoin payment volume over roughly eight months, a seven-fold growth rate, and that one-and-a-half-trillion figure whose time frame and definition are not specified. None of the three arrives with a named source institution, a methodology note, or a base period. In my 2024 study of the spot Bitcoin ETF inflows โ where my team of four spent three months quantifying fifteen billion dollars of institutional capital against realized volatility โ the single hardest part of the work was not the modeling. It was establishing what counted. Which custodians, which creation units, which settlement dates, which double-counted basis trades. A number without a definition is not evidence. It is a mood.
And then there is the regulatory clock, which is the piece the promotional framing most needs to ignore. The GENIUS Act framework, as described in the agenda material, takes effect on January 18, 2027. That is not a rounding detail. It is the load-bearing wall. A payment rail is not in production because a card network runs a corridor; it is in production when the legal finality of a transfer is settled, when the reserve backing is defined by statute rather than by a blog post, when redemption rights are enforceable and their timing is specified. Until the implementing rules exist โ and implementing rules reliably lag statute by a year or more โ the industry is operating in a provisional state that is best described as production-adjacent.
I came to this subject through central bank digital currency research, which means I spend my working hours on the least glamorous question in the field: what makes a claim on a ledger legally enforceable, and in whose forum. That question does not appear anywhere on the agenda. Which is, in its way, the most honest thing about the agenda.
The Rail Is Real. The Revolution Is Not.
Let me give the strongest version of the bull case, because it deserves one.
The correspondent banking model is genuinely awful in ways that are easy to forget if you have never had to move money through it. A cross-border payment in the traditional stack hops through a chain of intermediary banks, each of which takes a cut, each of which applies its own cut-off times and its own compliance screening, each of which introduces delay and opacity. Settlement lands at T+2 in many corridors, longer in the exotic ones. The sender cannot see where the money is. The recipient cannot predict when it arrives. The fees are layered in a way that makes them nearly impossible to itemize, which is precisely why they survive.
Collapsing that multi-hop chain into a single on-chain settlement leg is a real improvement. It removes intermediaries, compresses time, and produces an auditable trail. When the agenda describes stablecoins as a high-speed settlement layer running beneath the existing banking system, that is an accurate description of a genuine cost-structure optimization. I do not dispute it.
But notice what kind of improvement it is. It is a substitution of one settlement mechanism for a cheaper one, not a re-founding of what settlement means. And the truly hard parts of a cross-border payment do not disappear when you move the settlement leg on-chain. They migrate.
Foreign exchange conversion does not happen on the chain. It happens at the boundary, priced by a market maker, and that market maker carries inventory risk and quotes a spread. Sanctions and anti-money-laundering screening do not happen on the chain; they happen before the funds enter and after they exit, and they are the single largest source of friction and delay in the entire flow. Dispute resolution does not happen on the chain. Chargebacks, refunds, and reversals do not happen on the chain. Liquidity provisioning at the edges โ making sure there is local currency available to pay out at the destination โ does not happen on the chain.
During the DeFi Summer of 2020, I spent three months mapping roughly five hundred million dollars of liquidity movement across Uniswap and Aave, correlating it against Federal Reserve balance sheet expansion. The lesson I took from that exercise, and the one I have repeated in every piece of analysis since, is that the interesting part of any payment system is never the settlement leg. It is the liquidity at the boundary. That is where the cost lives, where the failure modes live, and where the margin is extracted. On-chain settlement improves the boring middle of the flow and leaves the expensive edges untouched.
So when the industry says it has built an invisible rail, I believe it. I also notice what invisibility costs.
An Article About Stablecoin Production That Never Names a Stablecoin
Here is the observation that reframed this entire agenda for me.
Across six sessions, dozens of speakers, and every quoted statistic, the source material does not name a single stablecoin. Not USDC. Not USDT. Not PYUSD. Not one. There is a passing reference to Monad Foundation as an attendee, and no discussion whatsoever of which chain the settlement actually runs on, which bridge or messaging layer moves value between environments, or how reserve attestation is performed.
A discussion of stablecoin production that never names a stablecoin is like a discussion of aviation safety that never names an aircraft. It is not a stylistic omission. It is the omission of the variable that dominates the risk.
Consider what actually determines whether an institution can build on a given stablecoin. It is not the chain's throughput. It is the issuer's reserve composition and the quality of the attestation covering it. Tether holds roughly seventy percent of the market by circulating supply, and in the entire history of that asset there has never been a full, independent, audit-opinion-grade examination of the reserves โ only attestations, which are a different instrument with a different scope and a different standard of care. The industry has collectively agreed to be polite about this. I have watched that politeness hold through two full market cycles and one catastrophic bear market, and I have come to regard it as the single largest unspoken liability in the sector. If you are a bank integrating a settlement rail, the reserve question is not a Twitter argument. It is your counterparty risk, and it belongs in the first paragraph of your risk memo.
Then there is the control surface. Major regulated stablecoins maintain blacklist and freeze functions. Tokens can be frozen at the issuer's discretion, typically in response to law enforcement or sanctions requests. This is, from a compliance standpoint, a feature โ it is precisely what makes the asset palatable to a regulated institution. But it is also a structural fact that the promotional framing never confronts: the settlement layer has an administrator, and the administrator has a kill switch, and the kill switch is operated by a private company responding to state requests. That is a perfectly reasonable design for a regulated payment instrument. It is not the design the word stablecoin evokes in most readers' minds, and the gap between those two things is where a great deal of misunderstanding lives.
I learned this distinction the hard way in the winter of 2022, when I ran twelve webinars on trust and verification for my former university's blockchain club as the market fell eighty percent. Three hundred people showed up across the series, and the question that came up most often was not about price. It was about custody: who actually holds this, and what happens if they stop. I built the curriculum around that question because it was the one that determined whether people panicked or thought clearly. The same question, transposed up a register, is the question this agenda refuses to ask. Who holds the reserves, who audits them, who can freeze the token, and in whose forum is a frozen token contested.
The Chain in the Room
Monad Foundation's presence on the agenda is the most technically interesting detail in the entire program, and it received no analysis at all. It was listed as a participant and left there.
Monad is a parallel-EVM Layer 1, built for high throughput. If bank-grade stablecoin settlement genuinely lands on infrastructure of that class, then three properties of that infrastructure become load-bearing for the credibility of the settlement layer: the size and distribution of the validator set, the degree of genuine decentralization in block production, and the censorship-resistance properties under adversarial pressure. Those are not abstract virtues. They determine whether a settlement finality claim survives contact with a regulator who asks what happens if a majority of block producers coordinate.
And here is the tension the agenda cannot resolve: institutional settlement has approximately zero demand for decentralization and very high demand for stability, predictability, uptime, and legal certainty. Those are not the same value proposition, and in some configurations they are in direct opposition. A chain optimized for institutional settlement is a chain whose operators are identifiable, permissioned in practice if not in name, and accountable to someone. That is a perfectly good design. It is also not what the on-chain settlement narrative sold for the last eight years.
There is a broader point here that I have made repeatedly and will make again. The omnichain narrative โ the idea that an application's value scales with the number of chains its contracts are deployed to โ is a manufactured metric. It exists because it is legible to venture investors and illegible to users. No merchant cares how many chains your contracts live on. No treasurer evaluating a settlement rail cares. What they care about is which chain their counterparty is on, because that determines whether the transfer clears. Chain choice in institutional settlement is a procurement decision, not a values statement. It will be made by a committee, documented in a vendor risk assessment, and revisited on a three-year cycle.
Which means the chain is a vendor. And vendors get replaced.
The Float Is the Business
Now the arithmetic that the promotional framing most needs you to skip.
Ask the only question that matters in payments: who gets paid, and for what.
The stablecoin issuer gets paid through float income. Reserves are held in short-dated government obligations and repo, and the yield on those instruments flows to the issuer. Revenue is, to a first approximation, reserves multiplied by the short rate. Volume does not enter the formula. This is not a payment business. It is a duration business wearing a payment business's clothes, and the distinction becomes acute the moment the rate path turns.
Run the scenario. Suppose the Federal Reserve cuts three hundred basis points across a cycle. Suppose simultaneously that stablecoin payment volume doubles, exactly as the agenda's growth claims suggest it might. The issuer's revenue still falls, because the doubling applies to a revenue line that is not rate-sensitive and the cut applies to the line that is. Growth in volume does not hedge a decline in rates. It is not a hedge at all; it is a different variable. Very few businesses have this property โ where the headline success metric and the actual revenue driver are structurally decoupled โ and almost none of them are described to the public in terms of the headline metric.
This is where the GENIUS Act's likely treatment of holder interest becomes commercially decisive. If the framework prohibits issuers from paying yield to holders โ as the prevailing reading of the legislative language suggests, subject to the implementing rules โ then the float is statutorily reserved to the issuer. That is a magnificent moat. It is also an enormous political exposure. The moment a stablecoin issuer's reserve book resembles a money market fund in size and function, the argument that it should be regulated like one, and share its yield like one, becomes very hard to resist. An entire economic model would need restructuring on the strength of a rulemaking.
Now trace the value through the rest of the chain, because this is where the crypto-native reader's intuitions tend to fail.
The card networks are the clearest winners and they know it. Visa and Mastercard treat stablecoins as an additional settlement corridor inside an unchanged revenue model. They earn network fees, foreign exchange spreads, and value-added services regardless of which ledger the money moves on. The strategic beauty of their position is that they are indifferent to the outcome of the settlement-layer war. Whichever stablecoin wins, whichever chain wins, the network toll still gets collected. That is a structurally superior position to owning the settlement layer, and it is why the networks have been discussing this on earnings calls for something like eight consecutive quarters without any of it appearing as a risk factor.
Mastercard's acquisition of BVNK is the most revealing data point in the entire agenda, and it is revealing in the opposite direction from the way it is usually read. A company that acquires a capability rather than building it is telling you something about its internal technical depth and its time horizon. Mastercard is not a slow-moving bureaucracy by payments standards โ it is one of the more competent acquirers in the sector โ and it still concluded that buying was faster than building. That is a statement about how hard the integration actually is.
The infrastructure vendors sit in the middle and they are already consolidating. Fireblocks, BVNK, Stablecoin Blueprint, Velocity โ the independent layer is being absorbed, and the acquisition price is the market's estimate of whether an independent future existed. I would read that estimate as roughly correct.
The banks are the most interesting and the most conflicted party in the room. Stablecoins are, functionally, a substitute for demand deposits in the specific use case of cross-border corporate treasury. Every dollar that moves through a tokenized settlement rail instead of a correspondent account is a dollar that is not sitting on a bank's balance sheet earning spread. Banks are participating in these sessions because the alternative to participating is being disintermediated without a seat at the table, which is a defensive posture dressed as strategic initiative. That is not cynicism; it is what incumbents rationally do. But it does mean that bank enthusiasm on a panel is not the same signal as bank capital allocation in a budget.
And then there is the category that matters most to the people reading this, and it is the category where the numbers get uncomfortable.
The Layer 1 tokens capture gas. Stablecoin transfers are cheap โ fractions of a cent on modern Layer 2 environments, low single-digit cents on a base layer at moderate load. Take the four hundred billion dollar figure at face value, annualize it to something near six hundred billion, and apply a generous blended fee assumption. You land, at the very top of the plausible range, in the low tens of millions of dollars of aggregate fee revenue. Much of that accrues to Layer 2 sequencers and to the bridging infrastructure rather than to a base-layer token. Against a market capitalization measured in the hundreds of billions, the transmission from stablecoin payment volume to Layer 1 token price is not small. It is indistinguishable from noise.
This is the arithmetic that the stablecoin narrative requires you not to perform. The growth of stablecoin payments and the price of the tokens people buy to express that thesis are connected by a coefficient close to zero. I have watched this mismatch persist for three cycles now, and I have come to believe it is the single most expensive category error in crypto investing.
Production Is a Word Without a Definition
Let me be precise about what commercial production means in payments, because the phrase is doing a lot of unexamined work.
When a payment company says a product is in general availability, it typically means: the product is available to a defined set of customers, in a defined set of corridors, with a published fee schedule, under a contract, with a support path. It does not mean global, unlimited, or fully featured. The industry's definition of general availability is generous to the point of elasticity, and this is not a criticism โ it is a mature industry's reasonable convention. You ship the corridor that works, you learn, you ship the next one.
But that convention has a consequence for how we should read the claim that a given network is in commercial production for stablecoin settlement. It means a specific corridor, with specific counterparties, at specific limits, is running. That is a genuine achievement and it is not nothing. It is also not the thing the promotional framing wants it to be, which is evidence that the general case has been solved.
I ran into exactly this problem in 2024, when my team spent three months quantifying the first fifteen billion dollars of institutional inflow following the spot Bitcoin ETF approvals. We published a whitepaper that argued, among other things, that the disclosure granularity of the vehicles determined what could legitimately be claimed about the market's structure. The lesson generalizes with uncomfortable precision. When the reporting is granular โ line items, dated flows, named counterparties โ you can make strong claims. When the reporting is a round number without a denominator, you can make any claim you like, which means you can make none.
So here is what would actually constitute evidence that the pilot phase has ended. I am listing these not as a rhetorical device but as an operational test, because I would genuinely revise my view if these existed.
An audit opinion on the reserves, from a firm whose name carries reputational weight, with a scope that covers composition and custody, published on a schedule. A published fee schedule for institutional stablecoin settlement, with the corridor-level spread disclosed, so that the economics can be compared against correspondent banking. A segment revenue disclosure from a card network showing stablecoin settlement contribution, rather than a qualitative mention in an earnings call. A named corridor with a named transaction limit and a named set of participating institutions, subject to third-party verification. An implemented rule under the GENIUS Act framework, with a compliance date inside the next four quarters rather than in 2027.
Every one of those items is boring. That is the point. The end of a pilot phase is never announced in a panel title; it is documented in a footnote.
The Four Hundred Billion and the Missing Denominator
Let me do the comparison that the promotional framing omits.
Four hundred billion dollars over roughly eight months annualizes to approximately six hundred billion. Set that against the reference points. Global cross-border payment flows, on the Bank for International Settlements' broad measure, run in the many hundreds of trillions annually. Global remittance flows, on the World Bank's measure, run somewhere in the seven to nine hundred billion range. Both of those reference figures need to be checked against the most recent official releases rather than taken from memory, and I want to be explicit that I am working from training-era knowledge here and the confidence on the exact comparison figures is moderate.
But the order of magnitude holds, and it holds robustly. On the most generous reading of the four hundred billion figure, stablecoin payment volume is on the order of a single-digit fraction of one percent of global cross-border flow, and roughly comparable in size to the global remittance market. That is the correct frame.
It supports neither the triumphant reading nor the dismissive one. It does not support disruption, because a one-percent share is not disruption. It does not support irrelevance, because remittance-scale is not nothing โ it is a real, functioning market with real users and real fee compression happening inside it. What it supports is a third reading, the only one that survives contact with the numbers: early, small, and growing at a rate that is genuinely fast but arithmetically easy to achieve from a base this small.
A seven-fold growth rate from a small base is not a signal of maturity. It is a signal of a small base. If you have moved two billion dollars and you move fourteen billion the next year, you have grown seven-fold and you are still, in the context of the global payment system, a rounding error. I have seen this exact statistical illusion used to sell three separate narratives across three cycles, and it works every time because the multiple is large and the denominator is invisible.
And here is the detail that tells you the industry knows it. One of the sessions was built around the one-and-a-half-trillion-dollar figure, and its title is a question. The framing interrogates the number rather than celebrating it. When your own conference panel is asking what your headline statistic actually means, you have already conceded the point.
The Contrarian Read: Decoupling, and the Two Things Nobody Scheduled
The consensus interpretation of this agenda is that stablecoins have arrived, that the institutional adoption curve is now vertical, and that the correct positioning is to own exposure to the theme. I want to argue the opposite in two specific ways, because the general form of the argument is too easy and therefore not useful.
First contrarian claim: stablecoin adoption is decoupling from crypto asset prices, and the coupling that does remain is a narrative coupling, not a cash-flow coupling.
Follow the money through the value chain I laid out earlier. The float accrues to issuers. The network fees accrue to Visa and Mastercard shareholders. The custody and movement fees accrue to Fireblocks and its peers. The deposit substitution cost lands on banks. The gas revenue โ the only piece that touches a token โ rounds to noise. Nowhere in that flow does a dollar reach a crypto asset holder except through the indirect channel of stablecoins functioning as DeFi collateral, which is a real but second-order effect and which does not scale linearly with payment volume.
The corollary is uncomfortable. Using the stablecoin adoption narrative to take a position in Layer 1 tokens is a category error, and it is the most common category error in the market today. The correct expression of a stablecoin-adoption view, if you are constrained to public equities, is the card networks and the custodians, and both have been pricing this in for roughly two years. The correct expression in crypto is the DeFi lending protocols where stablecoins serve as base collateral. The incorrect expression is almost everything else that gets bought.
Second contrarian claim: the GENIUS Act framework may be a survival threat to the decentralized stablecoin sector, not a rising tide.
This is the reading that the promotional framing cannot accommodate, because it treats regulation purely as an enabler. But read the requirements as a compliance officer would: one-to-one backing in high-quality liquid assets, monthly disclosure, independent examination, redemption timing requirements, restrictions on cross-border recognition. Each of those is a fixed cost. Fixed costs do not scale with the size of the issuer; they scale with the existence of the issuer. A regime built to license a small number of well-capitalized institutions with audited books and legal departments is a regime that eliminates the long tail by construction.
The likely equilibrium is an oligopoly of a handful of licensed issuers, most of them either bank-affiliated or already systemically important. That is a good outcome for payments. It is a bad outcome for anyone whose thesis depended on permissionless issuance, and it is an existential outcome for collateralized crypto-native stablecoins that cannot satisfy the reserve composition and disclosure requirements without abandoning the mechanism that makes them decentralized in the first place.
And now the silence. I have spent this entire piece listening to the silence between market cycles, and in this particular room the silence has a very specific shape: it is the shape of the names that were not on the agenda.
Stripe, with Bridge and its own settlement chain, is not in the source material at all. Circle, with USDC and its Arc settlement layer and the cleanest compliance posture in the sector, is not mentioned. JPMorgan's deposit token infrastructure is absent. Those are not minor omissions. Those are three of the most consequential participants in the exact market the agenda claims to survey, and their absence means the competitive analysis embedded in the framing is not a survey at all. It is a guest list. A guest list is a legitimate thing for a conference to publish. It is not a map of the industry, and it should not be read as one.
I have one more silence to name. Western Union appearing as a participant in a stablecoin settlement session is not a growth signal. It is a distress signal. Remittance is the single most exposed incumbent business in the entire payment stack โ gross margins in the mid single digits to low double digits on the fee line, in a market where an on-chain transfer compresses the marginal cost of the settlement leg toward zero. When the incumbent shows up to the panel, it is because the incumbent has run the numbers and does not like the answer. That is the most honest data point in the entire program, and it arrived without a slide.
What I Am Actually Watching
The next eighteen months of this story will not be decided in a keynote. They will be decided in documents that nobody puts on a stage.
The first is the implementing rulemaking under the GENIUS Act framework. The statute's January 2027 effective date is a date on a calendar; the compliance reality is set by the rules, and rules lag statutes. I will be reading the proposed rule text for three things specifically: the definition of permissible reserve assets, the treatment of holder yield, and the cross-border recognition provisions. Those three clauses determine the industry's structure more than any product announcement will.
The second is segment disclosure. The moment a card network breaks out stablecoin-linked settlement revenue as a reportable line rather than a qualitative sentence, the market will finally have a denominator. I would expect that to happen before any of the crypto-native players produce an audited reserve opinion, because public companies have disclosure obligations and private ones have communications strategies.
The third is the corridor-level economics. The first published fee schedule for institutional stablecoin settlement, benchmarked against correspondent banking in the same corridor, will settle the question of whether this is a cost revolution or a cost optimization. My prior, based on where the costs actually sit in the flow, is that the savings will be real and smaller than advertised, concentrated in the settlement leg and absent at the boundary where FX and screening live.
As for cycle positioning: this reads to me as mid-cycle infrastructure consolidation, not a top signal. The tell is that the conversation has become boring. Boring is what maturity looks like from the inside. The ICO summer of 2017 was never boring, and almost none of it survived. The DeFi summer of 2020 was never boring, and a meaningful fraction of it survived. The bear market of 2022 was unbearable, and the custody discipline it forced on everyone is the reason institutional capital showed up in 2024 at all. Narratives that survive their own boredom tend to be the ones with revenue underneath.
Which brings me back to the agenda, and to the one thing I keep returning to. If the pilot phase really had ended, the industry would not need to book six rooms in two days to explain it. It would be in the footnotes of an annual report, three lines long, and nobody would write an article about it.
That is the silence I am listening to. Not the absence of noise โ the presence of a specific, deliberate quiet where the audit opinion, the fee schedule, and the corridor disclosure should be. When those arrive, they will arrive without a panel. And that will be the moment the production era actually begins.