Money After AI: Meet the New Digital Dollar Built for the Internet "Stablecoins" | EP #200
Summary
Allaire defines a payment stablecoin narrowly: a one-for-one fiat claim, fully reserved and redeemable, running as cryptocurrency on public networks. The payoff is safer base-layer money with “openness, interoperability, global reach, programmability” and marginal transfer costs approaching zero. At recording, Diamandis put USDC at a $76 billion market cap, over 90% year-on-year growth, with Circle’s recent IPO raising $1 billion.
Allaire’s geopolitical call is that the U.S. can defend dollar primacy by exporting open, competitive stablecoin infrastructure that makes dollars more useful and supports demand for short-term Treasuries. Russia’s exclusion from dollar-system utilities freaked people out by showing that database access can be blocked. Diamandis separately raised exponentiating debt and the resulting challenge to the full-faith-and-credit proposition. Yet dollar trade settlement remains “60-some percent,” perhaps as high as 80%, leaving stablecoins as a potential advantage in the “financial utility arms race.”
USDC’s claimed safety case rests on transparent, short-duration sovereign backing rather than an opaque commercial-bank balance sheet. Roughly 90%—sometimes 85% to 93%—sits in the BlackRock-created Circle Reserve Fund, identified as USDXX, primarily holding U.S. Treasuries of 90 days or less, overcollateralized overnight Treasury repo, and cash. The average duration can be just 10 to 14 days, while Bank of New York Mellon, the “bankers’ bank,” custodies fund cash and $44 trillion of assets overall.
The economic fault line is full-reserve payment money versus fractional-reserve credit: banks can “borrow a dollar from you” and lend it out 12 times, while Circle’s payment-money model does not lend a dollar out 12 times. Allaire’s post-financial-crisis conviction is that payment money and lending money should be separated because free-floating internet IOUs would be “a recipe for total disaster.” Under the GENIUS Act, a commercial bank cannot directly issue a stablecoin, although its holding company can create a dedicated subsidiary.
Allaire argues regulated stablecoins augment central banks rather than replace them because Circle neither creates money nor sets interest rates. His counterexample is China’s e-CNY: despite government distribution mandates, “no one used it” because Alipay and WeChat Pay offered more utility. Europe’s estimated CBDC launch was 2029 and might slip, while the U.S. bet was private-sector, open-internet innovation; the Trump administration essentially banned a U.S. CBDC.
Allaire’s five-year forecast is that “the vast majority of stablecoin transactions” will be AI-intermediated. Globally distributed agents with capital need interoperable money, proofs and programmable controls that card networks cannot easily supply. x402-style rails can settle either a five-cent AI-token purchase or a billion-dollar oil transaction—the same way SMTP carries radically different payloads without caring what they contain.
The larger upside is an on-chain corporate form combining token capital, stablecoin treasury, provable governance, AI workers and human contractors. Allaire’s specimen is Hyperliquid, a perpetual-derivatives protocol reportedly operated by 11 people and producing well over $1 billion in revenue, with revenue returned to token holders and stakeholders. He expects “super predator corporations,” while hedging the timing and stressing that courts, asset enforcement and “prisons for the humans that do bad things” remain necessary.
Near-term monetization is arriving through digital-asset settlement, cross-border payroll and B2B flows, dollar savings, and corporate treasury before everyday checkout. Shopify was rolling USDC out to sellers with a 50-basis-point merchant incentive, and Stripe had made it available out of the box, but Allaire said e-commerce usage remained “very, very small” and widespread retail adoption was still a couple of years away. Emad Mostaque’s “static to supercharged” money also brings inflation and stability risks, making cryptographic auditability and provable agent controls central to the thesis.
Deep dive
1. Payment stablecoins are full-reserve fiat with internet superpowers
Allaire deliberately excludes assets that are merely “stable in name only.” His regulated definition is a fiat-denominated currency—dollar, euro, RMB or yen—represented as cryptocurrency on public blockchain networks and backed one for one by the corresponding fiat assets.
Stability comes from continuous creation and redemption at par, not from an algorithmic promise. Allaire’s claim is that full reserves make this money safer than commercial-bank digital money while preserving the internet’s “openness, interoperability, global reach, programmability” and near-zero marginal movement costs.
U.S. law calls the instrument a “payment stablecoin”: “money good to settle a transaction.” Diamandis’s opening scorecard put USDC at a $76 billion market cap, growing over 90% year on year, after Circle’s IPO raised $1 billion.
2. Dollar primacy now depends on utility as much as sovereign power
Allaire begins with network effects: after World War II, the dollar became embedded in global payment and trade-settlement systems, creating unmatched utility and liquidity. Reserve status therefore rests not only on state power but on how deeply the currency is plugged into financial infrastructure.
The 1971 break with gold changed the basis of confidence. Ballooning 1960s deficits and Vietnam War spending led Nixon to end convertibility, leaving the dollar supported by “full faith and credit,” sovereign creditworthiness and, as the hosts added, American hard power.
Russia’s invasion of Ukraine exposed the system’s political control points. SWIFT is a software messaging utility, while Treasury holdings are ultimately database records; sanctions effectively meant that “your read access has been blocked,” unsettling governments that had treated the dollar network as neutral and dependable.
Exponentiating debt adds another vulnerability, though Allaire sees no credible replacement yet: BRICS alternatives generate noise, while dollar trade settlement remains “60-some percent,” perhaps 80%. His policy answer is to export competitive stablecoin and blockchain infrastructure, strengthening Treasury demand and America’s position in a new “financial utility arms race.”
3. USDC’s reserve stack is built for visibility and instant liquidity
Allaire contrasts USDC’s daily look-through with conventional bank opacity. A depositor cannot inspect every asset supporting a bank’s obligations, and he argues that even regulators could not assemble a real-time picture because auditors generally sample records rather than continuously reconciling them.
About 90% of USDC backing—sometimes 85%, sometimes 93%—sits in the Circle Reserve Fund, created with BlackRock and identified as USDXX. Its holdings are published daily and consist primarily of short-duration U.S. government obligations, with Treasury maturities capped at 90 days.
Average duration can run just 10, 13 or 14 days. The portfolio also uses overcollateralized overnight Treasury repo: large banks borrow Circle’s cash and pledge more value in T-bills, leaving the fund holding the collateral if a counterparty fails to repay.
Fund cash is held at Bank of New York Mellon, which Allaire said custodies $44 trillion in assets. Of the remaining roughly 10%, about 98% sits with “bankers’ banks” such as BNY Mellon and State Street; a small portion is positioned globally to support 24/7 local liquidity. NYDFS restricts Circle to specified permissible assets.
4. Circle separated payment money from lending—and regulated from day one
Allaire’s blunt description of fractional-reserve banking is that a bank “borrows a dollar from you” and may lend it out 12 times; a run happens when depositors collectively ask for money that is no longer sitting there. His preferred architecture separates fully reserved payment money from explicitly chosen credit risk.
The Great Financial Crisis made that philosophical distinction foundational for Circle. Opacity and leveraged instruments convinced him that internet money could not consist of circulating IOUs: “You don’t want these IOUs floating around, free floating, moving around on the internet.”
Circle pursued technology and policy together. Before investing his own first dollar, Allaire hired regulatory advisers, operated state by state under money-transmission law, posted required bonds and held permissible investments. Nearly 12 years before this interview he was already testifying to the Senate, because realizing the technology required active policy work.
5. Private digital money won the utility contest in China
A regulated stablecoin inherits central-bank monetary policy rather than replacing it. Circle cannot set rates, manufacture dollars or replace the central bank’s ultimate settlement ledger; central banks possess what Allaire jokingly called the database “SQL insert statement capability.”
Facebook’s proposed “Zuck bucks” jolted governments in 2019: USDC had only about $500 million circulating, while Facebook had roughly 3 billion users and proposed synthetic money based on multiple currencies. The reaction helped send central-bank digital-currency research across a reported 122 governments.
China then pushed e-CNY into payment apps and banks, yet, in Allaire’s telling, “no one used it.” The problem was utility, not availability: Alipay and WeChat Pay kept innovating features such as voice-authenticated checkout, while a centrally built product could not compel consumer preference.
Allaire estimated that roughly 97% of financial transactions already pass through private intermediaries. The Trump administration essentially banned a U.S. CBDC; Europe continued toward an estimated 2029 digital-euro launch that might slip, despite bank resistance. His conclusion: open-source infrastructure, entrepreneurship and competitive markets are the stronger bet.
6. The prize is $10 trillion of intermediation, but banks face a conflict
Allaire describes the underlying financial system as worth hundreds of trillions of dollars, with roughly $10 trillion in intermediary revenue. Internet restructuring usually unfolds over 10 to 20 years—AI might accelerate it—and even the eventual platform winners can remain small beside the old regime for a long time.
The GENIUS Act prevents a commercial bank itself from issuing a stablecoin because deposits and loans are risky backing. A bank holding company can establish a separate issuer, but then confronts an incentive conflict: lend deposits 12 times for margin, or hold inert reserves while payments commoditize toward zero.
Allaire therefore expects hybridization: banks preserve deposits and lending while converting into stablecoins for internet settlement, programmability and institutional services. Stablecoins still behave like platform utilities, with developer and liquidity flywheels that create meaningful network-effect moats.
Tether’s larger position, in his account, grew from offshore Asian crypto markets and Bitfinex, where exchanges lacked dollar banking. Circle instead chose “U.S. first” and “regulatory first,” enabling relationships with major banks, asset managers and governments. GENIUS is a tailwind, but also invites what could become “a new stablecoin every week.”
7. AI agents turn stablecoins into an economic operating system
Circle began in 2013 with “programmable money” as the motivating idea, before Ethereum existed. Allaire imagined blockchains as “trust machines”—distributed compute engines producing cryptographically verifiable inputs, outputs, data and transactions—then spent roughly five years reaching a usable application layer.
His five-year call is explicit: “The vast majority of stablecoin transactions are going to be AI intermediated.” Agents with capital will hire humans, be hired by humans and transact with agents created anywhere, requiring globally interoperable settlement and proof systems rather than geography-bound card networks.
Layer-one blockchains become “economic operating systems” containing provable data, transactions and compute for entities facing one another without pre-existing trust. AI computation may remain off-chain, but chains provide the coordination and proving ground for identities, inputs, outputs and final economic state.
x402 targets the missing microtransaction layer. Allaire’s SMTP analogy captures the scale independence: the protocol does not care whether the payload is breakfast or a CIA dossier; similarly, stablecoin rails can settle a five-cent AI-token purchase or a billion-dollar commodity trade in fractions of a second.
8. Software-native corporations could become “super predators”
Allaire sees two simultaneous platform shifts: AI foundation models as more capable operating systems, and purpose-built blockchains as economic operating systems.
Salim connected these shifts to roughly 400 years of joint-stock corporations and argued that capital can now relate to machine labor rather than human labor. Together, the panel saw this as an opportunity to revisit corporate organization and governance.
A corporation could be instantiated in software: sell tokens to form capital, hold a stablecoin treasury, automate flows, vote on-chain, selectively disclose records through view keys and audit the books in real time. Both AI assignments and human commercial contracts could “manifest in code.”
Diamandis called this the first fully on-chain corporation, but Allaire preserved the boundary between execution and enforcement. Bad human actors still require courts, lawsuits and asset seizure; “there’s no AI court.” As an NYSE-listed company, Circle itself cannot simply say, “Poof, I’m on chain,” because regulation, legal questions and immature tooling remain.
Hyperliquid is Allaire’s working specimen: an open-source perpetual-derivatives protocol, extensible to markets such as sports or AI compute, reportedly run by 11 people and generating well over $1 billion in revenue. Revenue returned to token holders and stakeholders compounds adoption, suggesting future “super predator corporations”—though he said, “I don’t know what the time frame is.”
9. Hypervelocity requires proof, permissions and refunds above finality
Mostaque’s macro challenge was that U.S. monetary velocity had not recovered since COVID, while programmable money could go “from static to supercharged.” Faster circulation without matching output could affect inflation. Allaire agreed monetary theory may be upended; the smartest central bankers he briefs respond, “Oh my god.”
Allaire also remembers the downside of liberalization: permissive derivatives regimes and opaque balance sheets contributed to the Great Financial Crisis. His answer is not frictionless expansion alone but “radical transparency,” cryptographic proofs, real-time auditability and “agile policy making”—a phrase he acknowledged sounds oxymoronic.
Mostaque asked how Arc’s sub-second finality could coexist with refundability. Allaire clarified that base blockchain settlement is deterministic and irreversible. USDC transactions are final; separately, as a centralized regulated issuer, Circle must freeze sanctioned accounts across 28 blockchains.
Commerce protections sit above final settlement. Receivo carries invoices, receipts and other ISO 20022 metadata; the still-experimental proposed refund protocol would sit above settlement, using an insurance pool and decentralized risk market to handle fraud or customer-choice refunds. Circle’s open-source Secure Tool wraps OpenAI SDKs for agents so wallet APIs can impose permissions and provable monetary controls instead of letting an agent “party on.”
10. Cross-border dollars arrive before the Starbucks moment
USDC’s “bootstrap utility” was digital cash for 24/7/365 trading and working capital in digital-asset markets. It then became capital inside on-chain borrowing, lending and derivatives protocols, where conventional banking hours were an obvious mismatch.
The stronger recent growth, Allaire said, is cross-border settlement: payroll, payouts and B2B payment providers are adding stablecoin rails. Store of value is equally important because users in many countries prefer an internet-native dollar to local currency or local-bank dollars—“over-the-top internet money,” analogous to WhatsApp bypassing SMS fees.
Visa and Mastercard already support USDC debit cards, although purchases still travel over Visa and Apple Pay rails. Shopify was rolling USDC out to all sellers and offering merchants 50 basis points to accept it that quarter; Stripe had made USDC an out-of-the-box payment method.
Everyday e-commerce nevertheless remained “very, very small.” Mainstream wallets, simpler UX and refund protections still need work, putting widespread retail use “a couple years away.” At the opposite extreme, major electronic trading firms already settle multi-hundred-million-dollar transactions in USDC with some frequency.
11. On-chain treasury is the bridge to measurable economic output
For CFOs, Allaire expects on-chain treasury management to become a major category. Startups and fintechs are embedding USDC; Brex had launched it as a feature, while spin-outs from large ERP systems such as SAP were building dedicated treasury solutions.
The mechanism is capital efficiency: move instantly from a tokenized money-market position into stablecoin cash, settle programmatically across geographies, and retain real-time auditability. The missing pieces—enterprise tooling and regulatory treatment allowing an auditor to classify USDC as cash or a cash equivalent—were, in Allaire’s view, arriving with regulatory clarity.
Mostaque’s ten-year picture is “smart” or intelligent money flowing toward the highest-value uses through a self-balancing, self-driving economy. Jurisdictions that obstruct it could lose liquidity to those that permit it; the United States’ regulatory opening therefore becomes a competitive advantage if adequate control functions accompany the velocity.
The closing first principle was “money velocity without debt.” Yet Allaire rejected the idea that Circle’s founding mission was complete: he wants on-chain measurements showing that new economic velocity produces higher global GDP and prosperity. “We’re not there. We’re not even close to there right now.”