Stablecoins Were Just a Prototype: Machine-First Dollars and the Coming Asset Token Explosion

Ninety percent of the money moving through stablecoins right now is not moved by people. It is moved by software. Bots arbitraging price differences across exchanges. Smart contracts rebalancing liquidity pools every few seconds. Automated market makers routing value across blockchains without any human touching a button. When Visa published those numbers in April 2024, most commentators treated them as a gotcha: stablecoins are not real payments, they said. Just machines talking to machines.

Ribbit Capital read the same report and arrived at the opposite conclusion. The bots are not the flaw in the stablecoin story. They are the story. And they are just getting started.

This is Day 4 of our seven-part series analyzing Ribbit Capital's Token Revolution letter. Day 1 mapped the $41 trillion thesis. Day 2 explored KYA, the identity layer everything else depends on. Day 3 followed the race for memory tokens, the most intimate data asset ever created. Today we trace the money.

A Prediction That Was Too Small

In 2018, Ribbit Capital wrote to its investors about a fledgling idea that had caught the firm's attention. The concept was simple: a cryptocurrency pegged to the US dollar, offering stability in a market defined by chaos. The firm was cautiously optimistic.

The firm's dreams, by their own admission, turned out to be too small. The numbers in the stat grid above tell that story: a hundredfold increase in AUM, a position alongside sovereign nations as a Treasury holder, and monthly settlement volumes competing with the largest payment networks on earth. Global regulators are now working on frameworks to bring stablecoins formally into the international financial system.

Those numbers are impressive on their own. But they obscure what Ribbit considers the far more important development: stablecoins accidentally invented a new category of financial instrument, one built not for people but for machines.

The Machine-First Dollar

Traditional dollars exist as entries in bank databases. Moving them requires intermediaries: correspondent banks, clearinghouses, payment processors, each adding latency, cost, and permission gates. A wire transfer from New York to Tokyo involves multiple institutions, takes hours or days, and costs real money. The dollar itself is not programmable. It cannot carry instructions about what it should do when it arrives, or enforce conditions on how it gets spent.

Stablecoins changed this at a fundamental level. They are dollars that software can natively hold in a wallet, send to any address on earth, and transact in any amount, all without asking permission from an intermediary. The difference sounds incremental. It is not.

When you make a dollar programmable, you discover that the dollar was never the constraint. The constraint was the infrastructure around it: the banks, the rails, the compliance layers, the business hours, the jurisdictional boundaries. Remove those constraints and usage does not just increase. It changes in kind.

This is what the bot volume reveals. Those automated systems are not wasting resources. They are doing things that were previously impossible: settling trades in milliseconds, rebalancing portfolios across multiple blockchains simultaneously, providing liquidity to markets that would otherwise have none, and managing risk at a speed and granularity that no human treasury department could match. These are the early behaviors of an economy where software is a first-class financial actor.

Why the Critics Have It Backwards

The standard critique goes like this: comparing stablecoin volumes to Visa or Mastercard is apples to oranges. Most of the volume is just bots recycling capital, not real economic activity. Ribbit's response is direct: yes, the comparison is apples to oranges. That is precisely the point.

Card networks process consumer purchases. Stablecoins settle a fundamentally different kind of economic activity. Crypto arbitrage, liquidity provision, automated market making, cross-chain settlement: these are not inferior versions of buying coffee with a credit card. They are new categories of financial activity that only exist because the money itself became programmable.

And the human-initiated slice is not small, either. Even a fraction of the monthly settlement volume is profoundly large for a young technology running on immature infrastructure. Ribbit sees rising real-world stablecoin use cases across its own portfolio: vendor payments through Coinbase, cross-border remittances through Bridge and Sling, foreign aid disbursement through Fireblocks. The technology is young. The regulatory infrastructure is incomplete. The on-ramps are clunky. And it is still settling more value than most traditional payment networks.

The question Ribbit wants you to sit with is simple. If bots have accomplished all of this with just one programmable asset class (the dollar), what happens when you give them hundreds?

The Profit Signal

Before looking forward, Ribbit lingers on what stablecoins have already proved about economics.

Those numbers describe a company operating at a profitability-per-head that makes Wall Street's most efficient firms look bloated by comparison. Ribbit acknowledges stablecoin issuers may have hit peak margins (before rate compression from regulation or competition forces them down). But that caveat comes with a sharp counterpoint: Tether might also be the first demonstration of how an early lead on open, global networks compounds into durable advantages in liquidity, compliance, brand, and infrastructure.

The letter compares this to a pattern that has repeated throughout financial history. Being entrusted to manage funds, in whatever form they take, has historically been a sticky business with many adjacencies. Larry Fink has been talking about tokenization for a reason. When the CEO of the world's largest asset manager spends his public appearances discussing the same thesis a $12 billion VC fund is writing 41 pages about, it is worth paying attention to the convergence.

But the economic case for stablecoins is, in Ribbit's framing, just the appetizer. The simplicity of pegging a token to a dollar made stablecoins the ideal prototype for a much more complex project: tokenizing everything else.

From Dollars to Everything

A dollar, in theory, is a dollar. One unit. One value. Simple to represent, simple to verify, simple to settle. That simplicity made stablecoins the ideal first test case for programmable money. But as Ribbit pivots from prototype to production, they identify a challenge that grows exponentially with complexity.

This insight separates Ribbit's analysis from the generic "tokenize everything" argument you hear at every crypto conference. Most tokenization enthusiasts focus on the entitlement: the right to own a fraction of a building, or a share of a fund. Ribbit focuses on the information problem. A piece of real estate comes with zoning data, tax history, liens, maintenance records, insurance policies, tenant agreements, environmental assessments, market comparables. The token itself is the easy part. Making all of that contextual data machine-readable, verifiable, and composable is the hard part. And the hard part is where the money is.

This creates a valuable role for companies that facilitate asset compliance, accounting, valuation, reporting, and governance. Ribbit calls this "digital fund administration" and points to two of its portfolio-adjacent companies as positioned for this future: Juniper Square and Kanastra. Both are building what the letter calls "token factories" for high-value assets, systems that reinvent the efficiency with which you can create, administer, and trade tokenized instruments.

The opportunity is not just administrative efficiency. These platforms sit on a trove of valuable tokens themselves: identities, holdings, transactions, performance data. As the token factory thesis we explored earlier predicts, the companies that process the most tokens end up best positioned to create new markets, data products, and liquidity tools on top of them.

The Procurement Agent Scenario

The letter's most vivid illustration of what comes after stablecoins is a B2B scenario that reads like science fiction today and will read like a case study within five years.

Every step in this scenario depends on the capabilities that stablecoins proved were possible. Programmable execution. Permissionless settlement. Machine-readable terms. Composability across systems. The procurement agent does not need to call a bank. The working capital agent does not need to negotiate with a factor. The treasury does not need to reconcile across spreadsheets. The tokens carry their own context.

This is not a hand-wave about the future. The components already exist in isolation. Tokenized contracts run on Ethereum and Solana today. Automated treasury management operates across DeFi ecosystems. Agent wallets shipped from Coinbase in late 2025. What does not exist yet is the full stack operating together at enterprise scale. That is where the investment opportunity lives.

Programmable Money Gets Programmable Permissions

The scenario above depends on tokens that do more than represent a value. They need to encode behavior: what the token can be used for, who can hold it, how it responds to market conditions. This is where evolving blockchain standards enter the picture, and where Ribbit sees the gap between stablecoins-as-prototype and asset-tokens-as-production widening into a massive engineering opportunity.

Solana's Token Extensions and Ethereum's ERC-6900 and ERC-6551 standards allow developers to build permissions directly into a token. A token can enforce that only accredited investors hold it, that it cannot be transferred outside certain jurisdictions, or that it automatically distributes dividends on a schedule. These are not theoretical capabilities on a whiteboard. They are in production code on mainnet today.

Ribbit sees a future where tokens grow more dynamic: initiating transactions, responding to offers, updating their own metadata under certain conditions, or even negotiating with other tokens. A compute token might route itself to the highest-yield vault when idle. A real estate token might trigger a governance vote when ownership passes a concentration threshold. A licensing token for creative work might adjust its pricing based on usage volume.

This is the bridge between the stablecoin prototype and the fully tokenized economy. Stablecoins proved that value could live on a blockchain. Programmable asset tokens will prove that behavior, compliance, and intelligence can live there too.

The Agent Perspective

How will agents decide which tokens to use? Ribbit's answer is that agents will be relentless optimizers, choosing whatever instruments give them the most capability per transaction. And blockchains, despite a decade of bad UX press, may have the edge.

The argument inverts the entire history of blockchain UX discourse. For a decade, the crypto industry has been trying to make blockchains more user-friendly. Private key management is confusing. Transaction signing is opaque. Network selection is bewildering. All of these are real problems for humans. None of them are problems for software.

An agent will not forget its seed phrase. It will not confuse network addresses. It will not misconstrue what it is signing. But it will appreciate, deeply, the composability and accessibility of open networks like Ethereum, Solana, and Bitcoin. It will value the ability to interact with any protocol, any token, any contract, without needing an API key or a business relationship with a centralized provider.

Ribbit went a step further and asked the agents themselves:

But Ribbit is careful not to be dogmatic about decentralization. Agents will be "promiscuous" with payment and asset tokens, accessing and managing value wherever it exists. Centralized companies that serve key AI leaders and keep building for agents will capture significant value. Stripe, which acquired Bridge specifically for stablecoin infrastructure, is an obvious example. PayOS, building payment rails with agents as the primary customer, is another. Even governments are competing: Brazil's Central Bank is building DREX, a programmable asset layer designed to tokenize claims, entitlements, and credentials on top of its real-time payments network (PIX).

The winner will not be blockchains or banks or governments. It will be whoever makes their infrastructure most useful to the agents. And agents, unlike humans, have no brand loyalty, no inertia, no nostalgia for the payment method they grew up with. They will route through whatever system is fastest, cheapest, and most composable at the moment of transaction.

The Infrastructure Gap

Ribbit is explicit about how much work remains before the tokenized asset economy can operate at the scale it envisions. The letter identifies three specific infrastructure gaps in stablecoins alone, before you even begin to tokenize more complex assets.

The Stripe-Bridge acquisition is the clearest signal that the infrastructure gap is real and investable. Stripe, the most valuable private fintech company on earth, paid over a billion dollars for Bridge because it recognized that handling stablecoin demand (onboarding, servicing, and orchestrating stablecoin payments safely and compliantly) is genuinely hard. Ribbit predicts we will see Bridge analogs emerge in other markets and asset types, each solving the same class of infrastructure problems for a different geography or asset category.

This is where the opportunity map gets granular. It is not enough to say "tokenize everything." Someone needs to build the credible Brazilian real stablecoin. Someone needs to build the compliant on-ramp for Nigerian businesses. Someone needs to build the treasury management system that lets a CFO sleep at night while agents optimize across DeFi, CeFi, and TradFi simultaneously.

The DeFi Tailwind

As more assets are tokenized, one of the clearest beneficiaries will be the decentralized finance protocols that already know how to handle programmable tokens. Swapping, lending, borrowing, insuring, providing liquidity: these are all primitives that DeFi protocols built for crypto assets and that generalize naturally to any asset token.

Ribbit names specific companies it expects to benefit from the coming wave: trusted consumer brands like Coinbase, Revolut, and Robinhood that can serve as the front door for mainstream users and agents. Infrastructure protocols like Morpho (lending), Jito (MEV and staking on Solana), and Uniswap (decentralized exchange) that provide the plumbing for token-native financial operations. Ribbit frames this as a durable thesis: people and agents will always need safe ways to navigate the world of programmable money, and the companies that earn trust in this role will compound their advantage.

The convergence between Ribbit's portfolio companies makes the thesis self-reinforcing. Coinbase issues stablecoins, operates agent wallets, and offers crypto-backed lending through Morpho. Stripe (which acquired Bridge) processes payments and now handles stablecoin infrastructure. Robinhood launched a blockchain L2 and integrated prediction markets via Kalshi. Each company is building pieces of the same tokenized economy, and Ribbit holds positions across all of them.

The Biggest Prize: Markets That Don't Exist Yet

The final section saves its most speculative (and potentially most important) argument for last. Ribbit believes the biggest opportunity in the tokenized asset economy will not come from digitizing existing assets more efficiently. It will come from creating entirely new token markets that have no analog in the current financial system.

Consider GPU compute as a tradeable asset. Right now, if a company wants to lock in GPU capacity for future workloads, it signs long-term contracts with cloud providers, negotiates enterprise agreements, and manages capacity planning through procurement departments and spreadsheets. In a tokenized world, a compute contract becomes a token. That token can be traded on a secondary market. It can be fractionalized. It can be used as collateral. It can be automatically deployed to different workloads by agents monitoring demand in real time.

Or consider creative IP. Today, a musician who wants to license a song for commercial use enters a months-long negotiation involving multiple intermediaries: publishers, performing rights organizations, sync licensing agencies. In a tokenized world, the song exists as a token with embedded licensing terms. An agent representing an advertiser queries the token's metadata, confirms the terms fit the campaign, and executes the license in seconds. The creator's agent monitors usage, collects royalties, and adjusts pricing based on demand, all without either human picking up a phone.

Ribbit frames the directional insight cleanly: "One conclusion of this line of thinking is to find ways to be long digital IP." If every piece of unique intellectual property gains native tools for monetization, licensing, and enforcement through tokenization, then the value of owning original creative work increases structurally. The tools for capturing that value get better at the same time the demand for unique content (driven by agents that need training data and licensed material) goes up.

What Comes After the Prototype

The stablecoin experiment answered a question that the financial system had not thought to ask: what happens when money is rebuilt from the ground up for software? The answer, captured in the numbers at the top of this article, is a market that now rivals mid-sized countries in Treasury holdings, an issuer whose profit-per-head dwarfs the largest banks in history, and an ecosystem of bots conducting financial operations at a scale and speed that no human institution can match.

Ribbit's argument is that this was merely the test run. The dollar was the easiest asset to tokenize: one value, one peg, no complex metadata. Everything that follows will be harder and, if the pattern holds, exponentially more valuable. Real estate. Fund shares. Supply chain contracts. Compute capacity. Creative works. Insurance policies. Carbon credits. Every one of these carries more complexity than a dollar, and every one of them can be made more composable, more liquid, and more accessible to autonomous agents through tokenization.

The infrastructure to support this transition is immature. Most of the world's currencies lack credible stablecoins. Bank connectivity is inadequate. Treasury management tools were built for humans, not agents. Regulatory frameworks are incomplete. But these are gaps, not walls. And every gap is a buildable opening for the companies (and investors) positioned to fill it.

Tomorrow, in Day 5, we examine Vertical Token Systems: the companies building what Ribbit calls "BPO-dismantling, ERP-commoditizing platforms" that will deploy agent armies across entire industries. The stablecoins gave the agents money. The identity layer gave them credentials. The memory tokens gave them context. The vertical systems will give them jobs.

Day 4 of 7. Start with Day 1: The Token Revolution Overview or go deeper at The Pond.