ONE TOKEN, MANY MARKETS
Why genuine decentralisation requires monetary portability, not a thousand platform currencies
Keywords: tokenisation; transferability; decentralisation; interoperability; digital money; AI agents; cloud computing; micropayments; network effects; transaction costs; price discovery; settlement; platform economics; switching costs; monetary fragmentation
Thesis statement: A genuinely decentralised digital economy does not require every platform, AI service, cloud provider, data market, storage network, device ecosystem, and online community to issue its own currency. It requires the opposite: a broadly transferable settlement token whose monetary identity survives movement between competing services. The critical property is not that everything runs on one platform, one database, or even one ledger. It is that users and autonomous agents can leave a platform without first leaving the money they received there. A common transferable settlement asset reduces compulsory conversion, monetary switching costs, stranded balances, and the dimensionality of machine treasury management while improving cross-platform price comparison. Prices may still be denominated in dollars, euros, baht, or another common numeraire; the central claim is about portability of settlement, not the compulsory abolition of existing units of account.
The Babel of Digital Money
The digital economy has acquired a peculiar superstition: that decentralisation can be measured by counting tokens.
A new network appears, and it issues a token. A storage platform appears, and it issues another. An artificial-intelligence service wants an “ecosystem”, so it mints a third. A game, a data exchange, a distributed-compute marketplace, a creator platform, a social network, an identity system, an energy market, and an Internet-of-Things project each decide that the apparently indispensable first act of institutional independence is to manufacture a private monetary object.
The result is described as decentralisation. Economically, it can look remarkably like a shopping centre in which every shop requires customers to exchange ordinary money into a different species of gift voucher before approaching the till.
That is not decentralisation. It is monetary fragmentation with better typography.
The distinction matters because a market can be technologically distributed while remaining economically enclosed. A platform may expose open APIs, permit self-custody, publish its source code, and settle transactions across a distributed network, yet still impose a proprietary monetary perimeter if participants must acquire its native currency to enter and sell it to leave. Conversely, many independent firms can compete vigorously while accepting the same settlement asset. Institutional plurality does not require monetary plurality.
The central argument of this essay is therefore stronger than the familiar demand for “blockchain interoperability”. Technical interoperability is useful, but it is not the same thing as monetary portability. A bridge between twenty private currencies does not abolish fragmentation; it automates currency conversion. A wallet capable of holding two hundred tokens does not give its owner one form of money; it gives the owner a compact foreign-exchange desk. A routing protocol that can discover a path from token A to token Q may make conversion easier, but easier conversion is not equivalent to not needing the conversion in the first place.
The decisive property is transferability of the settlement object itself.
Imagine a token received for selling spare GPU capacity. If that same token can then be spent on storage, an API call, electricity, data, a human microtask, or an unrelated merchant service without mandatory redemption into the currency of each next platform, purchasing power has escaped the institutional boundary of the firm in which it was earned. If it cannot, the platform may have decentralised its database while retaining a monetary wall around its market.
This argument does not imply that every digital object should become identical. A kilowatt-hour is not a GPU-second. A gigabyte-month is not a dataset. A dataset is not an equity share. A bond is not a train ticket. A governance vote is not a payment. Tokenisation can represent heterogeneous assets, claims, permissions, credentials, securities, and contractual states. Distinct things should remain distinct when their legal or economic characteristics differ.
The claim concerns the settlement medium: the thing delivered in payment after buyer and seller have agreed what the heterogeneous good or service is worth.
That distinction is frequently obscured in token discourse. The ERC-20 standard illustrates the power of technical standardisation. It defines a common interface that allows fungible tokens to be transferred and reused by wallets and applications (Vogelsteller & Buterin, 2015). Ethereum’s current documentation likewise emphasises that token standards make different tokens easier for applications, wallets, exchanges, and services to handle consistently (Ethereum.org, 2026). This is valuable engineering. Yet the success of a standard interface raises rather than answers the monetary question: if applications can share a technical standard, why must every application also manufacture a separate currency?
The Internet became useful not because every website invented its own incompatible packet. It became useful because heterogeneous applications could communicate across shared protocols. Digital markets require an analogous economic layer: not one corporation, not one application, not one ledger, but a transferable settlement object capable of crossing institutional boundaries without compulsory monetary metamorphosis.
The most important word is therefore not token.
It is transferable.
Transferability Is More Than “Can Be Sent”
A digital object can be technically transferable without being economically portable.
Suppose a cloud platform pays a contractor in CloudCoin. The contractor can send CloudCoin from one address to another. In a narrow technical sense, the token is transferable. But suppose no unrelated merchant accepts it and no competing cloud provider prices services in it. The contractor must therefore sell CloudCoin for another asset before the proceeds can be used elsewhere.
The database entry moved. Purchasing power did not move cleanly across the institutional boundary.
Now change one fact. The contractor receives an asset accepted by an AI inference provider, a storage market, a merchant, a data vendor, and another contractor. The recipient can earn it in one market and spend it in another. The asset has not merely been transmitted. It has preserved its settlement function while changing economic context.
That is the stronger meaning of transferability used here.
It is related to, but not identical with, the monetary concept of singleness. The Bank for International Settlements defines monetary singleness around the ability of different forms of money to settle at par against a common unit of account and stresses the coordination benefits created when money is broadly accepted without users repeatedly evaluating the particular instrument offered to them (Bank for International Settlements [BIS], 2025). Its 2026 Annual Economic Report places interoperability, a common unit of account, singleness, and network effects at the centre of a functioning monetary architecture (BIS, 2026).
The present argument extends the logic into a platform and machine economy. The relevant question is not only whether different issuers’ monetary claims trade at par. It is also whether the settlement asset can leave one service domain and remain useful in another.
That gives a practical exit test:
A user should be able to leave a platform without first leaving the money received on that platform.
If the user must liquidate the platform’s currency before using the proceeds elsewhere, then the platform has imposed a monetary switching step. It may be cheap. It may be automated. It may take milliseconds. But it is still an additional market operation caused by the architecture of the platform rather than by the underlying good the user wanted to buy.
This is why a proliferation of tokens can create the appearance of decentralisation while generating the economics of enclosure. The user is free to choose among many gardens, provided the user changes money at every gate.
A common transferable settlement token changes the locus of competition. The platform must compete primarily on what it sells—compute, storage, inference, bandwidth, data, entertainment, expertise—because the customer’s working capital is not trapped in the platform’s private monetary unit.
Three Things That Must Not Be Confused
The argument becomes much cleaner once three distinct concepts are separated.
First, the unit of account. This is the denomination in which prices, contracts, accounts, and balance sheets are stated. A business can quote a cloud service at US$0.01 even if the final transfer uses another settlement asset.
Second, the settlement asset. This is the object transferred to discharge the payment obligation. A transaction denominated in dollars need not be settled by moving physical dollars; the legal and technical settlement mechanism can differ from the numeraire used to state the price.
Third, the ledger or payment rail. This is the infrastructure that records, validates, communicates, or settles the transfer.
These need not be the same thing.
The BIS makes the distinction particularly clear in discussing contemporary monetary systems: payments can share a sovereign unit of account while using different forms of money, and interoperability between platforms helps preserve the coherence of the system (BIS, 2026). The present proposal does not require every economic actor to abandon national accounting units, nor does it require every transaction to be recorded on one global database.
The core claim is narrower and, for that reason, stronger:
Where digital platforms would otherwise force users to hold and exchange proprietary platform currencies, a common transferable settlement asset can remove an unnecessary monetary boundary.
A universal token could eventually become a unit of account if its purchasing power were sufficiently stable and its acceptance sufficiently broad. But that is not logically required for the portability argument. A service might quote “US$0.004 per inference request” and accept the corresponding amount of the transferable token at execution. Another might quote in euros. A third might quote directly in token units. The settlement asset can be common even while commercial denomination remains plural.
This correction matters because a volatile token cannot magically create stable cost comparison merely by being universal. Common settlement reduces conversion and inventory complexity. A stable numeraire improves price comparison. The two benefits can coincide, but they should not be conflated.
The Arithmetic of Fragmentation
The simplest part of the argument is combinatorial.
Imagine n platforms, each insisting on a distinct settlement currency. If every currency needs a direct market against every other currency, the maximum number of bilateral currency pairs is:
B(n) = n(n − 1) / 2
With 2 currencies, there is 1 pair.
With 5 currencies, there are 10.
With 10 currencies, there are 45.
With 100 currencies, there are 4,950.
Real markets do not, of course, require every conceivable bilateral pair. They route through liquid hub currencies, exchanges, market makers, bridges, or common settlement arrangements. But that is not a rebuttal. It is the point.
If a hub asset is introduced because maintaining thousands of bilateral relationships is inefficient, the architecture has rediscovered the economic advantage of a common bridge. The question then becomes whether that bridge is merely an exchange intermediary between proprietary currencies or whether it can become the settlement asset users hold and spend directly.
Figure 1. Stylised interface burden. If every one of n proprietary currencies requires a direct market against every other, the number of bilateral pairs can rise as n(n − 1)/2. If the same n platforms instead accept a common transferable settlement asset, only n platform-to-settlement acceptance relationships are required in this simplified architecture. This is a conceptual comparison, not an empirical estimate of actual exchange infrastructure.
The Committee on Payments and Market Infrastructures (CPMI) reaches an analogous architectural conclusion in conventional cross-border payments. Its framework on payment-system interlinking explains that connecting systems can shorten transaction chains and can improve cost, transparency, and speed; it also distinguishes bilateral links from more integrated arrangements (Committee on Payments and Market Infrastructures [CPMI], 2022). The setting is regulated payment infrastructure rather than open token markets, but the systems insight is general: fragmentation creates interfaces, and interfaces require coordination.
The BIS’s 2026 analysis makes the digital-asset version unusually concrete. It identifies fragmentation across public permissionless blockchains and notes that even stablecoins carrying the same name on different chains can remain siloed because the ledgers do not natively communicate. In other words, nominal sameness is not enough if the settlement object is operationally fractured across incompatible environments (BIS, 2026).
A universal settlement asset must therefore be universal in an economic sense, not merely duplicated under the same ticker on many isolated systems.
Conversion Is Not Free Merely Because Software Performs It
Digital finance sometimes treats conversion as though automation abolishes cost.
It does not.
A software agent can execute a swap in milliseconds, but the swap can still contain a bid–ask spread, market impact, liquidity risk, network fees, bridge risk, oracle dependence, execution uncertainty, tax consequences, accounting consequences, and residual balances. Automation can reduce the labour cost of performing these actions. It cannot transform an economically unnecessary conversion into a free one.
The World Bank’s Remittance Prices Worldwide programme provides a useful real-world reminder. Its latest highlighted global statistic reports that sending remittances costs an average of 6.36% of the amount sent (World Bank, 2025). That figure should not be abused. The full cost is certainly not “the price of foreign exchange”. Compliance, distribution networks, capital, fraud, local competition, cash handling, and operating costs all matter. The relevant lesson is more modest: moving purchasing power across monetary and institutional boundaries consumes real resources.
Digital architecture should therefore be judged partly by the number of boundaries it manufactures.
If a payment must cross a boundary because the buyer is in Thailand and the seller is in Germany, there may be unavoidable legal, tax, and foreign-exchange consequences. If a payment must cross another boundary solely because the storage platform chose StorageCoin while the AI platform chose InferenceCoin, that additional conversion is architectural choice.
The correct question is not:
How cheaply can we swap among thousands of tokens?
It is:
Why are thousands of swaps necessary?
A Cost Equation for the Platform-Token Economy
The economic effect can be stated in a simple form suitable for machine procurement.
Let the delivered cost of buying service i be:
Cᵢ = Pᵢ + Fᵢ + Xᵢ + Hᵢ + Lᵢ + Rᵢ
where:-
Pᵢ = the underlying price of the service;
-
Fᵢ = explicit payment and network fees;
-
Xᵢ = compulsory currency-conversion cost, including spread and slippage;
-
Hᵢ = holding and rebalancing cost of maintaining the required currency inventory;
-
Lᵢ = liquidity and liquidation cost, including the cost of residual balances; and
-
Rᵢ = expected loss associated with settlement, bridge, routing, or conversion failure.
A common transferable token does not make Fᵢ, Hᵢ, Lᵢ, or Rᵢ vanish. Every payment infrastructure has operational costs. Liquidity remains important. Law remains important. Network congestion can remain important. A universal token can itself be badly designed.
What common settlement can remove is a large class of compulsory platform-to-platform Xᵢ terms.
That difference is easy to miss because modern interfaces hide conversion. The customer clicks once; the software performs four actions. Economically, there are still four actions.
This matters especially when Pᵢ is tiny. A one-cent conversion overhead attached to a $1,000 transaction is negligible. The same overhead attached to a one-tenth-of-a-cent machine purchase destroys the transaction.
The lower the value of the underlying service, the more absurd proprietary monetary boundaries become.
AI Agents Make the Problem Impossible to Ignore
The strongest case for a common transferable token may not be human retail commerce. It may be autonomous machine commerce.
Consider an AI agent instructed to produce a defensible market report. It may purchase search, proprietary data, model inference, compute, storage, translation, verification, chart rendering, and perhaps a small amount of human expert review. The optimal suppliers may be different firms in different jurisdictions.
Now imagine that every supplier requires its own currency.
The agent needs either pre-funded balances in every relevant token or permanent access to exchange infrastructure. Before buying the cheapest service, it must ask a second set of questions unrelated to the service itself:
Does it hold the required token? Is there enough liquidity to obtain it? What is the spread? What is the expected slippage? Is the bridge operating? What is the fee? Will there be an unusable residual balance? Does the conversion alter which supplier is actually cheapest? Should the agent hedge the token between acquisition and consumption?
The agent is no longer merely an autonomous buyer.
It is an autonomous corporate treasurer.
Suppose the agent may buy from k service ecosystems. Under proprietary settlement, its monetary state can require an inventory vector:
M = (m₁, m₂, m₃, …, mₖ)
where each mⱼ is the working balance of a different platform currency.
With one commonly accepted settlement asset, the machine can often reduce this to:
M = (m)
This does not eliminate foreign exchange between the token and national currencies. It does not eliminate collateral management. It does not eliminate credit. It does something simpler: it reduces the number of platform-specific monetary inventories the agent must manage merely to buy substitutable digital services.
That reduction in state dimensionality matters for automated systems. Every additional balance needs acquisition rules, minimum reserves, price feeds, accounting, security, liquidation logic, and exception handling.
Machines should become better at buying services, not better at tolerating badly designed money.
Monetary Dust Is a Form of Dead Capital
Closed monetary systems create another cost: residual balances.
Humans know the phenomenon from prepaid cards, game credits, transit balances, airline points, and foreign coins left in a drawer after a holiday. A proprietary digital-token economy can reproduce the same problem at machine scale.
An autonomous agent may buy 100 units of a platform token and spend 97.4. The remaining 2.6 units may be below an efficient conversion threshold, below a withdrawal minimum, or simply too small to justify the spread and fees required to liquidate them. Repeat this across hundreds of services and millions of agents, and monetary dust becomes stranded working capital.
A common transferable settlement token changes the fate of the residue. The 2.6 units are not “left over from Platform A”. They remain 2.6 units that can be spent on Platform B.
This is a small institutional change with a large conceptual consequence:
A balance stops belonging economically to the service in which it was earned.
That is what portable money should do.
Monetary Switching Costs and Balance Gravity
Platform economics usually discusses lock-in through proprietary data formats, APIs, learning costs, contractual commitments, network effects, and migration expense. Monetary fragmentation adds another mechanism: balance gravity.
Suppose a firm holds a material balance of PlatformToken A. Provider B offers the same underlying service at a slightly lower price. Switching is economically attractive only if the saving exceeds the cost of exiting A’s monetary system.
We can state the choice as:
Switch to B if: S > Eₐ + Aᵦ
where:-
S = expected saving from using B instead of A;
-
Eₐ = monetary exit cost from liquidating or repurposing A’s token; and
-
Aᵦ = monetary acquisition cost of obtaining B’s token.
The service may be perfectly portable while the money is not.
A common settlement token pushes Eₐ and Aᵦ toward zero for the platform-currency component of the decision. Other switching costs remain. Data may still need migration. Contracts may still impose notice periods. Staff may need retraining. But the customer no longer needs to liquidate one private money and acquire another merely to patronise a competitor.
That improves contestability.
A platform that cannot rely on monetary captivity must retain users by being better.
Price Discovery Requires a Common Denominator—But Not Necessarily a New One
The original attraction of money as a unit of account is informational. It permits heterogeneous goods to be compared through a common denominator.
If Provider A charges US$0.0040 for a standardised inference job and Provider B charges US$0.0044, the comparison is immediately intelligible. If A charges 19 AlphaTokens and B charges 7.4 BetaTokens, comparison requires exchange rates and assumptions about when and how conversion occurs.
A universal settlement token helps, but only if we are precise about what it helps with.
If both providers quote directly in the common token, comparison is direct in token units. If the token is volatile relative to the firm’s accounting currency, however, the firm may still prefer to quote and budget in dollars, euros, or baht. In that case the common token reduces settlement fragmentation, while the conventional currency remains the unit of account.
This is not a defect in the thesis. It is a refinement.
The digital economy does not need to solve every monetary problem simultaneously. It can obtain a substantial efficiency gain simply by separating the question “What is this service worth?” from the question “Which proprietary token does this platform force me to acquire?”
The BIS’s 2026 report is useful here because it treats a common unit of account, singleness, and interoperability as related but distinct supports for monetary coordination (BIS, 2026). The same analytical discipline should be applied to token systems.
One settlement asset does not automatically guarantee one stable numeraire.
But one settlement asset can still eliminate a remarkable amount of unnecessary conversion.
The Architecture: Many Services, One Portable Settlement Layer
The system envisaged here is deliberately plural above the monetary layer.
Figure 2. Conceptual architecture. AI inference, cloud compute, storage, data, energy/device markets, and human services remain separate and competitive. Prices may be quoted in an agreed accounting numeraire or directly in token units. The common settlement token supplies portability across domains; it does not require one service provider, one application, or one global database.
This distinction is essential.
One token does not mean one company.
One token does not mean one marketplace.
One token does not mean one application.
One token does not mean one ledger.
One token does not mean one legal jurisdiction.
One token does not mean one price.
It means that an economic actor can preserve a settlement asset while changing counterparties.
The BIS’s 2023 tokenisation blueprint explicitly contemplates multiple ledgers with specific use cases coexisting and being interlinked through APIs (BIS, 2023). That institutional proposal is not the same as the universal-token architecture advanced here, but it demonstrates the underlying systems point: economic integration does not require all activity to inhabit one database.
The rail and the money are separable design questions.
The ideal architecture therefore resembles an hourglass. At the top are thousands of services, contracts, applications, and business models. At the bottom may be diverse technical infrastructure. At the narrow waist is a settlement object whose rules are sufficiently stable and whose acceptance is sufficiently broad that value can pass between the domains without being reissued as platform scrip.
Decentralise the applications.
Decentralise the providers.
Decentralise access.
Do not decentralise the denominator into meaninglessness.
Standards Solve Interfaces; They Do Not Solve Monetary Fragmentation
Technical standards are indispensable because they reduce the cost of connecting systems.
ERC-20 is a good example. Its purpose is to provide a common interface for fungible tokens so that applications can work with them consistently (Vogelsteller & Buterin, 2015). The W3C’s Payment Request API similarly standardises communication between merchants, users, and payment methods while deliberately allowing multiple payment methods (World Wide Web Consortium [W3C], 2026).
Both demonstrate the value of standardisation.
But neither proposition should be confused with monetary singleness.
Standardising the shape of many monies does not turn them into one money.
A wallet can understand Token X and Token Y perfectly. Both can expose identical transfer functions. A browser can offer them in the same payment interface. An exchange can swap them with one click. If X and Y fluctuate against one another and each is required by a different platform, the user still holds two economic assets.
Technical composability solves an interface problem.
Monetary portability solves a balance-sheet problem.
This distinction is particularly important in discussions of bridges. A bridge may allow an asset representation to move between networks, but the user must ask what exactly has crossed. Has the same settlement claim remained directly redeemable and economically equivalent? Has a wrapped claim been created? Has counterparty risk changed? Has finality changed? Has the user simply exchanged one token for another that tracks it?
“Interoperable” is therefore not a sufficient description of monetary architecture.
The better question is:
After the interoperability machinery has done its work, does the user still need to sell one money and buy another?
If yes, the systems have been connected. The monetary fragmentation has not necessarily been removed.
Cross-Border Payments Show Why Architecture Matters
Cross-border payments provide a useful analogue because the world already knows what happens when payment systems, institutions, currencies, messaging standards, and compliance regimes are fragmented.
The CPMI’s work on interlinking payment systems emphasises that cross-system links can shorten transaction chains and improve speed, transparency, and cost efficiency (CPMI, 2022). Its later work on cross-border fast-payment-system links stresses that governance and oversight become harder precisely because multiple jurisdictions, currencies, and institutional arrangements must be coordinated (CPMI, 2024).
These reports concern regulated payment systems, not a universal open token. They therefore do not prove the proposal made here. They do, however, demonstrate a general economic fact: inter-system boundaries have operational consequences.
The World Bank’s remittance data show the consumer-facing manifestation of those consequences. Again, the global average cost of remitting money cannot be attributed solely to currency conversion. But it would be equally mistaken to infer that because compliance and distribution matter, monetary and payment fragmentation do not.
A globally transferable digital settlement asset would not abolish borders. It would not abolish sanctions law, tax, consumer protection, capital controls, identity requirements, licensing, or contract law. It would not make every transaction legal merely because the software can transmit it.
The claim is instead architectural:
Jurisdiction-specific rules need not imply platform-specific money.
A payment can be screened, reported, taxed, or restricted according to the applicable law without requiring the economic object being paid to mutate into a different proprietary currency at every commercial boundary.
That is a cleaner separation of concerns.
Micropayments Turn Friction Into a First-Order Problem
The case for common settlement becomes stronger as transaction size falls.
A $10,000 purchase can absorb a few cents of operational overhead. A $0.001 API call cannot.
Machine commerce is likely to contain enormous numbers of small economic events: model queries, sensor readings, storage retrievals, data fragments, verification services, tiny compute jobs, bandwidth, content access, agent-to-agent services, and automated royalties. At these scales, the payment system must minimise not merely percentage fees but the fixed cognitive and computational structure surrounding each payment.
An agent should not conduct a miniature foreign-exchange operation before buying one-thousandth of a dollar of computation.
Suppose a workflow makes 10,000 micro-purchases across 50 services. If each service requires a proprietary token, token acquisition becomes part of the workflow graph. The agent needs price feeds, liquidity feeds, route selection, inventory thresholds, residual-balance rules, and failure recovery for financial operations unrelated to the substantive task.
The payment architecture consumes intelligence.
A common transferable token turns settlement closer to a primitive:
If condition y is satisfied, transfer x units.
That simplicity matters because programmable transactions become most powerful when payment is composable with the underlying commercial action. The BIS’s tokenisation work emphasises the possibility of integrating messaging, reconciliation, settlement, and contingent execution on programmable platforms (BIS, 2023, 2025). A machine economy can use those capabilities most cleanly when every application is not also demanding its own money.
The ideal micropayment is not merely cheap.
It is economically boring.
Why “A Token for Every Use Case” Gets the Logic Backwards
There are perfectly legitimate reasons to create distinct digital assets.
A share represents an ownership claim. A bond represents a debt claim. A warehouse receipt can represent title to goods. A ticket can represent a permission to enter. A governance credential can represent a voting right. A loyalty point can represent a contractual marketing benefit. An identity credential can represent an attested status.
Those are different things because the underlying legal and economic rights are different.
The mistake is to infer that because different claims can be tokenised, every application requires a different money.
Suppose a storage provider sells disk capacity. The economically relevant product is storage. Its competitive dimensions may include redundancy, retrieval speed, geography, privacy, durability, reputation, and price. None of those characteristics logically requires the provider to issue StorageCoin.
The same is true for compute. A GPU marketplace can describe accelerator type, memory, benchmark performance, location, availability, latency, reliability, and trust characteristics. It may use cryptographic proofs or specialised contractual tokens. Payment can still occur in a common settlement asset.
Platform tokens often bundle three distinct functions:-
financing the platform;
-
governing the platform; and
-
paying for the platform’s service.
There is no economic law requiring one instrument to perform all three.
Bundling can actively confuse incentives. If a service token appreciates because of speculative demand, the effective cost of using the service may become detached from its underlying resource cost. If governance rights ride on the same token, a customer buying compute may simultaneously acquire political power the customer neither wants nor values. If token issuance finances development, service customers become entangled with a capital-raising mechanism.
A cleaner architecture separates functions.
Let capital instruments finance capital.
Let governance instruments govern where governance is needed.
Let title tokens represent title.
Let access credentials represent access.
Let money settle.
The “Universal Token” Must Not Become a Universal Platform
The strongest objection to one transferable token is obvious: would it simply create a larger monopoly?
It could.
A token is not decentralised merely because it is broadly accepted. If one company controls issuance, transaction approval, wallets, pricing, redemption, identity, rule changes, and access, then universality could produce concentrated power on a scale greater than the fragmented systems it replaced.
The design objective must therefore be monetary commonality without institutional monopoly.
That means separating the token from particular service providers. Competing firms should be able to accept it. Competing wallet implementations should be able to hold it. Competing processors should be able to build services around it where the protocol permits. Competing marketplaces should be able to price in it. A customer should not need the permission of the firm that originally paid the token in order to spend it elsewhere.
The monetary layer should be boring infrastructure.
That is not an insult. Infrastructure is most valuable when it allows innovation above it without demanding that every new application reinvent the substrate.
The universal-token thesis is therefore not “centralise all finance into one token company”. It is the opposite: make the settlement object portable enough that firms cannot easily use proprietary money as a fence around their customers.
A Twelve-Part Test for a Genuine Cross-Platform Settlement Token
Calling an asset “universal” does not make it so. A serious settlement token for global digital commerce would need to satisfy a demanding set of conditions.
1. Economic transferability. A recipient can spend received value outside the service in which it was earned without mandatory redemption into platform scrip.
2. Fungibility at the settlement layer. Equivalent units perform the same payment function rather than being economically stratified by arbitrary provenance.
3. Adequate divisibility. The denomination supports machine-scale purchases without economically destructive rounding.
4. Low marginal transaction cost. Small payments remain viable.
5. Rapid and intelligible finality. A seller can determine when payment is sufficiently settled to release the corresponding good or service.
6. Scalable capacity. The system can support volumes consistent with machine commerce rather than forcing most activity into proprietary off-ledger balances.
7. Open acceptance. Independent service providers can accept the token without being forced into a vertically integrated commercial franchise.
8. Stable and predictable rules. Businesses can write contracts without assuming that the basic settlement rules will be discretionarily rewritten every few months.
9. Auditable monetary rules. Participants can determine what they are receiving, how supply changes, and what conditions govern transfer.
10. Legal intelligibility. Commercial actors can determine what constitutes transfer, discharge, ownership, finality, and remedy in relevant jurisdictions.
11. Interoperability without compulsory currency exchange. Different applications or rails may exist, but moving between them does not automatically require buying a new proprietary money.
12. Application neutrality. The settlement layer does not privilege AI over storage, one cloud provider over another, or one merchant category over another.
This test is intentionally harder than “can the token be sent to another wallet?”
A universal settlement asset is public economic infrastructure in function even if it is not state-issued in form. It should be judged accordingly.
Network Effects Favour Shared Money More Than Shared Platforms
Money exhibits a fundamental acceptance externality: a unit is more useful to me when I expect more other people to accept it.
Platform-token rhetoric often notices this and then draws the wrong institutional conclusion. A project claims that as its ecosystem grows, its native token becomes more useful. True. But if every project follows the same strategy, the economy divides the network effect among thousands of incompatible monetary communities.
Imagine the principle applied outside digital assets.
Every supermarket issues its own currency. Every taxi company another. Every electricity supplier another. Every employer pays wages in its own token. Sophisticated exchanges make the system workable. Smartphones automate conversions. Algorithms discover the cheapest route through the currency graph.
Technically impressive.
Economically deranged.
The question would remain: why?
The technology required to make a poor monetary architecture tolerable should not be confused with evidence that the architecture is good.
A common transferable token captures the network effect at the settlement layer while leaving competition open at the service layer. Every additional accepting merchant makes the settlement asset more useful to existing holders. Every additional holder makes acceptance more attractive to merchants. The network effect therefore reinforces portability rather than platform captivity.
This is also why the BIS places such emphasis on monetary singleness and interoperability. Its institutional solution differs from the proposal here, but the underlying coordination problem is the same: money works better when acceptance is not continuously renegotiated instrument by instrument (BIS, 2025, 2026).
The Costing Revolution
The most underappreciated benefit of common settlement is not ideological.
It is accounting.
Businesses live by cost comparison. Autonomous agents will do so even more relentlessly.
Consider a firm deciding where to execute a workload. Its economic cost per verified unit of output might be written as:
K = C + E + S + B + V + T + Q
where:-
C = compute;
-
E = energy;
-
S = storage;
-
B = bandwidth;
-
V = verification;
-
T = transaction and settlement cost; and
-
Q = expected quality-adjustment or failure cost.
If each component is purchased using a different proprietary token, K is not simply the sum of service prices. Each term inherits an exchange-rate mapping into the firm’s accounting numeraire. Those mappings may change between quote and execution. Thin-token markets may impose slippage. Hedging may be impractical for microtransactions. Residual balances may distort future procurement because an agent can rationally prefer a worse supplier merely to consume tokens it already owns.
Common settlement removes one layer of noise from that calculation.
It does not guarantee that markets are competitive. It does not guarantee that the token is stable. It does not guarantee low network fees. What it does is make the payment asset less likely to be the reason two otherwise comparable service prices cannot be compared cleanly.
This is a profound point for automated procurement.
A machine should be able to ask, “Who provides the cheapest verified unit of output?”
It should not first have to ask, “In which proprietary currency am I already accidentally wealthy?”
Global Means Global—But Law Still Exists
“One thing that can be instantly done anywhere on Earth” is a compelling design objective, but it requires precision.
A settlement asset can be technically global without every transfer being legally unconstrained. Nations retain tax systems, sanctions, licensing regimes, consumer-protection rules, capital controls, anti-money-laundering obligations, and contract law. Any serious global architecture must coexist with those institutions or explicitly challenge them; it cannot simply pretend they are absent.
The goal should therefore be stated in economic rather than magical terms:
The settlement asset should be natively portable across geography wherever the underlying transaction is lawful and technically reachable, without requiring a new platform currency at each commercial boundary.
That is already a radical improvement.
It means a Thai household selling spare compute to a German buyer need not acquire GermanComputeCoin. A Singapore data service paid by a British AI company need not retain BritishInferenceToken. A storage provider that earns the common token can spend it with an energy supplier or a human contractor rather than first returning to an exchange.
The token follows commerce.
Commerce does not follow the token issuer.
The Hard Objections
A serious thesis becomes stronger when the difficult objections are admitted rather than concealed.
Volatility
A universal token that changes purchasing power violently may be a poor unit of account even if it is a highly portable settlement asset. Businesses can quote prices dynamically, but constant repricing is a cost.
This is why the revised thesis does not pretend that settlement and denomination are identical. A firm can maintain accounts in a relatively stable sovereign currency while settling selected digital transactions in the common token. If the token eventually becomes sufficiently stable and widely accepted to serve as a unit of account, that is an additional achievement rather than an assumption.
Liquidity concentration
A universal token needs deep liquidity if users regularly enter from or exit to national currencies. Concentrating liquidity in one asset can reduce fragmentation, but it can also concentrate market risk. Market structure therefore matters.
Governance capture
A common token whose rules can be changed opportunistically by a small group may become a single point of political and commercial control. Universality therefore raises rather than lowers the standard required for predictable rules and credible governance constraints.
Technical scaling
If the settlement infrastructure cannot scale economically, applications will recreate internal balances, payment channels, custodial credits, or proprietary off-ledger systems. Those mechanisms may be efficient, but if they become non-portable they can recreate the fragmentation the universal token was intended to reduce.
Privacy
A universal settlement layer must not become a universal surveillance layer. Broad portability combined with perfect public traceability could produce a commercial intelligence system of extraordinary scope. Privacy is therefore a first-order architectural requirement, not an optional feature.
Credit
Money is not the whole financial system. Firms need credit, netting, liquidity provision, collateral, and delayed settlement. One transferable token does not remove the need for financial intermediation. It merely prevents every commercial platform from turning payment into an unnecessary proprietary-currency problem.
Monetary sovereignty
Governments are unlikely to abandon national currencies merely because AI agents prefer a global settlement object. Nor must they. National currencies can remain dominant units of account, tax units, and banking denominations while a transferable digital asset serves specialised cross-platform settlement functions.
“Why not stablecoins?”
Stablecoins illustrate both the attraction and the difficulty. They aim to combine digital transferability with a familiar unit of account, but current architectures remain fragmented across issuers and blockchains. The BIS’s 2026 report specifically highlights this fragmentation and the fact that same-named stablecoins on different chains are not automatically interoperable (BIS, 2026). The lesson is not that stable value is undesirable. It is that a familiar denomination does not by itself create universal settlement portability.
None of these objections defeats the central argument.
They define the work required to make it credible.
A Better Rule for Token Design
The token industry has spent years perfecting the art of explaining why every new project needs a token.
The presumption should be reversed.
Before creating a new monetary asset, ask:
What economic function cannot be performed using an existing transferable settlement asset?
If the answer is governance, issue a governance right.
If the answer is title, issue a title claim.
If the answer is an investment security, structure the security.
If the answer is loyalty, issue loyalty points.
If the answer is access, issue an access credential.
If the answer is merely “customers need to pay us”, a new currency has not yet been justified.
This shift would move the intellectual centre of digital-asset design away from token proliferation and toward market architecture.
The objective would no longer be to manufacture artificial monetary scarcity around every application. It would be to make heterogeneous applications economically interoperable while preserving competition among them.
That is less glamorous than launching a new coin.
It is also vastly more important.
The Market We Actually Want
Consider the end state.
A household owns spare GPU capacity. It offers compute into a global market. An AI company buys that capacity and pays in a transferable settlement token. The household uses part of the proceeds to buy cloud storage from another provider. The storage provider pays a data supplier. The data supplier pays a translator in another country. The translator buys an unrelated digital service. A software agent earns the same token for one task and spends it on another.
At no point is anyone forced to redeem into the proprietary currency of the previous platform merely to continue participating in commerce.
The token has travelled through different industries, counterparties, applications, and jurisdictions while retaining its settlement identity.
Prices remain competitive because providers can quote against a common numeraire or directly in the common settlement asset. Entry becomes easier because a new supplier does not have to persuade customers to acquire a new currency before trying its service. Exit becomes easier because a departing user takes working capital away intact. Autonomous agents optimise service cost rather than token inventory. Residual balances remain spendable rather than becoming dust. Service providers compete for customers instead of monetising the friction of escape.
This is decentralisation in an economically meaningful sense.
It contains many firms.
Many owners.
Many applications.
Many contractual forms.
Many prices.
Many technical implementations.
Many competing ideas.
It does not require many monies.
Conclusion: Decentralise Everything Except the Denominator
The great error of the token era has been to confuse multiplication with decentralisation.
A thousand currencies do not necessarily create a thousand centres of economic freedom. They can create a thousand tollbooths. Every proprietary platform currency can become a small wall: easy to enter while subsidies are abundant, annoying to leave when liquidity is thin, and permanently demanding another conversion from anyone who wants to carry value elsewhere.
The more radical architecture is simpler.
Let AI providers compete.
Let cloud providers compete.
Let storage networks compete.
Let data markets compete.
Let energy sellers compete.
Let wallets compete.
Let payment processors compete.
Let marketplaces compete.
Let autonomous agents roam among them.
But give them a settlement object that can travel.
The decisive property is not that a token exists on a blockchain. It is not that it can be wrapped, bridged, swapped, routed, or displayed in a wallet. The decisive property is that purchasing power survives movement from one economic domain to another without compulsory monetary metamorphosis.
The deepest benefit is not ideological purity. It is the removal of unnecessary state from the economy: fewer balances to manage, fewer conversions to price, fewer residuals to liquidate, fewer monetary switching costs, fewer interfaces between the buyer and the thing the buyer actually wants.
A decentralised digital economy therefore needs fewer currencies, not more.
It needs a common transferable settlement layer against which heterogeneous markets can remain heterogeneous. It needs money that can leave a platform when the customer does. It needs a settlement asset that a machine can earn in one context and spend in another without becoming a foreign-exchange trader between every operation.
Prices may still be denominated in dollars, euros, baht, yen, or whatever unit contracting parties choose. Ledgers may remain technically diverse. Firms may remain organisationally distinct. Jurisdictions will remain legally plural.
The token does not have to homogenise the world.
It has to be able to move through it.
That is the paradox worth taking seriously: monetary portability can increase institutional decentralisation because it weakens the monetary fences around platforms.
One token.
Many markets.
Not because one organisation should own the economy, but because no platform should need to own the customer’s money in order to sell the customer a service.
References
Bank for International Settlements. (2023). Blueprint for the future monetary system: Improving the old, enabling the new. In Annual Economic Report 2023. https://www.bis.org/publications/aer-2023/blueprint-future-monetary-system-improving-old-enabling-new
Bank for International Settlements. (2025). The next-generation monetary and financial system. In Annual Economic Report 2025. https://www.bis.org/publications/aer-2025/next-generation-monetary-financial-system
Bank for International Settlements. (2026). Anchoring trust in money: Innovation beyond stablecoins. In Annual Economic Report 2026. https://www.bis.org/publications/aer-2026/anchoring-trust-money
Committee on Payments and Market Infrastructures. (2022). Interlinking payment systems and the role of application programming interfaces: A framework for cross-border payments. Bank for International Settlements. https://www.bis.org/publications/interlinking-payment-systems-and-role-application-programming-interfaces-framework-cross-border-payments
Committee on Payments and Market Infrastructures. (2024). Linking fast payment systems across borders: Governance and oversight—Final report. Bank for International Settlements. https://www.bis.org/publications/linking-fast-payment-systems-across-borders-governance-and-oversight-final-report
Ethereum.org. (2026). ERC-20 token standard. https://ethereum.org/developers/docs/standards/tokens/erc-20/
Vogelsteller, F., & Buterin, V. (2015). ERC-20: Token standard (Ethereum Improvement Proposal No. 20). Ethereum Improvement Proposals. https://eips.ethereum.org/EIPS/eip-20
World Bank. (2025). Remittance Prices Worldwide.
https://remittanceprices.worldbank.org/
World Wide Web Consortium. (2026, June 22). Payment Request API (Candidate Recommendation Draft). https://www.w3.org/TR/payment-request/