A forecast is not an architecture. That is the first rule of due diligence, and it is the rule the market forgets every cycle.
Morgan Stanley's research desk put a number on the table: tokenized assets reach ten trillion dollars by 2030. The press cycle treated the figure as a milestone. My terminal treated it as an anomaly. Not because ten trillion is large. Because the number arrived without a specification attached.
I have spent twenty-eight years watching systems get described before they get built, and I have learned to read the description as a liability until the execution layer exists. A ten-trillion-dollar claim is not a market-size claim. It is a claim about settlement finality, identity resolution, custody architecture, cross-chain interoperability, and oracle integrity. The note, as it circulated, contained none of these. It contained a number.
That absence is the signal. When a Tier-1 investment bank publishes a ten-trillion-dollar forecast for a category and cannot, or will not, name the protocol standards that will carry it, you are not reading a research product. You are reading positioning. And positioning, in my framework, is metadata. Execution is final; intention is merely metadata.
So let me do what the note did not do. Let me open the category at the code level and ask the only question that matters: what would actually have to ship for ten trillion dollars to settle on-chain?
What Was Actually Said
Strip the narrative and the published content reduces to four points. Morgan Stanley forecasts tokenized assets reach ten trillion dollars by 2030. The growth may transform financial markets. It may drive innovation. It may increase institutional participation. A source attribution closes the set.
That is the entire payload. There is no named protocol. No token standard. No custody design. No settlement mechanism. No regulatory pathway. No on-chain data. The forecast is a size claim with no substrate beneath it.
This matters more than the size. A forecast is a commitment about mechanics, whether the author admits it or not. Ten trillion by 2030 is not an opinion about sentiment. It is an implicit assertion that the following all exist simultaneously: a legally recognized token standard for securities, a compliant identity layer that satisfies KYC and AML across jurisdictions, a custody model acceptable to qualified custodians, a settlement asset for the cash leg of every trade, and a cross-chain fabric that does not become an attack surface. The note asserts none of this. It skips to the outcome.
I want to be fair to the format. A short research note is not an engineering whitepaper. But there is a difference between omitting implementation detail for brevity and omitting it because the analysis never reached that layer. The tell is the absence of any named rail. Every serious tokenization effort names its stack: this chain, this standard, this custodian. The forecast names none. That is not brevity. That is a market-size projection standing in for an architecture.
And that is where the information gain lies for a reader. The value of a forecast is not the number. It is the architecture the number implies. When the architecture is missing, the reader is not receiving a conclusion. The reader is receiving a conclusion with the reasoning deleted. My job is to reconstruct the reasoning and check whether it holds.
The Historical Precedent Nobody Cites
Here is the part of the story the forecast leaves out, and it is the most important part.
This is not the first time the industry has been promised a multi-trillion-dollar tokenized securities market. In 2017 and 2018, the security token offering, the STO, was going to be the next wave. tZERO raised capital on the thesis. Polymath built a security token standard. Harbor, Securitize, and a dozen others promised to bring real-world assets on-chain. The standards were written. The custody partnerships were signed. The conferences were held.
Almost none of it reached scale. The reason was not technology. The technology of 2018 was adequate to tokenize a fund. The reason was that the legal framework, the custody acceptance, and the settlement asset never arrived together. The STO wave died of inheritance failure: it tried to bring traditional assets on-chain without bringing the traditional legal and operational framework with them. The assets had no legal home on the chain.
This precedent matters because the 2025 forecast assumes the inheritance problem is solved. It is not. It is marginally closer. The token standards matured. ERC-3643 is genuinely better than what existed in 2018. The custody infrastructure matured. Qualified custodians now handle digital assets as a matter of routine. But the two binding constraints, legal recognition and the settlement asset, remain substantially unresolved. The forecast treats a five-year-old partial solution as a completed one.
I do not raise the precedent to dismiss the category. I raise it because the category has a track record, and the track record says the binding constraint is not enthusiasm and it is not capital. It is the unglamorous integration of law, custody, and settlement. A forecast that ignores the track record is not a forecast. It is a wish with a decimal point.
The Anatomy of Tokenization: What Ten Trillion Requires
Tokenization is not a synonym for issuing a token. This is the confusion that produces bad forecasts. A tokenized bond is not a new asset. It is the same bond with a new settlement rail. The economics do not change. Coupon, maturity, seniority, all unchanged. What changes is the plumbing.
The plumbing has five layers, and every one of them is a place where a ten-trillion-dollar forecast can fail.
Layer one: the token standard. ERC-20 does not work for securities. It has no transfer restrictions, no identity binding, no native handling of corporate actions. A security token must enforce who can hold it, where it can move, and under what conditions. The industry solved this with standards like ERC-1400, which introduces partitions and a controller role, and ERC-3643, the T-REX protocol, which binds every transfer to an on-chain identity and a compliance contract. In ERC-3643, the transfer function does not simply move balances. It calls a compliance check before the balance updates. If the receiving identity is not verified, or the compliance module rejects the transfer, the transfer reverts. The token is permissioned by construction. This is correct for securities. It is also the first place where the architecture diverges from the crypto-native ethos of permissionless transfer.
Layer two: identity. Every holder needs an on-chain identity contract. In ERC-3643, that is an ONCHAINID. Plus a registry of trusted claim issuers who attest to KYC status, accreditation, jurisdiction, and investor classification. This is not a checkbox. It is a parallel system of record. It must resolve identity across borders, survive corporate restructuring, and remain auditable for a decade. The identity layer is where most tokenization pilots quietly die, because it is expensive, it is jurisdictionally fragmented, and it collides with privacy law. You cannot immutably record personal data and also honor a right to erasure. The reconciliation is zero-knowledge proofs, and institutional-grade zero-knowledge identity is not ready.
Layer three: custody. A tokenized asset held by an institution does not sit in a retail wallet. It sits with a qualified custodian, behind multi-party computation key management or hardware security modules. The custody model determines who can move the asset, who can freeze it, and who carries the liability when a key is compromised. This is where my own work on machine-to-machine custody standards became relevant. The hard problem is never the token. It is the key lifecycle. A custodian that holds keys for a ten-trillion-dollar asset base is a systemic risk node, and the standards governing its key management are the real infrastructure of the market.
Layer four: settlement. Here is the problem the forecasts never mention. You can tokenize the asset. You still need to settle the cash leg. Delivery-versus-payment requires a settlement asset that both sides accept. Central bank money is not on-chain. Commercial bank money is not on-chain at scale. Stablecoins are on-chain but carry issuer and regulatory risk. The cash leg is the unsolved half of the equation, and it is not a technical detail. It is the difference between a demo and a market.
Layer five: interoperability. Tokenized assets will not live on one chain. They will live on Ethereum, on permissioned networks, on bank-controlled ledgers. Moving value between them means bridges, and bridges are the most attacked surface in the industry. The cross-chain fabric is not a solved problem. It is an active battlefield.
That is the architecture. Now count the places it can break. Five layers, each with its own failure mode, each requiring coordination across legal, financial, and technical domains that have never coordinated before. A ten-trillion-dollar forecast is a claim that all five layers mature on schedule. The note does not even acknowledge that they exist.
The Gap: Three Orders of Magnitude
Now the data. On-chain RWA value, excluding stablecoins, sits in the low tens of billions of dollars. Tokenized Treasuries, the fastest-growing, most institutionally credible subcategory, home to BlackRock's BUIDL, Franklin Templeton's BENJI, and Ondo's products, measured in the single-digit billions. Let me hold the number loosely, because it moves, but the order of magnitude is stable. Roughly fifteen billion dollars, give or take.
The forecast is ten trillion. The ratio is roughly six hundred and sixty-six to one. Over five years, that implies a compound annual growth rate near 267 percent. Not 267 percent cumulative. 267 percent per year, every year, for five years straight.
I have audited enough systems to know what that growth rate implies. It does not imply a trend. It implies a phase change. A 267 percent annualized rate sustained for five years is not the continuation of a curve. It is the discovery of a new one. And phase changes do not happen because a bank publishes a number. They happen because a constraint is removed.
The constraint here is regulatory inheritance, not technology. Tokenization's technology is largely built. ERC-3643 works. Custody infrastructure works. The identity layer works, in fragments. What does not exist is a coherent legal framework that lets a tokenized security be recognized, transferred, and settled as a security across jurisdictions. That is a legal problem wearing a technical costume.
And here is the read the forecast hides. If the constraint is legal, then the winners are not the crypto-native protocols. They are the institutions that already own the legal rails. The forecast, by staying at the level of market size, avoids the question of who captures the market. That avoidance is the analysis.
The Stablecoin Contrast
There is a useful comparison that clarifies why tokenized assets have not scaled the way the forecast assumes: stablecoins.
Stablecoins did reach scale. They crossed into the hundreds of billions. They found product-market fit. And they did it because they solved a specific, acute problem, dollar access and cross-border settlement, with a clean architecture. One asset, one issuer, one settlement function, no identity layer on the token itself, no corporate actions. A stablecoin is a bearer instrument. It does not need a compliance contract to transfer. It does not need an on-chain identity to hold. It solved the cash leg by being the cash leg.
Tokenized securities are the opposite. They need the identity layer. They need the compliance module. They need corporate action handling. They need a cash leg that is separate from the asset leg. Every one of those requirements is a friction that stablecoins do not have. The comparison is not flattering to the ten-trillion-dollar forecast. Stablecoins found their use case because they removed friction. Tokenized securities add friction, legally required friction, and then have to justify it with efficiency gains that have not yet materialized at scale.
This is the analytical core of my skepticism. The forecast borrows the growth curve of stablecoins and applies it to a category with structurally more friction and structurally fewer native use cases. The borrowing is unearned.
The Regulatory Substrate
Tokenization's binding constraint is legal, so the regulatory layer deserves its own section. And here the forecast is making a silent bet.
Tokenized securities require a definitional home. Is a tokenized Treasury note a security? Under U.S. law, almost certainly. It satisfies the Howey test without strain. That means it falls under SEC jurisdiction, which means it requires registration or an exemption, a transfer agent, a broker-dealer for distribution, and a qualified custodian. In the European Union, MiCA provides a framework, but MiCA's treatment of tokenized financial instruments is layered on top of existing securities law, which creates its own integration problem. The two regimes do not fully interoperate. A tokenized asset that is compliant in one jurisdiction is not automatically compliant in the other.
This fragmentation is the quiet killer of the forecast. A ten-trillion-dollar market implies cross-border flow. Cross-border flow of securities implies mutual recognition of token standards, identity attestations, and settlement finality across regimes. That mutual recognition does not exist. The forecast assumes it into being.
Here is the structural tell. Morgan Stanley is an SEC-regulated institution. Its published research passes through compliance review. A forecast that a ten-trillion-dollar tokenized asset market arrives by 2030 is, by necessity, a forecast that the regulatory framework becomes permissive rather than restrictive. The bank cannot publish a forecast premised on its own regulator tightening. So the forecast is, in part, a signal about the bank's expectation of regulatory direction. That is useful information. It is also self-interested information, and it should be weighted accordingly. Sell-side research has a structural optimism bias. The desk that publishes the forecast is often the desk that will sell the product. I weight it as a directional signal, not as a valuation input.
The Blind Spots: Where Tokenization Breaks
Now the part the forecast cannot see because it never looked at code. I will walk through the failure modes I have found in adjacent systems, because they generalize.
Admin keys are not power. They are liability. Tokenized securities require administrative control. Issuers must be able to freeze accounts, reverse erroneous transfers, comply with court orders, and manage corporate actions. This is not optional. It is a legal requirement. The consequence is that every compliant security token has an admin surface, a privileged role that can alter balances and restrictions. That surface is a single point of failure. In my 2021 work on NFT marketplace contracts, I found that royalty enforcement modules introduced exactly this kind of privileged path, and the reentrancy exposure it created was real. The lesson generalizes. The more compliance logic a contract carries, the larger its attack surface. Compliance is not a feature layer. It is the execution context.
Upgradeability is a trap dressed as a feature. Most tokenized asset contracts are upgradeable proxies, because securities law changes and issuers need to adapt. But an upgradeable contract means the code you audited is not the code that will run tomorrow. Inheritance is a feature until it becomes a trap, and proxy inheritance is the purest form of that trap. The logic can be swapped by whoever holds the upgrade key. In a permissioned, institutional context, that key is held by a custodian or an issuer. The trust model collapses to trusting the admin. That is a traditional trust model wearing blockchain clothing.
The oracle problem is unsolved for off-chain assets. A tokenized fund needs a net asset value. That NAV comes from off-chain computation, fed on-chain by an oracle. The oracle is a trusted input. If the oracle is wrong or manipulated, the token price is wrong. For tokenized Treasuries the risk is low, because the underlying is stable and priced by a deep market. For tokenized private credit, real estate, or private equity, the assets that make up the bulk of the ten-trillion-dollar forecast, the NAV is a mark-to-model estimate, not a market price. The oracle problem is the entire valuation problem. The forecast assumes it away.
Cross-chain bridges remain the soft target. I will not rehearse the bridge exploits. The industry has a decade of them. The point is structural. Tokenized assets want to be multi-chain for liquidity. Every bridge adds a trust assumption. A ten-trillion-dollar market cannot rest on bridges that lose nine figures to a single exploit.
The identity layer is a privacy and compliance collision. On-chain identity means personal data on a public ledger, or a zero-knowledge proof system that is far from institutional readiness. The travel rule, the GDPR, and the right to be forgotten do not reconcile cleanly with immutability. This is not a bug you patch. It is a design contradiction.
Each of these is a reason the forecast can miss. The note mentions none of them. That is not a research gap. It is a research choice.
Who Captures the Value
Let me make the value-capture question concrete, because this is where the retail audience gets misled.
Tokenized assets are assets. They are not tokens in the speculative sense. A tokenized Treasury note pays a coupon to its holder. It does not pay a governance token holder. The value created by tokenization flows to the asset issuer, the custodian, the transfer agent, the trading venue, the financial intermediaries. It does not, by default, flow to holders of RWA governance tokens.
This is the value distribution problem that the RWA narrative consistently omits. Ten trillion dollars of tokenized assets does not mean ten trillion dollars of value accrues to token investors. It means ten trillion dollars of assets move onto new rails, and the tolls are collected by whoever owns the rails. If those rails are permissioned networks run by consortiums of banks, the crypto-native investor captures almost nothing. If the rails are public, the picture changes. The forecast does not tell you which. It cannot, because it does not name the rails.
The value capture asymmetry is structural, not incidental. A protocol that provides ERC-3643 tooling to a major asset manager is a vendor. A vendor with a token is still a vendor. The token does not capture the asset's cash flows. It captures, at best, a fee on the vendor's license. That is a real business. It is not a ten-trillion-dollar business.
The Counterintuitive Angle: Success Is the Risk
Here is the angle that cuts against both the bulls and the bears. The bear says tokenization fails. The bull says it succeeds and everyone wins. I say the most likely outcome is that tokenization succeeds, and that success is precisely what hollows out crypto-native value.
Think about the mechanism. If tokenization works, it works because it inherits the legal and operational framework of traditional finance. The token standard inherits the transfer agent's role. The custody model inherits the qualified custodian. The settlement layer inherits the clearinghouse. The compliance module inherits the broker-dealer. Every inheritance is a place where a traditional intermediary keeps its function and captures its fee.
The crypto-native protocols, meanwhile, become suppliers. They provide the smart contract library, the identity primitive, the bridge. Suppliers get paid once. Intermediaries get paid on every transaction, forever. This is not a conspiracy. It is the natural consequence of building a compliant system. Compliance is defined by the incumbent framework, so the incumbent framework captures the compliant flow.
This is why I am skeptical of the RWA governance token thesis specifically. The token does not capture the asset's cash flows. The second-order effect is worse. If tokenized assets migrate to permissioned or consortium chains, and there is a strong institutional argument that they will, because banks prefer control to censorship resistance, then the public chains do not even capture the settlement demand. Ethereum becomes a notary, not a settlement layer. The forecast, read charitably, is a forecast about the size of the tokenized asset market. It is silent on who owns the market. Those are different questions, and the second is the one that determines whether any crypto investor benefits.
What to Watch Instead of the Number
The forecast is not useless. It is directionally informative about a long-term trend. Institutional assets are moving onto programmable rails, slowly, and the direction is one-way. What it is not is a tradeable signal, a token thesis, or a substitute for architecture review.
So watch the substrate, not the number. Three signals matter.
First, on-chain RWA value, tracked honestly, excluding stablecoins and excluding double-counted wrapped assets. If it does not grow by an order of magnitude before 2027, the forecast's implied growth rate is already broken. The number to watch is not ten trillion. It is whether the curve bends at all.
Second, the cash leg. The moment a major jurisdiction recognizes a settlement asset, central bank digital currency, or a regulated stablecoin, as acceptable for delivery-versus-payment, the binding constraint is removed. That is the real catalyst. Not a bank's research note. Watch the settlement asset, not the forecast.
Third, the rail choice. When the first top-ten asset manager tokenizes a fund, watch whether it lands on a public chain or a permissioned one. That single decision tells you where the value will settle for the next decade. A permissioned landing is a confirmation of the inheritance trap. A public landing is the first genuine signal.
I will close with the line I keep returning to, because it is the whole thesis compressed. Inheritance is a feature until it becomes a trap. Tokenization is the industry inheriting the entire architecture of traditional finance. Its standards, its intermediaries, its compliance, its custody. The inheritance is what makes the ten-trillion-dollar number plausible. It is also what makes the crypto-native capture of that number improbable.
The forecast may be right about the size. It is silent on the ownership. And ownership, not size, is the only thing a market participant should care about.
Execution is final; intention is merely metadata. Morgan Stanley gave us the intention.