Settlement Is Not Clearing: The Dangerous Confusion at the Heart of Tokenization Discourse
Another week, another tokenization report lands in my inbox. This time from Amundi, one of Europe's largest asset managers.
https://about.amundi.com/files/nuxeo/dl/4c73d033-5b89-439b-b702-e4851bd8eec4
150 pages of market analysis
regulatory mapping
ecosystem cataloguing
Professional production values.
Extensive sourcing.
And yet, like so many before it, it perpetuates a fundamental confusion that's becoming genuinely dangerous as institutional capital begins to take tokenization seriously.
The confusion is simple to state but apparently difficult to grasp: settlement is not clearing.
This distinction matters enormously. Getting it wrong doesn't just produce bad research, it produces bad infrastructure decisions, misallocated capital, and ultimately systemic vulnerabilities that will manifest precisely when we can least afford them.
I honestly don't understand why this keeps being got wrong.
The Claim That Won't Die
Here's a direct quote from the Amundi report, describing BlackRock's BUIDL fund settlement process:
"This eliminates traditional settlement delays (e.g., T+2) and counterparty risk."
And later:
"The instantaneous nature of transaction settlements mitigates the risk that a counterparty may fail to meet its obligations."
This claim that 'atomic settlement eliminates counterparty risk' appears in virtually every tokenization report, pitch deck, and conference presentation I encounter. It has become an article of faith when each of these documents relies on each other as a source.
It is, at best, a category error.
Atomic settlement, the simultaneous, irreversible exchange of assets, eliminates settlement risk. This is the risk that one party delivers while the other fails to, creating a timing mismatch. It's a real risk, and eliminating it is valuable.
But settlement risk is a narrow subset of counterparty risk. It says nothing about:
Pre-settlement credit exposure—the risk during the life of a trade before settlement occurs. If I've agreed to sell you an asset in three months at a fixed price, and the market moves 40% against you, you might not show up to settle. Atomic settlement doesn't help if you default before settlement day.
Mark-to-market exposure—unrealized profit and loss on open positions. If you're my counterparty on a perpetual swap and you're underwater by $50 million, that's counterparty risk I'm carrying right now, regardless of how settlement will eventually work.
Default risk—the fundamental question of who absorbs losses when someone can't pay. Atomic settlement assumes both parties have the assets to exchange. It has no answer for what happens when they don't.
For a money market fund subscription—exchanging USDC for BUIDL tokens—this distinction barely matters. Both parties have the assets, the exchange is immediate, done.
But the tokenization discourse routinely extrapolates from this narrow use case to make sweeping claims about eliminating counterparty risk generally. This is where the confusion becomes dangerous. It is dangerous because this is used to suggest the growth of these assets, when derivatives trading not spot trading dominates the markets.
The Word "Clearing" Is Doing Too Much Work
Part of the problem is that "clearing" means different things in different contexts, and the tokenization discourse conflates them freely.
In payments, clearing refers to the process of transmitting, reconciling, and confirming payment instructions before final settlement. When Visa "clears" a transaction, it's validating and routing information. The settlement, actual movement of money between banks, happens later, often in batches.
In securities, clearing is the post-trade, pre-settlement process of confirming trades and preparing for delivery versus payment. A clearing house matches buyers and sellers, confirms terms, and orchestrates the settlement process.
In derivatives, clearing is something else entirely. A derivatives clearing house, a Central Counterparty or CCP, interposes itself between buyer and seller, becoming the buyer to every seller and seller to every buyer. But that's just the beginning. The CCP then:
Collects initial margin from both parties based on potential future exposure
Marks positions to market continuously or daily
Issues variation margin calls to cover unrealized losses
Maintains default waterfalls specifying who pays when losses exceed collateral
Runs auctions and manages positions when counterparties default
Coordinates with regulators during market stress
This is not a settlement function. This is ongoing credit intermediation, the continuous management of credit risk throughout the life of a position. It requires judgment, governance, and human beings making decisions about whether to liquidate, how to auction, when to invoke emergency powers.
When tokenization reports mention "faster settlement & clearing" as if they're the same thing, they're glossing over a distinction that represents trillions of dollars of risk infrastructure built up over decades. You cannot smart-contract your way out of credit risk, I've tried the trust level required is off the charts.
The 24/7 Problem No One Is Solving
The tokenization narrative loves 24/7 markets. "Trade anytime, anywhere." "Global access." "No more waiting for business hours."
But who manages risk at 3am on a Sunday? Treasuries don't trade as spot, they trade as derivatives. Obviously this is hyperbole but less so than statements that think atomic settlement is clearing.
Traditional CCPs operate on business hours for good reasons. Margin calls require assessment. Defaults require coordinated responses. Auctions require participants. Regulatory consultation requires regulators to be awake. I do push to innovate here, real-time risk, securitised (and then tokenised) tail risk can push this forward but this is where we are today.
The standard answer is "smart contracts can automate it." But smart contracts cannot:
Make judgment calls about partial liquidations versus full close-outs
Negotiate with distressed counterparties who might cure a default if given hours rather than seconds
Coordinate with regulators during market stress to prevent cascading failures
Invoke emergency powers that require board decisions
Assess whether a price oracle is being manipulated versus reflecting genuine market moves
The big issue though is that bayside doesn't trust this level of complexity to smart contracts so we are dead before we try. Today anyway.
We have seen what happens when automated liquidation systems operate without judgment. The October 2025 crypto crash. The recurring "auto-deleveraging" events on perpetual swap exchanges. These aren't bugs in the automation, they're the automation working as designed, producing outcomes that no rational human would have chosen.
24/7 markets are coming whether we like it or not. But the infrastructure to manage credit risk in 24/7 markets, real infrastructure, not just faster settlement, doesn't exist yet. The existing FCMs are not a model for this. Reports that celebrate 24/7 capability without addressing this gap aren't advancing the conversation; they're obscuring the hard problem.
The Collateral Use Case Buried in the Footnotes
To its credit, the Amundi report includes a use case assessment table. Buried within it is an honest admission. The "transaction collateral" use case—using tokenized assets as margin in derivatives and lending—is rated "1-Theoretical."
The explanation: "Regulatory uncertainty around collateral acceptance; limited market infrastructure; valuation and liquidity risks in volatile markets."
This is the honest part. The infrastructure to use tokenized assets as derivatives collateral, with all the margin call, liquidation, and default management infrastructure that requires, doesn't exist. It's not a technology problem. It's an institutional and legal architecture problem that technology alone cannot solve. The entire CCP->FCM architecture isn't capable of supporting this in the short term.
But this honest assessment contradicts the report's broader narrative about atomic settlement eliminating counterparty risk. If atomic settlement solved the problem, the collateral use case would be further along. The fact that it isn't reveals the gap between settlement infrastructure and clearing infrastructure.
And collateral is not just a use case for the majority of assets, it is the use case. We bundle large companies into indices, trade these indices supported by derivatives, then use liquid assets as collateral, and trade these as derivatives as well. Add FX and that's practically the whole market. Everything else is a rounding error.
Why This Keeps Happening
These reports aren't written by stupid people. They're produced by competent research teams at sophisticated institutions. So why does the settlement/clearing confusion persist?
Part of it is scope limitation. The Amundi report explicitly states it covers "tokenization of fund shares" and "does not address directly the portfolio management of tokenized assets." Fair enough. But it then makes sweeping claims about counterparty risk that only hold within that narrow scope, without flagging the limitation.
Part of it is incentive structure. Reports that identify hard, unsolved problems aren't as marketable as reports that describe revolutionary transformations. "The third revolution in asset management" generates more press coverage than "significant infrastructure gaps remain unaddressed."
Part of it is that derivatives clearing is genuinely obscure. Most people in financial services have never thought deeply about what a CCP actually does, because they've never had to. The infrastructure works invisibly—until it doesn't.
And part of it is the AI-assisted writing that increasingly characterises these reports. The language of tokenization discourse has become formulaic: "leveraging blockchain technology," "enhancing transparency and efficiency," "revolutionary potential." These phrases can be assembled into apparently authoritative documents without requiring deep understanding of the subject matter. The result is a proliferation of content that looks sophisticated but doesn't advance knowledge. We all use AI now, it corrects flow, improves reading and generally makes the writing better, but it you don't know the subject before asking AI to help this is the junk that comes out. It is best to improve how you present and to check you not to draft from scratch.
What Actual Progress Would Look Like
I don't write this to dismiss tokenization. The technology is real, the efficiency gains in settlement are real, and the distribution thesis, reaching digital-native investors who want on-chain access, is commercially sound.
But if we're serious about extending tokenization beyond fund share transfers into derivatives and lending, where the real volume and revenue opportunity lies, we need to stop pretending that settlement infrastructure solves clearing problems.
Actual progress would involve:
Honest problem definition. Acknowledge that credit intermediation in 24/7 markets is an unsolved infrastructure problem, distinct from settlement efficiency.
Institutional design work. Who collects margin? Who issues margin calls? Who manages defaults? Who absorbs losses that exceed collateral? These aren't technology questions: they're governance questions that require legal entities, regulatory frameworks, and economic models.
Integration with existing infrastructure. The hybrid models, off-chain risk management with on-chain settlement, are more realistic than fully on-chain approaches, but they require solving interoperability problems that aren't glamorous enough to attract attention.
Regulatory engagement. CCPs are among the most heavily regulated entities in finance for good reason. Any entity performing CCP-like functions in tokenized markets will eventually face CCP-like regulation. Building infrastructure that assumes regulatory exemption is building on sand.
The Cost of Confusion
The tokenization wave is coming. Institutional capital is entering. Infrastructure decisions being made now will shape markets for decades.
If we build that infrastructure on a foundation of confused thinking, believing that faster settlement solves credit risk, that smart contracts can replace judgment, that 24/7 markets need no 24/7 risk management, we will discover the gaps at the worst possible moment. In a crisis. When it's too late to fix them.
We've been here before. The 2008 crisis revealed that the industry had confused liquidity with solvency, correlation with causation, and sophistication with understanding. The post-crisis reforms, mandatory clearing, margin requirements, resolution planning, were built on hard-won clarity about what risk actually means. They don't mean we need to freeze the choices about how we operate based on 2008 though.
Tokenization gives us a chance to build new infrastructure right the first time. But only if we're honest about what problems remain unsolved. Settlement is not clearing. Atomic DvP is not credit intermediation. And another 150-page report recycling the same misconceptions is not progress.
We can do better. We need to do better. The alternative is to learn the same lessons again, the hard way, with other people's money.