The single most common confusion in tokenisation marketing is that fast settlement and atomic settlement are the same thing. They are not, and the conflation costs operators credibility in the first counterparty conversation. This part isolates the two properties, explains why atomicity (not speed) is the structurally interesting one, and frames the chapter that follows: DvP, PvP, the precise vocabulary of T+N, and the capital arithmetic that makes intraday settlement economically meaningful.
Two properties, often conflated
Speed is a duration. It is how long the workflow takes from trade execution to final settlement. Seconds, minutes, hours, days. Speed is observable on a stopwatch and is the metric that shows up in marketing decks because it is easy to demonstrate.
Atomicity is a structural property. It says that settlement is all-or-nothing: either both legs of the trade complete together, or neither does. There is no observable interval in which one leg has settled and the other has not. Atomicity is invisible until something goes wrong, at which point it is the only property that matters.
A 10-second settlement can be non-atomic. A T+3 settlement can be atomic. The two axes are independent. A US treasury settling T+1 through Fedwire securities and Fedwire funds is atomic in the regulated sense even though it takes a day, because the linkage between the two legs is enforced by the rule book of the system. A wire transfer hop between two custodians, completed in seconds, is not atomic if the asset booking and the cash booking are separate operations in separate ledgers.
The legal significance of atomicity
The reason atomicity matters is the settlement window. In a non-atomic transaction, there is a period (sometimes seconds, sometimes hours, historically days) during which one party has performed and the other has not. One side has the asset but has not paid; or has paid but does not yet have the asset. That window is the definition of settlement risk.
Settlement risk has a credit dimension and a liquidity dimension. The credit dimension is the obvious one: if the counterparty defaults inside the window, the performing party loses the open exposure. The liquidity dimension is subtler: even with no default, the performing party has tied up either an asset or cash for the duration of the window, and that capital cannot be redeployed. Both dimensions price into the cost of doing business.
Atomic settlement closes the window. There is no interval in which one party has performed and the other has not, because the protocol that updates one balance is the same protocol that updates the other, in a single indivisible operation. The credit exposure goes to zero in the settlement window. The liquidity drag is reduced to whatever is needed to fund the gross legs at the moment of settlement.
For a treasury PM or a clearing operations head, this is a material change in the credit profile of every counterparty interaction. The pre-tokenisation answer to settlement risk was institutional: build a central counterparty (CCP) that interposes itself between buyer and seller, mutualising the risk through margin and a default fund. The post-tokenisation answer is structural: design the protocol so that the risk window does not exist in the first place.
The CCP analogy
Cleared derivatives use a CCP precisely to eliminate settlement risk between counterparties. When two banks trade an interest rate swap through LCH or CME Clearing, the CCP novates the trade: the original bilateral contract is replaced by two contracts, each between one of the original parties and the CCP. Now neither party has counterparty risk to the other; both have risk only to the CCP, which is over-collateralised through initial and variation margin and backed by a default fund.
The CCP is an institutional solution to a structural problem. It works, but it is expensive: every member posts margin, contributes to the default fund, and pays clearing fees. The CCP itself is a systemically important entity that requires its own supervisory regime and its own resolution planning.
Atomic on-chain DvP achieves the same risk-elimination outcome through a different route. The smart contract is the indivisible operation: it either performs both legs or neither. There is no counterparty risk in the settlement window because there is no settlement window. No CCP is required for that specific risk, because the protocol structurally cannot leave one leg unsettled.
This is not a claim that tokenised rails make CCPs obsolete. CCPs do other things (multilateral netting across many trades, default management, liquidity provision in stress events) that atomic DvP does not address. The narrower claim is that the specific function of eliminating settlement-window risk between two counterparties can be achieved structurally on a tokenised rail without the institutional overhead of a CCP. For derivatives, that probably means CCPs evolve. For cash-securities markets, where there is no equivalent CCP layer, it means a category of settlement risk simply disappears.
What this chapter covers
The rest of the chapter takes this distinction seriously. Part 2 covers DvP and the BIS three-model taxonomy, the framework regulators use to classify settlement designs. Part 3 extends the same logic to FX through PvP and the post-Herstatt history. Part 4 fixes the operational vocabulary so that "real-time" and "atomic" and "T+0" stop getting used interchangeably. Part 5 grounds the abstract argument in the capital arithmetic that explains why intraday settlement actually matters to a treasury function: a worked example of futures clearing collateral and the published cost-reduction figures from production tokenised rails.