The legal layer used to be the part of tokenisation conversations that everyone deferred. Through 2024 and 2025 it became the part that everyone has to lead with. Part 4 sets out why finality is a 2026 priority, then walks through the half-dozen confusions that recur in every onboarding conversation a tokenised settlement system has with a TradFi treasury, a prudential supervisor, or a regulator's policy desk.
Why it matters now
Tokenisation has arrived at the point where the operational engineering is solved and the legal layer is the bottleneck. Through 2024 and 2025 the live infrastructure (designated central securities depositaries (CSDs), real-time gross settlement (RTGS) systems, central counterparties) and the tokenised infrastructure (consortium chains, regulator-led pilots, GSIB tokenised deposit platforms) have started running real flows alongside each other. Where they coexist, the finality question gets asked explicitly in every onboarding conversation. The BIS and CPMI are working through how Principles for Financial Market Infrastructures (PFMI) Principle 8 applies to tokenised infrastructure, and Hyun Song Shin's research on the unified ledger argues, in substance, that tokenisation pays off precisely when central bank money, commercial bank money, and tokenised assets share a single finality envelope rather than each carrying their own (BIS Annual Economic Report 2024 Ch III; Shin speech, 25 June 2024).
The Basel Committee under Pablo Hernández de Cos and now Ben Gully has set the prudential frame through the basel sco60 cryptoasset standard. The standard's classification conditions for Group 1a tokenised traditional assets include legal enforceability of the on-chain record, which is finality-adjacent (Basel SCO60). A bank seeking favourable capital treatment has to be able to demonstrate, on demand, that the chain entry is legally operative, which runs through the system's finality designation or its contractual basis. IOSCO has been developing parallel guidance for securities markets. The combined effect is that for the first time, an institution evaluating a tokenised settlement system can ask the finality question against a defined set of standards rather than against vibes.
The downstream consequence is operational. Where a bank has to demonstrate Group 1a treatment to its supervisor, the finality opinion is part of the demonstration. Where a treasurer wants to use a tokenised cash leg in atomic delivery versus payment, the finality opinion has to be in place before the trade desk gets approval. Where a CSD wants to extend an existing designation to a tokenised settlement layer, the conversation with the designating authority is about whether the chain integration changes the system's PFMI compliance and its insolvency posture. None of these conversations existed in coherent form three years ago. All of them are routine in 2026.
Common confusions
Operational finality is not legal finality. Block inclusion or transaction posting tells you the system has processed the transfer. It does not tell you whether the transfer is protected against insolvency clawback under the law of the system. The two questions are independent and need to be answered separately for any institutional settlement system.
Probabilistic finality is not legal finality. A high probability that a transfer will not be reorganised is not, in any major jurisdiction's law, equivalent to the transfer being final. Regulators do not give partial credit for very-likely-final settlement. The CPMI's tokenisation taxonomy is explicit on this point (BIS CPMI tokenisation report).
Designation is not certification. Designating a system gives it finality protection under the relevant law. It does not certify the operator's product, the soundness of every individual transfer, or the absence of operational failures. Designation is a legal hook, not a quality stamp.
Atomic does not mean final. Atomic delivery versus payment (DvP) coordinates the timing of two transfers. The transfers themselves are final or not based on their underlying settlement regimes, not on the atomicity of the coordination. Atomic settlement on top of two systems without finality gives you very fast simultaneous failure rather than slow asymmetric failure, which is a smaller category of progress than atomic-DvP marketing tends to suggest.
Smart contract irreversibility is not legal finality. A smart contract that cannot be reversed by ordinary chain operation is operationally robust. It is not, by virtue of its irreversibility, immune to insolvency clawback or court-ordered reversal in jurisdictions whose law would reach the underlying claim. The chain's resistance to roll-back is one input. The legal status of the system is another.
Finality on one leg does not give you finality on the trade. A bond settlement system can be designated under Settlement Finality Regulations (SFR) 1999 while the cash leg settles in a non-designated system or in a non-domestic currency through correspondent banking. The trade as a whole is only as final as its weakest leg.
The Settlement Finality Directive (SFD) and its national implementations do not all do the same thing. They have a common architecture, designation, insolvency override, netting protection, but the timing and scope details vary enough that a multi-jurisdictional tokenised settlement system needs a finality opinion under each system's law, not a single pan-EU or pan-APAC opinion. This is one of the reasons cross-border tokenised settlement is harder than the chain technology suggests.
A finality opinion is not a one-time artifact. When the system rules change, when the participant set changes, when the underlying chain technology changes, the opinion has to be refreshed. The Uniform Commercial Code (UCC) Article 4A and Federal Reserve Regulation J framework in the US is unusually stable across operational changes. Most other regimes require a fresh look when the system materially changes. This is part of why operators of tokenised settlement systems prefer architectures that look like an extension of an existing designated system rather than a new system seeking new designation.