The previous four parts made the case for tokenisation. This one is the honest pushback. The failure modes below show up in pilots, in regulator briefings, and in client meetings often enough that any operator who cannot articulate them lacks credibility. The right read of tokenisation is not as a universal upgrade but as a particular set of trade-offs against the existing rails.
It does not solve credit risk on the underlying
A tokenised deposit at Bank A carries Bank A's credit risk, full stop. The chain entry does not transmute the unsecured loan from depositor to bank into a secured claim. A tokenised bond carries the issuer's credit risk; a tokenised private-credit interest carries the underlying borrower's; a tokenised real-estate position carries the property's vacancy and impairment risk. The chain records who owns the claim. It is not a substitute for an analysis of whether the claim is good.
This sounds obvious until it shows up in the wrong meeting. A treasury team comparing a USD deposit at a regulated bank to a USD-pegged stablecoin issued by a non-bank entity is comparing two instruments with materially different credit profiles, even if the chart shape near par is identical. Chapter II on stablecoin types takes the comparison more carefully. The November 2025 contagion across DeFi (decentralised finance) curated vaults was, at its core, a credit event (see the curation-layer DeFi theme).
It does not fix the legal-control question if the jurisdiction has not figured it out
The whole framing rests on the chain entry being the legally operative record. Chapter I, the legal-control question sets out why this matters and how the four major financial jurisdictions are converging on a "control" or "possession-equivalent" test. The convergence is real but uneven. The US has UCC Article 12 amendments enacted across most states; English law has the Property (Digital Assets etc) Bill in flight; Singapore has a workable overlay through the SFA and trust principles; Hong Kong is the least codified.
A jurisdiction without a clear answer leaves the issuer carrying legal-opinion risk that an issuer in a more codified jurisdiction is not. A tokenised bond issued under English law, held by a Hong Kong custodian, traded on a Singapore venue will need three legal opinions, each carrying a residual ambiguity a stress event could surface. The CPMI tokenisation taxonomy is explicit that the legal envelope is the load-bearing piece.
It does not eliminate operational risk; it relocates it
Banks have spent decades building operational-risk machinery for the post-trade stack: reconciliation teams, exception-handling workflows, BCP, messaging redundancy. A tokenised system inherits some of this and replaces some with new failure modes: smart-contract bugs, validator failures, key-management failures, oracle failures. Basel SCO60 prices this with a 2.5% infrastructure-risk add-on for Group 1a tokenised exposures. That is the regulator's quantification of the residual.
The 2022 Binance Bridge hack, the 2024 Wormhole issues, and various smaller exploits across L2s are reminders that the supply chain of a tokenised asset on a public chain includes the bridge operator, validators, the smart-contract developer, the wallet provider, and the custodian, each a potential point of failure. The off-chain equivalent has a smaller supply chain, with the trade-off being the absence of the capabilities Part 1 listed.
"Tokenisation does not abolish operational risk. It moves it from places where we have spent thirty years learning how to control it to places where we have spent five years learning how to control it."
Adapted phrasing reflecting the framing routinely used in central-bank financial-stability reviews; see BIS CPMI tokenisation report for the formal version.
It does not automatically improve liquidity
Tokenisation moves an asset onto a programmable ledger. It does not, by that act, create a secondary market. A tokenised private-credit interest is still a private-credit interest. A tokenised bond on a permissioned chain with no active secondary venue is, for practical purposes, less liquid than the same bond on a designated bond-trading platform.
Franklin Templeton's peer-to-peer transfer feature for FOBXX, enabled in 2024, illustrates the point. The technical capability to transfer a share between two Benji-onboarded holders does not, by itself, create a deep secondary market. The market exists or it does not. On-chain composability is a property of the chain, not of the asset, and a deep secondary market still requires the usual ingredients (multiple bidders, willing sellers, neutral price discovery, designated market makers).
It does not bypass regulation; it generally adds another layer
The most consistent misread among new entrants is that putting an asset on a public chain routes around the regulator. For any institutional issuer, the regulator now supervises both the off-chain wrapper and the on-chain mechanism. MiCA imposes both an issuer regime and a custody regime; the GENIUS Act federalises a payment-stablecoin licensing layer on top of state money-transmitter rules; the HK Stablecoins Ordinance creates a new HKMA perimeter; Japan's PSA amendments add an FSA registration layer for fund-transfer service providers. These are net-additional supervisory burdens with a clearer perimeter as the trade-off, not regulatory holidays.
What to read next
If the headline answer in Part 1 surprised you, Chapter I is the right next stop. If the architectural choice in Part 3 is the question that surfaced, Chapter V takes it deeper. If the cross-border story in Part 4 mattered, Chapter IV on settlement finality explains why the legal layer is what does the work, and Chapter VI on atomic DvP takes the operational property that ties the two legs together. Chapters II (stablecoins), III (bank money), VII (tokenised deposits), VIII (wholesale CBDC), and IX (agentic commerce on tokenised rails) build out from this base.