Three GatherHub reports this week describe the same shift in different forms. A Financial Stability Institute brief shows how regulation can isolate a stablecoin issuer without capturing every activity elsewhere in its corporate group. M0 lets several branded digital currencies share infrastructure, reserves or service providers. The Securities and Exchange Commission, meanwhile, proposes allowing distributed ledgers into the official record of US shareholders while requiring one transfer agent to retain control of that file.
The common change is not the disappearance of intermediaries. It is their separation into modules: brand, contract, reserve, validation, distribution, recordkeeping and custody can sit with different actors. That architecture can make some components replaceable and some data easier to inspect. It also creates a question the word “blockchain” cannot answer: who is accountable for the whole when one module fails, departs from the law or stops providing an exit?
Marc Norat’s original contribution is to compare the three cases on one accountability map. The result is less dramatic than a promise of disintermediation but more useful: technology distributes execution; customer protection still depends on an identifiable owner at every critical boundary.
Three developments, one layered architecture
The first case concerns legal scope. In its 27 August brief, the FSI compares stablecoin-issuer rules in the European Union, Hong Kong, Singapore, the United Kingdom and the United States. These frameworks generally restrict the issuer to a few core functions: issuing, redeeming and managing the reserve. The restrictions do not always capture a parent or sister company that provides lending, staking or cryptoasset custody.
The second case is technical and commercial. M0 supplies shared contracts and roles to stablecoins carrying other brands. An extension can define access, chains and the destination of reserve income; an issuer holds assets and mints units; validators attest to a value of offchain collateral. The modularity is real, but two visible tokens can still depend on the same base asset, issuer or liquidity channel.
The third case concerns the legal register for securities. In a 421-page proposal published on 1 September, the SEC seeks to modernize transfer-agent rules that have not been substantively overhauled in roughly four decades. A distributed ledger could form part of the official securityholder file. Yet one agent would retain exclusive control for each issuance, give the regulator independent access and remain accountable for accuracy, continuity and correction.
The cases involve different assets, jurisdictions and stages of maturity. Their shared feature is precise: each separates the surface used by the customer from the layer carrying accountability.
GatherHub’s accountability map
Comparing these architectures requires four consistent questions: which element becomes modular, which point remains indivisible, what evidence exists today and which failure has yet to be covered?
| Case | Visible or replaceable module | Accountability point | Observable evidence | Missing test | |---|---|---|---|---| | Stablecoin group | Issuance, custody, staking or lending split across companies | Reserve, redemption and supervision of intra-group links | Five frameworks compared by the FSI | Supervisor’s ability to contain a loss at an affiliate | | M0 | Brand, extension, chain, issuer and yield allocation | Reserve assets, validator signatures and redemption rights | $181.7m of M tracked on 4 September | Live issuer replacement without breaking parity or rights | | SEC register | Conventional database, cloud or distributed ledger | One agent controlling the official file | Detailed proposal, forms and obligations | Treatment of an immutable ledger the agent does not control alone |
The grid reveals a principle. A module can be technically substitutable without being legally interchangeable. Replacing a contract does not automatically carry holder rights with it. Changing validators does not move assets held at banks. Replicating a history across many nodes does not identify who can correct a disputed entry.
Modularity therefore reduces some forms of lock-in while multiplying interfaces. Every interface must identify the authority, available information, fallback procedure and party absorbing a loss. Without all four, separating functions relocates risk instead of reducing it.
Stablecoins expose the perimeter problem
The FSI brief we examined on Tuesday does not document reserve misuse. It identifies a possible contagion channel. A licensed subsidiary may hold conservative assets while an affiliated company takes more risk. A problem at the affiliate can reach the issuer through a shared provider, an intra-group claim, operational dependence or a loss of confidence.
Our reading of the US GENIUS Act confirms the textual limit. Section 4(a)(7) constrains the activities of the “permitted payment stablecoin issuer.” A supervisor may consider the issuer’s financial activities, including those of its own subsidiaries, but the statute does not explicitly establish consolidated supervision over every parent and sister company in a nonbank group. By contrast, the Financial Stability Board’s recommendations address the entire stablecoin “arrangement” and all of its functions.
The distinction is practical. A licence attached to one entity answers “who may issue?” Group supervision also asks “where can risk travel?” Structural innovation becomes useful only if supervisors can follow related-party transactions, common providers and continuity arrangements beyond the licensed subsidiary.
It would be wrong to infer that every affiliate is unregulated. Some are banks, custodians or service providers with their own licences. The important gap is the possible absence of a common view, not a universal absence of rules.
M0 shows modularity working — and concentrating
The critical M0 profile published on Wednesday lets us observe the architecture in production. Its documentation separates extension, distribution and issuance. It automates the mint ratio, the rate paid by issuers and the allocation of income to approved accounts. Audits cover the core, extensions and several multichain components.
The reserve does not become directly observable, however. Issuers report the value of offchain assets and authorized validators sign that figure. An independent 2024 review warned that core security depended in part on governance incentives and at least one honest validator. The high-severity signature-counting flaw found then was marked as fixed; it is not evidence of a current vulnerability. It does illustrate why a smart-contract audit does not certify issuer solvency or the speed of a bank redemption.
Independent data also qualify the growth story. DeFiLlama tracked $181.7 million of M at 00:00 UTC on 4 September, down from $294.2 million on 4 August: a $112.5 million or 38.2% decline. Ethereum carried $155.2 million, or 85.4% of the tracked supply. DeFiLlama also marks M as “doublecounted” because some of this base reappears through extensions.
Those figures do not prove commercial failure. They show why product count, underlying monetary supply and distribution by chain must remain separate measures. A modular infrastructure can add partners while its common base contracts. The structural test will be a real issuer change: unchanged rights, continuous liquidity, preserved parity and a clearly transferred responsibility.
The SEC permits distribution under single control
The SEC proposal examined on Thursday offers the clearest counterpoint. It does not try to spread accountability among nodes. It permits technical diversity while reconcentrating legal responsibility with the recordkeeping transfer agent.
That logic reflects a fundamental difference between technical history and property rights. A blockchain can make a sequence of transactions visible and difficult to alter. It does not always establish whether a token is itself the security, instructs an offchain register, represents a claim against a custodian, or merely reproduces financial performance.
The SEC therefore proposes counting issuer-supported tokenized securities separately from those created by third parties. Transfer agents would also identify tokenization providers, maintain access to records without a provider’s intervention and test continuity plans. The opening concerns the medium used for recordkeeping, not the dilution of regulatory responsibility.
The document remains a proposal, open to comment and subject to change. Its most consequential issue is unresolved: how should “exclusive control” work where the file exists solely on an immutable blockchain that the agent cannot control by itself? The answer will determine which networks can become part of an official register and under which permissions.
What actually changed this week
This week did not produce a financial system without a centre. It made the separation between technical components and economic obligations more explicit. Stablecoins can distribute functions across companies; M0 can separate issuance, brand, yield and validation; a transfer agent can distribute data across a ledger. In each case, the reserve, redemption or official file retains an accountability point.
That is progress, provided it is measured correctly. Chain, contract and partner counts are not enough. Market participants need disclosure of intra-group exposures, non-double-counted supply, redemption rights, exit timing, validator concentration and the entity authorized to correct the record.
The next milestones are observable. For stablecoins: related-party rules and consolidated supervision. For M0: an issuer migration executed under real conditions and extension-level data. For the SEC: the final control test and treatment of immutable ledgers.
The week’s conclusion is therefore simple: finance is becoming composable faster than it is becoming collectively accountable. Modularity improves the architecture. It protects the holder only when every boundary specifies who holds the assets, who honours the exit, who controls the register and who answers when those accounts diverge.