On 1 September 2026 the US Securities and Exchange Commission proposed a rewrite of its transfer-agent rules — the plumbing that decides who is written down as the owner of a share. The proposing release (34-106246) does something I did not expect to read this decade: it would expressly permit a transfer agent to use distributed ledger technology as part of the master securityholder file, or as the entire file.

The master securityholder file is the authoritative list of registered owners. Not a mirror of it. The list.

Meanwhile, according to Solana’s own institutional RWA reporting, 97% of all on-chain tokenised-equities spot volume to date had settled on Solana as of late July 2026. Almost none of that volume is a share. It is a claim on a company that holds a share, and the difference is the whole subject of this piece.

Why the distinction matters now

Until this year you could treat “tokenised equity” as one category, because only one version existed in any volume: an offshore wrapper. That is no longer true. Three structurally different things now share the name, and in 2026 all three acquired regulatory momentum within about eight months of each other. If you cannot tell them apart, you cannot answer the only question that matters when something goes wrong: what do you actually hold a claim against?

First principles: ownership is a record, not a token

Share ownership in most markets is a hierarchy of records. At the top is the issuer’s register, maintained by a transfer agent, which names the registered holder. In the US that name is usually a depository nominee rather than you. Below it sits your broker’s book entry, which gives you a security entitlement — a property interest in your broker’s position, not title to a specific share. Below that, if you have bought synthetic exposure, sits a contract with somebody who promises to track a price.

Tokenisation does not collapse this hierarchy. It picks a layer and puts it on a ledger. Which layer it picks is the design decision that everything else follows from, and it is almost never stated in the marketing.

The three layers, with numbers

Say you hold what a screen calls 100 tokenised shares of a company trading at $50. Same ticker, same price, same wallet. Now the entity you bought from fails.

Layer one: the wrapper. Kraken’s xStocks programme, issued by Backed and settled on Solana for non-US retail, is documented in its own product legal overview as a bearer debt instrument — a tracker certificate giving economic exposure. It confers no shareholder voting rights. The documentation is candid that collateral may not always consist of the underlying shares. So your $5,000 is an unsecured claim in an issuer’s insolvency, and you queue with other creditors. You never appear on the company’s register, because you were never meant to.

Layer two: the tokenised entitlement. In December 2025 the SEC’s Division of Trading and Markets granted the Depository Trust Company no-action relief for a three-year pilot tokenising security entitlements in certain highly liquid securities. On 18 March 2026 the Commission approved a Nasdaq rule change (SR-NASDAQ-2025-072) letting the exchange trade certain securities in tokenised form, building on that pilot. Here the token represents the entitlement, so your $5,000 sits inside the customer-property regime rather than the creditor queue, and your votes travel up the intermediary chain as they always did. The register still names the depository nominee. The chain has become a faster way to move an entitlement, not a new kind of ownership.

Layer three: the register itself. This is what the September proposal contemplates. If a transfer agent’s master securityholder file is maintained on a ledger, the token is not evidence of a claim on an intermediary — the entry is the record of registered ownership. Your $5,000 is not exposed to the failure of a wrapper issuer or a broker, because neither is standing between you and the register. That is a genuinely different instrument from layers one and two, and it is the only one where “the blockchain is the source of truth” is a legal statement rather than a slogan.

Three tokens. One screen price. Three completely different answers to a single insolvency.

The mechanics are further along than the law

What strikes me, reading the proposal against the technology, is that the technical primitives for layer three already exist and have for a while. Token-2022 on Solana supports transfer hooks that enforce allowlists and per-wallet eligibility, so only verified participants can hold or move a restricted asset, and confidential-balance extensions that keep positions private while remaining auditable to authorised parties. Those are recognisably the controls a registrar needs: restricted transfer, eligibility, auditability, legend enforcement. The proposal even contemplates tightening restrictive legends, possibly through smart contracts.

So the constraint was never cryptographic. It was that nobody had authority to treat the ledger as the book of record. That is what a rule change can grant and a library cannot.

What the proposal does not do

It is a proposal. It was published in the Federal Register on 4 September 2026 with a 60-day comment period, which puts the close in early November. Nothing is permitted yet. It applies to the roughly 273 registered transfer agents in the US and to nobody else, and technology-neutral drafting is not an endorsement: the release is explicit that a transfer agent using DLT in its master file presents different risks from a traditional mutual fund agent, and it would add Form TA-2 reporting on DLT use plus expanded safeguarding and cybersecurity obligations. Permission arrives attached to supervision.

The staff had already said the obvious thing in a joint statement in January 2026: tokenised stock remains an equity security, and offers and sales remain subject to registration or an available exemption. The Division of Trading and Markets keeps the same position in its FAQ on crypto asset activities and distributed ledger technology. Putting a share on a ledger changes its plumbing, not its legal character.

The strongest argument against my framing

Layered claims are not a crypto pathology; they are how securities markets already work. An American Depositary Receipt is a wrapper. A nominee account is an entitlement. Investors have held both for decades, accepted the intermediary risk knowingly, and been served well by intermediaries who handle corporate actions, tax and recovery competently. On that view, tokenised wrappers are simply an old structure with new settlement, and disclosed clearly if you read the prospectus — xStocks’ EU distribution runs on a base prospectus approved by Liechtenstein’s FMA under the EU Prospectus Regulation, which is more disclosure than most retail products carry.

I accept most of that. My objection is narrower: the disclosure is accurate and the distribution is not. A product documented as a bearer tracker certificate is presented on a trading screen as a company’s ticker at the company’s price. The gap between the legal document and the buying experience is where retail harm accumulates, and no amount of prospectus rigour closes it.

I also want to be careful about the pace. The Commission is reported to have proposed repealing Reg NMS Rule 611 and Rule 610(e) in June 2026, which commentators read as removing a structural barrier to on-chain equity trading, since an automated market maker cannot satisfy Rule 611’s trade-through protection by construction. But the separate innovation exemption for tokenised stocks has been reported as delayed in May and cancelled in August. I have not yet read those two items in the primary releases, so treat both as reported rather than verified, and treat analyst expectations of a finalised repeal in early 2027 as expectations rather than a timetable.

Where this leaves Ghana

Ghana’s Virtual Asset Service Providers Act, 2025 (Act 1154) was passed by Parliament on 19 December 2025 and signed on 30 December, with oversight split on activity lines between the Bank of Ghana and the SEC. The Bank has created a Virtual Assets Department, admitted six firms to a one-year sandbox, and said in July 2026 that implementing guidelines were well advanced. The SEC is piloting gold-backed and tokenised gold products with the Ghana Gold Board and plans to license independent custodians for them.

Notice which layer that framework addresses. Act 1154 licenses intermediaries — the people who hold and move virtual assets on your behalf. Starting with gold under independent custody is a sensible choice precisely because a commodity claim can be made honest with a custodian and an audit. Equities are harder, because the authoritative record does not belong to a VASP. Trades on the Ghana Stock Exchange settle through a central securities depository on a T+3 basis, and the register is the depository’s and the registrar’s business, not a licensee’s.

Which means a tokenised GSE share, if anyone builds one before the equity-record question is settled, will be a layer one wrapper by default. Not by intent — by the absence of a rule that lets it be anything else. That seems worth saying now, while the guidelines are still being drafted, rather than after the first product launches.

What I currently believe

I think layer three is the only version of tokenised equity that changes anything structurally, and that most of what is called tokenised equity today is layer one wearing layer three’s language. I think the SEC proposal is more consequential than the Reg NMS headlines it was buried under, because a settlement-speed change is an efficiency and a book-of-record change is a change in what ownership is.

What I have not worked out: whether a register on a public ledger can satisfy the safeguarding and cybersecurity obligations the same proposal imposes without being permissioned so heavily that it stops resembling a public chain. I also do not know how corporate actions, transfer-agent liability and error correction work when the record is append-only and the mistake is already final. I am reading the release properly and will write about that separately.

And I would treat any figure for the size of this market with suspicion. Reported tokenised-equity totals for July 2026 range from a few hundred million dollars to billions depending on whether wrappers, funds and treasuries are counted, and projections like the Citi Institute base case of $5.5 trillion by 2030 do not publish a method I can check. The distinction between the three layers is verifiable from primary documents. The market size is not.