I keep coming back to a slightly awkward question. If a fund share is sitting on a public blockchain, but the official ownership book still lives in a back-office system that predates most of the people trading that share, what exactly got tokenized? The wrapper, or the record? That distinction sounded academic until a chief executive of a large asset manager stood on a Singapore stage and said, in effect, that a lot of the industry is still selling the wrapper.
Jenny Johnson, chief executive of Franklin Templeton, used a panel at TOKEN2049 to draw a hard line between products that look tokenized and products whose books and records actually run on a chain. She called many rival structures digital twins. The phrase is not new inside her firm. It is, though, unusually blunt coming from a conference stage shared with a major exchange co-chief and a network chief executive, on a session billed around tokenized assets, liquidity, and onchain settlement.
Perhaps the most interesting aspect is not the insult. It is the test she is proposing. A token that merely mirrors a position held somewhere else is a receipt. A token that is the position, because the official shareholder file is built from blockchain records, is a different animal. Investors, treasurers, and competing product teams are going to have to decide which animal they are actually buying.
What the Digital Twin Critique Really Means
A digital twin, in the way Franklin Templeton has described it in its own research, is a token linked to an ownership record that stays offchain. The chain shows a representation. The legal shareholder file sits in a conventional system. Move the token and you may still need a transfer agent, a custodian, or an internal ledger to catch up before anyone treats the move as final.
A digitally native model flips that order. Official transaction and ownership information lives inside a blockchain-integrated system. The chain is not a billboard for a database elsewhere. It is part of the book.
I’ve found that people hear “onchain” and assume the entire shareholder register is public, names and all. That is not how a regulated U.S. fund works, and it is not how Franklin describes BENJI either. Privacy and official status can coexist. The harder question is which system wins when the two disagree.
A stage built for this argument
The setting mattered. TOKEN2049’s agenda placed Johnson alongside Binance co-chief executive Richard Teng and Canton Network chief executive Yuval Rooz. That is a strange trio if you still think tokenization is a crypto-only hobby. An asset manager, a global exchange, and a network aimed at institutional settlement, all talking about the same plumbing.
Conference remarks travel fast and lose precision even faster. The early write-up of her comments framed BENJI transactions at about $1.13 against roughly $150 through traditional fund processing. Hold that number. A more detailed comparison she gave months earlier does not say the same thing. We will get there, because the cost story is where a good slogan can outrun the spreadsheet.
A token linked to an offchain book is a mirror. A token that is the official share record is the thing being mirrored.
That is the claim, stripped of conference lighting. Everything else, yield clocks, collateral lines, acquisition currency, is downstream of it.
Why rivals bristle at the label
Calling someone else’s product a twin is a marketing move as much as an engineering critique. Plenty of managers have good reasons to keep the legal record where regulators, auditors, and transfer agents already know how to examine it. A token on top can still speed distribution, improve collateral mobility, or let a client see a position without waiting on a nightly file.
Fair. And still not the same product. If your operations team can reverse a wallet transfer only by editing the offchain book, the chain did not settle the fund. It advertised it. Johnson is betting that buyers will eventually price that difference, the way they price share classes and custody arrangements today.
How BENJI was built, and what “official” actually covers
BENJI traces back to 2021 and the Franklin OnChain U.S. Government Money Fund, known by the ticker FOBXX. Franklin presents it as the first U.S.-registered mutual fund to use a public blockchain as its official system of record for processing transactions and recording share ownership. It started on Stellar. It has since been deployed across a long list of public networks.
The current Benji site lists Stellar, Polygon, Arbitrum, Avalanche, Aptos, Ethereum, Base, Solana, and BNB Smart Chain. Retail access is not uniform. Some networks are institutional routes. That spread is easy to mock as chain-collecting. It is also a practical answer to a client base that does not live on one ledger.
The structure is more careful than a public cap table with names attached. An August 12 staff letter from the U.S. Securities and Exchange Commission describes a split. Franklin Templeton Investor Services keeps an internal system with private shareholder information. Blockchain records carry anonymous transaction data. The two are connected in real time to form the fund’s official shareholder file.
So the chain is inside the official system. It is not the whole official system. That nuance gets lost in slogans, and it is the nuance that makes the model regulatable.
The transfer agent still holds the pen
Blockchain maximalists sometimes hear “system of record” and picture an immutable register nobody can touch. The staff letter says otherwise. The transfer agent retains administrative controls. It can correct unauthorized transactions or errors, freeze or migrate wallet records, and restore the official record when necessary.
In my experience, that sentence is where serious allocators relax and where purists get annoyed. Both reactions are reasonable. A money-market fund cannot shrug at a compromised wallet the way a meme coin can. Control is the feature, not the bug, if the alternative is a frozen shareholder file and a lawyer on the phone at 2 a.m.
- Private identity data stays in the transfer agent’s internal system.
- Anonymous transaction data sits on the public chain.
- The two records reconcile in real time into the official shareholder file.
- The transfer agent can correct, freeze, migrate, or restore wallet records.
- Blockchain entries are part of the official system, not a side display.
Read that list twice if you are comparing products. A rival can put a token on the same chains and still fail every line except the second.
The cost comparison that needs a slower reading
Johnson used processing cost as part of the Singapore argument. The first conference report put BENJI at $1.13 against roughly $150 per transaction on traditional rails. That gap is the kind of number that gets screenshotted. It is also the kind of number that should make you ask what is being counted.
An earlier, more detailed statement points somewhere else. On Franklin Resources’ January 2026 earnings call, Johnson said the firm had compared roughly 50,000 transactions across its legacy transfer-agency system and the Stellar blockchain. According to that call, the traditional system cost roughly $1.50 per transaction. Running all 50,000 transactions on Stellar cost approximately $1.13 in total.
Sit with that. One dollar and thirteen cents for the batch, not per share movement. Against a dollar and a half each on the old system, the chain looks almost absurdly cheap. It does not, by itself, support a $150 legacy cost per transaction. Unless Franklin later publishes a separate methodology for the conference figure, the $150 number should be treated as unverified stage shorthand.
I would rather have the smaller, documented gap than the cinematic one. A dollar-fifty versus a fraction of a cent, at scale, is already a business case. Inflating it to $150 invites a skeptic to throw out the whole slide.
| Comparison | What was said | How to treat it |
| Conference report | About $1.13 versus roughly $150 | Unverified until a method is published |
| January 2026 earnings call | About $1.50 each on the legacy system | Per-transaction legacy figure |
| Same call, Stellar batch | About $1.13 total for roughly 50,000 transactions | Batch chain cost, not a per-trade fee |
| Useful takeaway | Chain processing can be dramatically cheaper | Quote the call, not the unverified gap |
Cost is also not only gas. Transfer agency includes exceptions, anti-money-laundering reviews, statement production, and the unglamorous work of fixing a wrong account number. A chain that makes the happy path nearly free can still leave the exception path expensive. Anyone modeling a full conversion should price both.
Second-by-second yield is the sleeper feature
The yield claim is documented on its own timeline, separate from the cost spat. In June 2025 Franklin announced an Intraday Yield feature. The system calculates proportional yield down to the second when a tokenized security moves between investors. Yield can then be distributed each calendar day, including weekends and holidays.
Traditional money-market funds are not built for that clock. Shares often earn on a daily accrual with cutoff times that feel prehistoric next to a stablecoin moving at 11:40 p.m. on a Sunday. If you are a treasury desk parking cash between trades, a weekend is not a rounding error. It is idle yield.
Does second-level math change the annualized rate in a way a retail saver will notice? Often, no. Does it change fairness when a large block moves at 9:01 a.m. versus 4:59 p.m.? Yes. The seller stops earning the moment the share leaves. The buyer starts. No overnight argument about who owned the accrual.
Intraday yield, in plain language: Share moves at a timestamp Accrual splits to the second Distribution can hit every calendar day Weekends and holidays are not dark
That is a product feature competitors can copy only if their recordkeeping can see the move in time. A twin that updates overnight cannot pay to the second without inventing a second book. And once you have invented the second book, you are halfway to the model Johnson is defending.
Platform size is not the same as one fund’s size
Numbers in this market get mashed together. Earlier tracking put the BENJI platform near $2.5 billion in tokenized assets. On October 8, RWA.xyz listed Franklin Templeton Benji Investments at about $2.60 billion in distributed asset value, across four products, and ranked the platform second among tokenized U.S. Treasury fund platforms by that measure.
Do not confuse the platform with the standalone U.S. fund. The same tracker put BENJI itself near $760.6 million and iBENJI near $1.71 billion. Franklin’s conventional fund page reported FOBXX net assets of $686.64 million as of August 31. Dates differ. Definitions differ. A headline that says “the fund is $2.6 billion” is blending sleeves.
Still, second place on a treasury-token leaderboard is not a pilot. It is a franchise. Scale is what lets a collateral desk, an exchange, and an affiliated-fund program all point at the same share without inventing a bespoke wrapper each time.
Collateral is where tokenization stops being a demo
Franklin has been pushing those shares into institutional trading arrangements. One route lets eligible institutions pledge fund shares on Bybit while drawing USDT or USDC trading credit. An earlier Binance arrangement lets eligible institutions use Benji-issued money-market shares as off-exchange collateral, with the assets remaining in regulated custody.
MoonPay Trade is another path. Eligible clients can move between supported stablecoins and Franklin tokenized money-market exposure inside that institutional trading setup. The pattern is consistent. The share stays a regulated fund share. The token makes it legible to venues that already speak stablecoin.
This is the use case that makes the twin-versus-native argument concrete. A collateral desk wants to know, quickly, that the pledge is the official share and that a transfer agent will not quietly disagree with the wallet. If the token is only a mirror, the desk has to trust a reconciliation file. If the token is inside the official file, the desk is closer to treating the chain state as evidence.
Collateral does not care about your white paper. It cares whether the thing pledged is the thing owned.
A blunt way to read the institutional use case
None of this removes fund risk, issuer risk, or the chance that a venue’s credit line is mispriced. It removes a layer of “whose spreadsheet wins” risk. That layer is boring until it is the reason a margin call stalls.
Staff relief is not a commission blessing
August brought a narrower regulatory opening. SEC staff granted no-action relief for arrangements in which affiliated Franklin funds could invest in BENJI while Franklin Templeton Investor Services acts as custodian for the shares. The relief sits on controls: authorization, separate blockchain wallets, daily reconciliation, recordkeeping, and checks by an independent accountant.
Staff stressed what these letters always stress. The position is an enforcement stance. It has no legal force as a rule. It is not Commission approval of the arrangement. Anyone treating a no-action letter as a green light for the whole industry is reading a document that says the opposite on its face.
Even so, the letter matters inside the complex. Affiliated funds getting a supervised path onto the tokenized sleeve is how a manager proves the product to itself before it asks the outside world to rebuild operations around it. Internal adoption is a quieter signal than a conference quote, and often a better one.
- Authorization controls on who can move shares.
- Separate blockchain wallets rather than a pooled blur.
- Daily reconciliation between systems.
- Recordkeeping that an examiner can follow.
- Independent accountant checks on the arrangement.
Those five items are a template rivals will study even if they dislike the digital-twin label. Regulators tend to bless process, not branding.
When the token became deal currency
Franklin has also used BENJI outside subscriptions and collateral. When it announced the acquisition of crypto investment manager 250 Digital in April, it said BENJI tokens would form part of the payment consideration. The deal closed on June 22. Franklin Crypto was stood up with the acquired investment team and liquid crypto strategies.
Paying with your own tokenized fund share is a small sentence with a large implication. The seller had to accept that the token was consideration, not a voucher to be swapped back into a wire before the signature page. That does not make BENJI a currency. It does show an internal willingness to let the share travel as value between sophisticated parties.
Deal currency is a high bar. Lawyers price uncertainty. If the consideration had been a loosely mirrored token with a fuzzy shareholder file, I doubt it would have survived the purchase agreement without a pile of escrow conditions. The fact that it was used is not proof of perfection. It is evidence that Franklin’s own counsel could describe the instrument.
A network bug that should not be pinned on the fund
One claim around the original Stellar deployment needs a separate box. The Stellar Development Foundation disclosed a network state-archival bug discovered in October 2025. It had corrupted 478 data entries. Most were repaired. Eighty-four still needed mitigation by affected protocols or issuers. The foundation said the problem was contained by October 10 and resolved by October 23.
The official post-mortem does not identify Franklin Templeton or BENJI among the affected entries. The 478 corrupted records cannot be attributed to BENJI from the primary account available. The foundation said it responded by tightening monitoring, validator coordination, testing, code review, and work with outside security auditors.
Why mention it at all? Because public-chain recordkeeping inherits public-chain failure modes. A fund that chooses that path owes investors a plain description of what happens when the chain hiccups, who can restore state, and how the transfer agent’s pen interacts with a corrupted entry. Silence here would be worse than a careful non-attribution.
Multi-chain deployment cuts both ways. A bug on one network is not a bug on nine. It is also nine sets of client software, nine operational runbooks, and nine chances for a wallet integration to drift. Breadth is a feature for clients. It is a cost center for the control team.
What a buyer should ask before calling a fund tokenized
If you allocate to these products, or you build them, the Singapore critique is a due-diligence checklist wearing a headline. Skip the tribal language. Ask operational questions.
Where is the official shareholder file produced? If the answer is “our transfer agent system, and the token reflects it,” you own a twin, which may still be fine for your mandate. If the answer is “the official file is assembled from chain records plus a private identity system, reconciled in real time,” you are closer to the native model. Get the reconciliation timing in writing.
Who can freeze, migrate, or restore a wallet record, and under what notice? A fund with no admin key is a science project. A fund whose admin key is undocumented is a surprise. You want the key, the policy, and the audit trail.
How is yield cut when a share moves mid-day? Daily accrual with a 4 p.m. convention is a legitimate product. It is not intraday. Do not pay intraday prices for end-of-day math.
What does a transaction actually cost once exceptions are included? Quote batch chain fees if you like. Also quote the fully loaded transfer-agency cost for a broken instruction. The earnings-call comparison is the honest starting point, not a conference ratio nobody has footnoted.
Can the share be pledged, and does the pledge survive the venue’s custody model? Off-exchange collateral that stays in regulated custody is a different risk from a token sitting in a hot wallet at a trading firm. Read the custody sentence before you read the yield.
Liquidity, settlement, and the panel that framed the fight
The TOKEN2049 session was not a product demo. It was framed around tokenized assets, liquidity, and onchain settlement. Those three words pull in different directions. Liquidity wants a deep secondary market. Settlement wants finality. Tokenized assets want a legal object that both can point at.
An exchange co-chief cares about whether a money-market token can sit beside stablecoin credit without creating a hole in the margin system. A network chief executive cares about whether institutions will move real books onto shared rails. An asset-manager chief executive cares about whether the share she issues remains a share after it leaves her transfer agency and enters someone else’s wallet stack.
Johnson’s twin critique is a settlement argument dressed as a product argument. If ownership finality is offchain, onchain settlement is a suggestion. The suggestion can still be useful. It is not the same as a primary record.
I’ve watched this debate cycle through other assets. Warehouse receipts, dematerialized securities, central securities depositories. Every generation thinks the new ledger replaces the old one, then discovers the old one was also a bundle of legal rights, tax lots, and exception handling. Blockchain does not repeal that bundle. It offers a new place to store part of it.
Money-market funds are the awkward perfect test
Why start with a government money fund instead of a flashier strategy? Because the asset is boring on purpose. Investors already understand T-bill yield, stable net asset value expectations, and the idea of cash parking. If tokenization cannot improve that product, it will struggle to improve a complex one.
FOBXX is also a regulated mutual fund, which drags the experiment into transfer-agent rules, custody rules, and disclosure rules that a offshore wrapper can dodge. That drag is the point. A model that only works outside the U.S. fund code is a different business. Franklin is trying to show the code can bend around a public chain without snapping.
The August 31 net-asset figure and the October onchain platform figure will keep diverging as sleeves, share classes, and international wrappers multiply. Treat platform AUM and fund AUM as cousins, not twins. Ironic wording, given the speech, but accurate.
Where rivals can still win without copying the book
A fair reading leaves room for other designs. Some clients want a token that never leaves a permissioned environment. Some regulators will prefer a central securities depository that anchors a chain rather than a chain that anchors the depository. Some distribution desks will choose the twin because their existing transfer agency contract has three years left and ripping it out is not the highest-return project on the list.
Distribution can beat architecture for a while. A beautifully native fund that no platform will list loses to a mirrored fund that a treasury portal already supports. Johnson’s bet is that architecture compounds. Portals can be rebuilt. A shareholder file is harder to migrate once millions of daily accruals depend on it.
There is also a branding risk in the critique itself. If every competitor hears “digital twin” as a slur, they will stop sharing operating data. The industry needs more published cost studies like the 50,000-transaction comparison, not fewer. A single manager marking its own homework is not a market standard.
A practical way to read the next headline
The next firm to announce a tokenized treasury fund will lead with chain logos and a yield quote. Before you forward the release, run a short filter.
- Is the blockchain the official transaction record, or a mirror of one?
- Who holds private shareholder data, and how fast does it join chain data?
- What admin controls exist, and who audits their use?
- Is yield cut to the second, the day, or the old convention?
- Are platform assets being quoted as if they were one fund?
- Can the share be pledged while staying in regulated custody?
- Has any regulator written even a staff letter, and what does that letter refuse to bless?
Pass those and the product can still disappoint. Fail them and the token is a brochure. Brochures have their place. They should not be priced like infrastructure.
The opinion underneath the panel
Here is where I land, having sat with both the stage line and the earnings-call math. The digital-twin distinction is real, and it is the right fight for this market to have in public. Too many launches use “tokenized” for a PDF with a contract address stapled on. Investors deserve a plainer vocabulary.
I am less impressed by unverified per-transaction theater. The documented comparison, roughly $1.50 on the legacy path versus about $1.13 for a 50,000-transaction Stellar batch, is already strong. It does not need a $150 ghost to make the point. If anything, the ghost weakens it.
Second-level yield, multi-chain reach, and collateral that stays in regulated custody are the parts I would watch over the next year. Speeches fade. A treasury desk that will not go back to a Friday cutoff does not. That habit, once formed, is harder for a twin model to serve without becoming native in all but name.
Franklin is not the only firm that will get to define the category. It is, for the moment, one of the few willing to say the quiet part on a main stage: a token that does not carry the official record is a copy. Copies can be useful. They are not the same object. The market can disagree. It should at least stop pretending the two are interchangeable.
Questions worth keeping on the desk
Will other managers publish their own batch-cost studies, with legacy and chain figures side by side? Will staff relief stay a Franklin-specific path, or will examiners start expecting the same wallet separation and daily reconciliation everywhere a fund token moves? Will intraday accrual become table stakes for treasury tokens, the way T+1 became table stakes for equities?
And the awkward one from the opening. If your token and your shareholder file ever disagree, which one does your counsel call the share? Answer that, and the rest of the branding takes care of itself.
Johnson’s Singapore remarks will be clipped, quote-tweeted, and half-remembered as a cost boast. The durable piece is narrower. Official records, anonymous chain data, a transfer agent who can still correct the file, yield that notices the second a share changes hands, and a platform measured in billions that should not be confused with any single sleeve. That is a product argument. The industry can meet it with a better design, or with a clearer label. Either response would be an improvement on the mirror.
