Three exchanges. One asset. Eleven weeks apart.
Binance opened the door in February 2025, accepting Franklin Templeton's tokenized money market fund shares as off-exchange collateral custodied by Ceffu. OKX followed in April, with Standard Chartered holding the assets. This week, Bybit became the third — parking shares with its own ByCustody platform and mirroring their value into USDT and USDC trading credit.
The announcement says the arrangement "reduces counterparty risk."
I have audited collateral systems since 2017, when I pulled an integer overflow out of a token minting function before a public sale opened. Read that sentence again. Then read Bybit's February 2025.
Nothing in this product moved on-chain. Nothing settled. The entity promising to honor the collateral once held the largest stolen balance in crypto history.
Franklin Templeton's Benji ledger now sits behind three of the largest exchanges on earth. That is the story nobody is publishing.
The mechanics matter more than the marketing. Franklin Templeton is not a startup. Founded in 1947, it manages roughly $1.7 trillion. Its tokenized money market fund shares represent a pro-rata claim on short-duration Treasuries — a real, interest-bearing asset. Its transfer agent platform, Benji, handles on-chain registry and dividend distribution.
Bybit's structure is simpler than its diagram. A qualified client pledges FT shares through ByCustody. Bybit does not take delivery. It credits a mirror value inside its own trading system and extends USDT or USDC margin against it. The shares keep paying yield. The client gets capital efficiency — one asset, two functions. Access is gated to qualified clients, which tells you the retail version is not the point yet.
Bybit is also promising a second product: tokenized wealth offerings distributed on Mantle, its affiliated public L2. Details were withheld. That silence will matter later.
This is not new technology. This is securities lending, re-implemented. The question was never whether it works. The question is who carries the liability when it stops working. The floor is a lie; only the ledger that backs it.
Start with the balance sheet, because that is where the truth lives.
Every collateral arrangement answers one question: if the borrower defaults, who eats the loss? In a repo book, the clearinghouse. In a bank custody model, a regulated institution with capital requirements and a supervisor who reads its filings. In OKX's arrangement, Standard Chartered occupies that seat. A bank.
In Bybit's, ByCustody does. ByCustody is a Bybit subsidiary.
Institutional credit desks price this precisely. The custodian's balance sheet is the collateral's quality. A money market fund share backed by Treasuries is close to risk-free. A money market fund share whose redemption depends on one exchange's solvency is not. The asset did not change. The counterparty did.
Which brings us to February 2025. Bybit lost roughly $1.4 billion to Lazarus Group — the largest single exchange breach on record. The exchange survived, reimbursed users, and rebuilt. That is a real operational achievement. It is also a permanent entry in every institutional risk memo written since. Now that same entity asks clients to trust a mirror ledger it controls, with no public audit, no disclosed haircut, no published liquidation waterfall.
This is not elimination of counterparty risk. It is a transfer of it — from "my assets could be stolen from the exchange" to "my claim on the exchange's internal ledger could be impaired." The first risk is visible, and insurable where possible. The second is an accounting entry with a logo. The floor is a lie; only the whale. Here, the whale is the spread.
Consider the undisclosed parameters. Haircut, the discount applied to fund shares. LTV, the ratio determining margin. Financing rate, the interest charged on credited USDT. None appear in the announcement. Without them, "capital efficiency" is unfalsifiable. If the financing rate exceeds the fund's yield, the client runs a net negative carry while congratulating themselves on optimization.
In 2022 I watched UST decouple from LUNA reserves 48 hours before the terminal cascade. The lesson was not that algorithmic stablecoins fail. The lesson was that a claim is only as good as the mechanism that redeems it. Mirror ledgers are claims.
Then there is Benji's architecture. Franklin Templeton is a US-registered asset manager; its ledger is almost certainly permissioned — authorized nodes, not a public chain. That matters, because the same announcement mentions tokenized wealth products distributed on Mantle, a public L2. Two technical stacks, one press release, no distinction drawn. Readers will conflate them. Developers should not.
And the regulatory piece — the actual headline. CFTC staff have now permitted futures commission merchants to invest customer funds in tokenized assets. That single sentence does more for this model than any exchange partnership. It legitimizes the collateral type. Benji stops being a product and becomes infrastructure, the way Circle became infrastructure. Franklin Templeton distributes one asset across Binance, OKX, and Bybit simultaneously. The exchanges compete. The issuer compounds. The floor is a lie; only the whale sits on the balance sheet.
The consensus read is that RWA is maturing. The forensic read is that RWA is being rented.
Strip the blockchain layer and the structure is a margin loan against a Treasury portfolio, intermediated by a broker that never takes possession. TradFi has run this since the 1980s. The token adds round-the-clock settlement, programmability, and — most importantly — a story. Mechanical innovation: small. Distribution innovation: not.
The blind spot is causality. Observers will credit the next RWA rally to announcements like this one. But three exchanges adopting the same issuer's asset within eleven weeks is a distribution decision, not a demand signal. Franklin Templeton chose multi-platform issuance for the reason it diversifies any counterparty exposure: concentration is the risk. Bybit is the third mover, roughly four months behind Binance, two behind OKX. The floor is a lie; only the issuer. Follow the issuer's incentives, not the exchange's press release.
Watch four signals. One: ByCustody's legal structure — an isolated SPV or trust means the risk is contained; a line item on the group balance sheet means it is not. Two: disclosed haircut and financing spread. Three: the size and architecture of the Mantle wealth product, when details arrive. Four: whether a second issuer — BlackRock's BUIDL, Fidelity's fund — replicates the pattern.
If they do, tokenized fund collateral becomes the default margin asset in CeFi. If the parameters stay hidden, it becomes a liquidity story dressed as a custody story. One of those is real. The other settles tomorrow.