How did we end up with $36 billion sitting in tokenized claims on the US Treasury, with almost no honest discussion of what happens at the redemption layer when you actually want your dollars back?
Most "RWA is the future" pieces stop at the issuance side. The tokens exist. The yield accrues on-chain. The custodian holds T-bills in a segregated account. All true. None of it tells you what happens between the moment you click *redeem* and the moment the wire lands. That gap is the thing I keep failing to find written down honestly. So this piece is a timeline of how we got here, and what the redemption layer actually is once you walk through it event by event.
November 2022: FTX Pauses Withdrawals and the Custody Question Becomes Public
Before November 2022, "where are my dollars" was a question most retail crypto users had never seriously asked. The exchange page showed a USD balance. The balance was a number. The number could be wired out, eventually, and people did, and the system mostly worked.
Then FTX paused withdrawals on a Tuesday and the question stopped being academic.
What got buried in the bankruptcy coverage is the part that matters for the RWA argument: the "USD balance" on an FTX account screen was never a dollar. It was a database row representing a claim on a pooled account that turned out not to contain what it was supposed to contain. The mechanism between *number on screen* and *dollars in your bank* had a custodian, a banking partner, a reconciliation process, and an unwritten assumption that all of it would work the same way next Tuesday as it did last Tuesday.
That assumption broke. And the pitch for tokenized real-world assets — for the entire BUIDL / BENJI / OUSG / USDY category that did not yet exist at scale — was downstream of exactly this realization. The promise was: instead of a database row at an exchange, you hold a token that represents a verifiable, on-chain claim on a real Treasury bill held by a real qualified custodian. The chain becomes the reconciliation layer. The custodian becomes auditable in something close to real time.
The pitch was good. The pitch is still good. The pitch is also not the redemption layer.
March 2023: USDC Depegs and the Redemption Plumbing Fails Live
In the weekend of March 10–13, 2023, Silicon Valley Bank failed. Circle had roughly $3.3 billion of USDC reserves on deposit at SVB. Redemption windows for USDC were closed for the weekend because Circle's wire rails ran through commercial banks that were not open on Saturday. USDC traded as low as around $0.87 on Saturday before recovering by Monday.
This is the dress rehearsal for every conversation about tokenized treasury redemption that anyone is having in 2026. It is the cleanest case study in the file.
A "fully reserved" dollar token, issued by a regulated US entity, with monthly attestations, broke its peg for about 48 hours because the underlying redemption mechanism was a wire transfer from a commercial bank that was closed. Not the chain. The chain kept running. The token transferred peer-to-peer at full speed all weekend. What failed was the leg between the smart contract and the dollars.
Two things I want to flag from this episode, because they propagate through everything that comes next.
First: the *token* was fine. The contract executed. Balances were accurate. The mismatch between "token" and "what the token represents" only surfaced at the redemption boundary.
Second: the recovery was political-institutional, not protocol-level. The FDIC announced full backing of SVB depositors on Sunday evening. USDC repegged on Monday morning. Nothing about the smart contract changed. The repeg was a US Treasury statement. Worth sitting with.
March 2024: BlackRock Launches BUIDL and Institutional Tokenization Gets Real
On March 20, 2024, BlackRock launched the BlackRock USD Institutional Digital Liquidity Fund — BUIDL — on Ethereum, with Securitize as the token administrator. BNY Mellon as the custodian for the underlying assets. Anchorage Digital, BitGo, Coinbase Custody, and Fidelity Digital Assets among the named digital asset custody providers in the ecosystem.
This is the moment "tokenized treasury" stops being a 2022 panel-talk concept and starts being a product that institutions actually allocate to.
But read the redemption mechanics. BUIDL holders do not redeem to a bank account by sending the token to a contract address. Redemption goes through Securitize. The token is burned on-chain. The cash settlement is a wire from the fund's account to the investor's whitelisted bank account. The wire runs through normal banking hours. There is a daily redemption window. There is an off-chain reconciliation step. None of this is hidden — Securitize documents it — but the marketing language consistently emphasizes the on-chain side and downplays the wire-transfer leg.
If you want to know whether your tokenized Treasury actually behaves like a Treasury at the redemption layer, the question to ask is not "is the custodian reputable" (BNY Mellon, yes, obviously) but "what is the worst-case path from token burn to dollars-in-my-bank, and who controls each leg of it." For BUIDL: same-day redemption is the design target. Stress scenarios are not in the marketing.
October 2024: Ondo, Franklin, and the Quiet Multiplication of Redemption Models
By late 2024, the category had a half-dozen brand-name products. Ondo's OUSG (which sits on top of BUIDL for part of its book) and USDY. Franklin Templeton's FOBXX, the BENJI token. Superstate's USTB. Each one has its own redemption design. Each one has its own off-chain reconciliation. Each one has its own banking partner. Each one has its own KYC-at-exit policy.
This is where the *checklist* version of the redemption analysis becomes useful. When you evaluate one of these products, here is what you should be reading for, beyond the yield headline:
What is the redemption window — minutes, hours, business day, longer? Who is the bank that wires the dollars out, and is it a single bank or a panel? Does the issuer reserve the right to gate redemptions during stress, and where is that disclosed? Is your KYC re-verified at redemption or only at issuance? What is the minimum redemption size, and does a partial redemption have different terms? Does the underlying T-bill custodian have FDIC pass-through coverage on the cash leg, and what happens to in-flight wires if the custodian's bank fails on a Friday?
I am not going to tell you the answer to all of these for all of those products in this piece, because the grounding I am working from is honest and I do not have a uniform comparison table I trust. What I will tell you is that the answers are not the same across products, and the marketing pages treat them as if they were.
June 2026: $36 Billion and the Standardization of Friction
Which brings us to now. The category has crossed $36 billion. The institutional credibility problem of 2022 is solved. The custody side is genuinely robust — Coinbase Custody as a NY DFS Trust Company, Fidelity Digital Assets as a NY DFS Trust, Anchorage Digital as an OCC Federal Trust Charter. These are the strongest qualified-custody postures the US system produces. You can audit them. You can read their regulatory filings. The custody layer is not the weak link.
The redemption layer still is.
Here is the math teardown. Take a $1,000,000 position in a tokenized Treasury product yielding roughly 5.2% annualized — replace that number with whatever the current 4-week bill is paying, the structure of the calculation does not change. You initiate redemption Friday at 3:00 PM Eastern. Token burn is effectively instant. The issuer processes the redemption instruction on the next business day, Monday. The wire instruction reaches the custodian's bank Monday afternoon. The wire settles Tuesday morning.
Total elapsed: four calendar days. At 5.2% annualized on $1,000,000, four days of foregone yield is approximately $570. Add a wire fee of around $25 to $50. Now consider the tail risk: during those four days, your position is not in T-bills and it is not in cash in your bank account — it is a redemption claim against the issuer's settlement chain, with a wire in flight through a single commercial bank. If that bank has its SVB weekend during your four days, you are an unsecured creditor in a bankruptcy, not a Treasury bill holder.
For $1,000,000, the friction is $570 plus tail risk. For $100,000, it is $57 plus the same tail risk, which now dominates. For $10,000, the tail risk is the only thing on the page.
The number that goes unsaid in every tokenized-Treasury pitch deck is: *what wire-transfer infrastructure am I exposed to during settlement*. Because that is the thing the token does not abstract away. That is the thing the chain cannot do for you.
What It All Means
The custody layer of tokenized treasuries is real. The issuance layer is real. The on-chain transferability is real. The yield is real. None of that is the argument I am making here.
The argument is that "tokenized" and "redeemable" are different properties, and the distance between them is exactly the wire-transfer leg from the custodian's commercial bank to your commercial bank — which is the same wire-transfer leg that USDC depended on in March 2023, and the same one that failed for 48 hours when its commercial-bank counterparty failed. Tokenization did not eliminate this leg. It did not shorten it. It made everything *upstream* of it auditable in a way that earlier custody arrangements were not, and that is genuinely valuable. But the redemption boundary is a wire transfer. And a wire transfer is a wire transfer.
The self-custody analogy is sharp here. When you hold BTC at a Ledger or Trezor device, the redemption layer is *you signing a transaction*. There is no third party between the key and the asset. When you hold a tokenized Treasury, the redemption layer is *the issuer's banking relationships*. The token is a claim. The claim is only as fast and as reliable as the slowest commercial-bank leg in the settlement chain.
This does not mean tokenized treasuries are a bad product. They are not. For institutional cash management, the package — qualified custody, on-chain transferability, programmatic accounting, daily redemption — beats the alternatives on most dimensions. What it means is that the marketing language equating *tokenized* with *cash-equivalent* is doing real work that the underlying mechanism does not earn. There is a wire. The wire takes time. The wire goes through a bank. The bank has a balance sheet. That is the part of the picture the chain does not redeem.
The forward question is not whether RWA tokenization keeps growing — it will, the institutional fit is too strong. The question is whether the redemption-layer disclosures evolve to the point that a buyer can actually price the wire-transfer risk into the yield they accept. That work is not where this piece ends. It is where the next one starts.
FAQ
Why does the redemption layer matter if the on-chain custody is verified?
Because on-chain verification proves the issuer holds the underlying T-bills. It does not prove the issuer can deliver dollars to your bank account on a given timeline. Those are different guarantees. The token attests to the asset; the wire transfer delivers the value. Custody verification is necessary but not sufficient — the redemption boundary is where settlement actually happens, and that boundary runs through a commercial bank with normal banking hours and normal counterparty risk.
Is a tokenized Treasury safer than holding USDC?
It depends on which risk you are trying to avoid. Tokenized Treasuries from issuers like BlackRock or Franklin Templeton sit on top of qualified custodians — NY DFS Trust companies like Fidelity Digital Assets and Coinbase Custody, or the OCC-chartered Anchorage Digital — which is stronger than the typical stablecoin reserve posture. But the redemption leg still runs through commercial banks, and that is exactly the leg that failed for USDC in March 2023. Custody risk is lower. Wire-transfer-leg risk is the same.
How long does redemption actually take in practice?
The marketing answer is same-day or T+1. The honest answer is: it depends on when you initiate, which bank rail the issuer uses, whether your KYC needs re-verification, whether the issuer has gating provisions, and whether the wire crosses a weekend or a US bank holiday. Friday-afternoon redemption frequently means Tuesday settlement. Treat any duration shorter than two business days as a best-case scenario, not a contractual guarantee.
What is the wire-transfer concentration risk people are not talking about?
If a tokenized Treasury issuer routes all redemption wires through a single commercial bank — and most do, because the operational setup is simpler — then for the duration of any wire in flight, your effective counterparty is that bank. If the bank fails during your settlement window, your wire is not a Treasury exposure, it is a general claim against the bank's estate. This is precisely the SVB weekend mechanic. Cash in flight is bank exposure, not Treasury exposure.
Does holding via a qualified custodian like Coinbase Custody change the redemption mechanics?
Custody and redemption are separate problems. A qualified custodian holds the asset securely and meets a regulated standard for segregation, audit, and reporting. The redemption mechanics — how you convert that custodied position back into dollars in your bank account — are determined by the issuer's settlement design, not by the custody arrangement. Coinbase Custody holding the underlying T-bills does not change the fact that the dollars come out via a wire on someone's commercial-bank rail.
Can the issuer gate redemptions during stress?
Most institutional tokenized Treasury products reserve the right to suspend or gate redemptions in extraordinary circumstances — this is standard fund-document language inherited from the traditional money-market fund world. Read the offering document, not the marketing page. The right to gate is almost always there. The trigger conditions are usually broad and discretionary. This is not a tokenization issue; it is a fund-structure issue that the tokenization layer inherits without modifying.
Is self-custody an alternative to tokenized Treasuries?
Not directly. Self-custody on a Ledger, Trezor, or GridPlus Lattice1 secures crypto-native assets where the key controls the asset. Tokenized Treasuries are claims on off-chain T-bills held by a custodian; even if you hold the token in self-custody, redemption still goes through the issuer's banking infrastructure. Self-custody removes intermediary risk on the *token*. It does not remove intermediary risk on the *underlying claim*. The two layers have to be evaluated separately.