The Fed's own research desk put a number on tokenized deposits: up to $700 billion in bank lending capacity, the same week three more banks committed to the rails.

Neither figure appears on a single bank's product page. They are the cost side of a trade only one side has been pricing.

The note bundles a scenario estimate, a standard disclaimer, and a quiet callback to the same desk's own taxonomy piece from six weeks earlier. Separated out:

One scenario, two dollar figures, and a direct line back to the desk's own earlier work. The line back is the one worth reading closely.

Here is the mechanism the two figures are pricing.

• The token is legally boring. The behavior around it isn't. A tokenized deposit is still a deposit by the same desk's own definition. What changes is how fast and how automatically it can leave for a better rate, the input both dollar figures depend on.
• Both numbers are one mechanism counted twice. Interest-rate risk and maturity transformation are two sides of the same balance-sheet math: banks fund long, fixed loans with short, flighty deposits. Tokenized rails don't create that mismatch. They shorten the short side further.

A regional Fed research desk did what outside critics haven't managed: attach a number. The systemic case against bank tokenization has mostly come from stablecoin skeptics drawing deposit-flight comparisons, while the banks marketing the products describe them as behaving exactly like an ordinary deposit. This note comes from inside the Federal Reserve System's own research apparatus and publishes a quantified estimate instead of a qualitative worry.

It lands the same week three separate commitments turned the risk from hypothetical to scheduled. Wells Fargo's fall launch, the 25 August formation of the 39-association BankChain Alliance, and JPMorgan's JPM Coin already running in production on Base are not proposals anymore. They are dated commitments the Dallas Fed's scenario now has something concrete to price against.

No capital or liquidity rule references either figure, yet. Basel's LCR and NSFR treat deposit outflow risk through categories built before deposits could move at the speed of a wire that never sleeps. This note is the argument for revisiting that treatment. It is not evidence that anyone has.

• It's a scenario, not a forecast. The 10% shifts in rate-sensitivity and weighted average life are illustrative inputs the authors chose to size the mechanism, not observed data or a base-case adoption projection.
• Standard Fed disclaimer applies. The views are Levy's and Ramaswamy's, not the Dallas Fed's or the Federal Reserve System's. This is research, not supervisory guidance or a proposed rule.
• The instrument itself isn't new risk, by the same desk's own account. Per the 14 July companion piece, a tokenized deposit stays on the issuing bank's balance sheet, settles at par, and sits inside the existing supervisory perimeter. The concern here is behavioral, not legal status.
• Nothing forces a bank to change how it treats a tokenized deposit today. The note argues for revisiting capital and liquidity treatment. It does not itself change any requirement a bank has to meet.

Wells Fargo's fall launch is one data point in a wider build-out. The Clearing House is targeting the first half of 2027 for a shared interbank tokenized deposit network, and 39 state banking associations, representing 3,283 banks and $21.8 trillion in assets, spent 25 August forming BankChain Alliance around the same goal, with a technology partner still unselected. JPMorgan needs neither: JPM Coin has run as a live deposit token on the Base blockchain for over a year.

tracee's July briefing on SWIFT's tokenized deposit pilot already flagged Wells Fargo among the 17 banks testing 24/7 settlement rails. The Dallas Fed note is the first serious attempt to price what that speed costs the rest of the balance sheet, not just what it adds to the payments side.

Unlike a payment stablecoin under the GENIUS Act, a tokenized deposit never leaves the FDIC-insured perimeter, which is exactly why this note matters. The risk it describes can't be regulated away by treating tokenized deposits as something other than ordinary bank funding. It has to be handled inside the same capital and liquidity rules banks already operate under.

Tokenized deposits are, and remain, ordinary bank deposits: on balance sheet, FDIC-insured, settling at par. The Dallas Fed's 25 August note doesn't dispute that. It prices what happens once that ordinary deposit can move between banks in seconds instead of days. Up to $700 billion in interest-rate-risk capacity and $580 billion in maturity-transformation capacity are scenario estimates, not observed losses, but they arrive the same week Wells Fargo, BankChain Alliance, and JPM Coin turned the underlying behavior from a research assumption into a shipping product. The gap between the two is now the thing to watch.

Watch three things:

• Whether prudential regulators start citing rate-sensitivity or deposit duration explicitly in capital or liquidity rules, rather than treating a tokenized deposit the same as an ordinary one for LCR and NSFR purposes.
• Whether Wells Fargo's fall launch or JPM Coin's growth produces real deposit-switching data, which would move the Dallas Fed's 10% inputs from assumption to observation.
• Whether BankChain Alliance's technology choice, still pending, ships with any rate-sensitivity guardrails built in, or leaves that question to the individual member banks.