I keep coming back to a simple question that most market chatter skips. If a firm wants to mint a dollar token under the new federal framework, who actually walks into the room with the application? Not the wallet vendor. Not the marketing consortium. Not the software shop that designed the minting keys. The answer, at least on the Federal Reserve path published in late September, is narrower than the slogans suggest, and that narrowness is the whole story.
Who Can Qualify To Issue Under The New Fed Path
Two packages landed on September 24. One is an application procedure for an insured state member bank that wants a subsidiary to issue payment stablecoins. The other is a much longer operating proposal covering reserves, capital, redemption, custody, and the messy plumbing that sits behind a promise to pay one dollar. Both are drafts for comment. Neither is a license. That distinction matters more than the page count.
I’ve found that people collapse three legal routes into one phrase, bank stablecoin, and then act as if every dollar token issuer can file with the same desk. They cannot. Statute language, as restated in the application notice, describes three domestic categories: a qualifying subsidiary of an insured depository institution approved by its primary federal regulator, a federal qualified issuer approved by the national bank supervisor, and a state-qualified issuer approved by its state regulator. Different doors. Different files. Different clocks.
On the Fed route, the applicant is the insured state member bank. The contemplated issuer is a controlled subsidiary. A national bank uses another primary regulator. An uninsured state member bank does not use this particular insured-bank procedure. The notice even points that uninsured institution toward its home state stablecoin regulator while existing Federal Reserve duties continue. Dry, yes. Also the first filter that decides whether a headline about “applying to the Fed” is even true.
Control Is Not A Logo On A Press Release
The proposed rule borrows familiar bank holding company ideas of control. Own or vote at least 25 percent of a class of voting securities. Control a majority of directors. Or exert a controlling influence that the Board determines after notice and hearing. Those tests look boring until a group of lenders tries to share one issuer.
A multi-bank venture raises a practical problem. Which bank controls the company? Which supervisor reviews it? The Board asks that outright. It says it may accept one application on behalf of several insured state member banks if the venture counts as a subsidiary of each. It does not say every joint venture automatically meets that test. A bank with a thin minority stake and a commercial firm calling the shots is not the same animal as a true subsidiary.
A named bank on a consortium slide does not settle who controls the issuer.
That point is timely. A large group of financial institutions recently pledged to form a stablecoin company, with a hoped-for launch in the first half of 2027, subject to conditions. A pledge is evidence of intent. It is not evidence that the eventual legal entity will use the insured state member bank route. Banks can collaborate through a company that seeks another license. Structure still decides the filing desk.
The Clock Does Not Start When The Envelope Lands
Here is where I get a little impatient with tidy timelines. There are two clocks, not one. The Board would tell an applicant within 30 days whether the filing is substantially complete and what is missing if it is not. The 120-day decision period runs from the submission date of a substantially complete application. If the Board does not decide a complete file within that window, the proposal restates the statute’s deemed-approval idea.
Filing a letter on day one does not lock approval on day 120. The submission date is the date the Reserve Bank received the last material needed for substantial completeness, not the later date of the completeness notice. Leave out information required to judge statutory factors and the file is not complete. A material change can even reset a file that once looked finished. Deteriorating finances. A rewritten business plan. Anything that leaves the information on hand insufficient.
So a company cannot start the clock by sending a half-built plan and calling it a filing. Nor can the Board deny a complete application for any reason it likes. Denial is limited to a finding that the applicant’s activities, including those of the proposed issuer, would be unsafe or unsound based on statutory factors. A denied applicant can seek a hearing and appeal. That is not a rubber stamp. It is also not an open-ended stalling machine.
- Submitted is only a letter on a desk.
- Substantially complete is the moment the 120-day count can begin.
- Approved is a decision, or a deemed approval after a complete file sits too long.
Those three statuses are not interchangeable. A press note that says an application was filed tells you almost nothing about when review actually started. In my experience, that is the sentence most likely to age poorly.
The applicant sends the letter to the appropriate Reserve Bank, which forwards a copy to the Board. It has to be signed. It has to describe the proposal, state the action sought, and explain why approval meets the statutory factors. Some existing supervisory information can reduce duplication. It does not replace a business plan, a governance map, or a redemption process for a brand-new issuer sitting under an old bank.
What The Application Actually Asks For
The disclosure load is heavy on purpose. Business plan. Financials. Policies. Agreements. Governance. Material third-party relationships. Who does what across the program. A bank can outsource code and rails. The Board still wants to see the issuer’s operating structure and the parent’s oversight of that structure. Pre-filing conversations are invited for complex deals. Those talks are not approval in a quieter room.
Perhaps the most interesting aspect is how ordinary this looks once you strip away the token language. Supervisors want to know who can mint, who can move reserve assets, who screens customers, and who is on the hook when a weekend redemption request hits a market that is closed. That is banking homework with a blockchain costume, not a new species of permission.
Reserves Back Coins. Capital Backs The Firm.
The operating proposal splits two piles of money that commentators love to mash together. Eligible reserves back outstanding tokens on a one-to-one basis. Capital is a second layer meant to absorb risks to the issuer’s own survival and certain exposures. Calling a fully reserved issuer “capital-free” confuses those accounts. I think that confusion will produce some expensive surprises.
The Fed proposes a $5 million initial minimum during a three-year de novo period, indexed to nominal U.S. GDP. The applicable minimum would be the higher of that floor and a calculated risk-based requirement. The Board could set a different amount in specified cases, including when the formula does not match real exposures. Five million is not an application fee. It is not a permanent cap. It is a proposed starting floor for a newly approved Board-supervised issuer.
One line in the long notice makes reserve design measurable. A proposed 2 percent capital charge would apply to uninsured eligible deposit claims held as reserve assets. The logic is bank credit risk. If the reserve bank fails, recovery can lag or shrink. The notice even recalls a well-known 2023 episode in which a major dollar token issuer held a very large uninsured cash pile at a failed lender. History as a warning, not as a verdict on any current balance sheet.
| Hypothetical uninsured deposits | 2% capital component |
| $250 million | $5 million |
| $1 billion | $20 million |
| $3.3 billion | $66 million |
Those figures are illustrations of one proposed component. They are not complete capital bills. Even so, they show when the $5 million floor stops being the interesting number. A billion dollars sitting uninsured at a bank already produces a $20 million charge before operational risk and other add-ons. The proposal asks whether 2 percent should instead sit somewhere between 1 and 4 percent, or vary with the credit standing of the deposit bank. At 1 percent the same billion-dollar example is $10 million. At 4 percent it is $40 million. Those ranges are questions for comment, not adopted law.
Other buckets sit in the same framework: undercollateralized reverse repurchase agreements, eligible funds, operational risk, non-reserve assets. Measurement periods differ. Treating the deposit example as the entire bill would be false precision. Treating reserve-bank choice as a branding decision would be worse.
Why Short Bills Do Not Cancel Operating Risk
A common argument says short government securities carry little credit risk, so capital talk is overdone. Fine, as far as it goes. A redemption desk still has to work on Saturday. Software access, failed wires, custody controls, reconciliation, and customer screening all cost money even when the reserve portfolio looks pristine. The proposed operational-risk charge therefore leans on inputs other than the market value of bills. Quarterly measurement is on the table for the revenue-based piece, with a question about whether another frequency would work better.
Cash at a bank and bills in a portfolio are not a binary choice in real operations. An issuer needs settlement balances to pay people. It can hold another slice in permitted liquid instruments. Promise instant cash-out while parking every dollar in paper that must first be sold, and you are betting on the sale-and-payment chain during a scramble. Keep large uninsured deposits and you may have cash on a normal Tuesday while eating the proposed credit-risk charge. Neither mix is automatically right. Both follow from treating liquidity and bank exposure as different problems.
Custody Is A Chain Of Authority, Not A Slogan
Proposed language on covered custodians aims at protection for reserve property and for the private keys that permit issuance. Picture an issuer that uses one bank to hold Treasuries and a separate technology firm to manage minting permissions. Someone still has to map who controls each asset, who can authorize movement, and how outstanding coins are reconciled with eligible backing. That is why the application wants material third-party contracts. “The reserves are safe” is not a map.
The Board also describes room to raise or lower the de novo capital requirement when a different amount looks sufficient. It asks whether three years is the right runway and whether the indexed $5 million floor should sit higher or lower. A budget that treats five million as a fixed forever cost misses both the higher-of test and that reserved authority. The real number will come from the final rule and from the issuer’s actual book.
Other agencies have already moved on parallel tracks. Similar starting floors across proposals do not erase differences in jurisdiction, process, or final text. None of these drafts should be dressed up as a finished license for a named firm. I will keep saying that until someone publishes an approval letter.
The $10 Billion Line Is A Second Gate
State supervision remains a path for eligible issuers below a statutory scale line. Once consolidated outstanding issuance passes $10 billion, the proposal sketches a 360-day transition into the federal framework, unless the issuer stops net new issuance while above the line or obtains a waiver that lets state supervision continue.
Crossing that line would trigger notice to the Board within five calendar days. The notice would name the supervising state, the outstanding amount, the crossing date, and whether net new issuance has stopped. A capital analysis would follow within 270 days. A waiver request, if any, would be due within 240 days under the proposed timetable. This is not a sticker swap. The issuer would need to meet applicable federal requirements on a clock.
Imagine an issuer at $9.9 billion. A $200 million net print takes it to $10.1 billion under a simple snapshot. The proposal asks whether measurement should instead use a rolling average and whether issuance by nonconsolidated affiliates should count. Those questions are open. It is early to claim that splitting tokens across subsidiaries will keep a program under the line forever. The Board is asking, in so many words, how affiliated issuance should be counted.
- Notify quickly after a crossing.
- Prepare a capital analysis on the longer inner deadline.
- Decide among transition, halt of net new issuance, or a waiver request.
A coin above $10 billion does not become illegal at midnight. The draft gives a transition, a possible waiver, and an off-ramp that stops net growth. A growing state issuer still faces the calendar earlier than a startup hunting its first license. Growth, reserve mix, and capital can all move while the firm prepares for a different supervisor. That is a planning problem, not a morality play.
A Redemption Promise Needs A Working Door
Eligible assets would have to back outstanding coins one for one. The Board would also want a public redemption policy that states a timeframe, fees, minimum size, and procedures. Timely redemption may not exceed two business days after a request, subject to applicable requirements. Onboarding and screening still apply. A trader who can sell a token in seconds on an exchange is not automatically the same person as an eligible customer redeeming with the issuer.
That gap is easy to miss when the market price hugs a dollar. Secondary trading shows what buyers and sellers will accept. It does not answer who holds a contractual claim on the issuer and through which pipe the cash actually moves. Banking partners, custodians, and transfer systems sit in that pipe. If the pipe is vague, the promise is vaguer than the peg chart implies.
Market price is a vote. Redemption is a contract plus operations.
Safekeeping language reaches reserve assets, tokens used as collateral, and private keys used to issue payment stablecoins. Certain Board-supervised custodians holding covered assets would face rules meant to keep customer property away from a custodian’s creditors. That is a different animal from generic wallet software that never controls the customer’s keys. The proposal asks where those boundaries should sit. Fair question. The industry will argue about it for months.
A Board member accompanying the notices stressed safeguards against runs and payment-system risk. The best case for the approach is operational. Clear redemption terms, eligible liquid reserves, capital where bank deposits are uninsured, and documented custody make a promise easier to judge before stress hits. The fairest worry from a new entrant is the upfront work and the uncertainty while several agencies finish rules that are supposed to fit together. Both readings can live in the same text. Final requirements still depend on comments and later decisions.
What These Drafts Still Cannot Tell You
There is no published list of approved issuers under these new proposals. The application notice does not reveal who will file through a state member bank, a nationally supervised entity, or a state regulator. A charter, a pending application, a partnership announcement, and permission to issue under a finished regime are four different milestones. Mixing them is how rumors get a head start.
Open items sit on the face of the notices. Final capital calibration. Whether the $10 billion test uses a snapshot or an average. How multi-bank issuers document control. How final rules across agencies line up. The length of the paperwork reflects unanswered questions as much as settled policy. A claim that a specific issuer already qualifies would need organizational documents, supervisory status, an actual application, and a regulator decision. We do not have that package in public.
Watch list: Federal Register date that starts the 60-day comment window Survival of the $5 million floor and 2% deposit charge Public decisions naming a bank and its proposed subsidiary Ownership papers on any multi-bank issuer Outstanding issuance near the $10 billion mark
There is a checkable way to follow this. Publication in the Federal Register starts the stated 60-day comment period. Final text decides whether the proposed floor and deposit charge survive. Application notices and decisions will show which banks actually seek approval. Ownership documents will show whether a bank controls a joint issuer. Outstanding issuance disclosures will show which state programs are approaching the scale line.
Practical Reading For Anyone Building A Token Program
If you work inside a bank that might file, start with control and completeness, not with token branding. Who owns the issuer. Who votes. Who appoints directors. Who can force a change in the minting process. Then build a file that can survive the 30-day completeness screen. Missing agreements and a fuzzy redemption flow are how clocks fail to start.
If you work at a state-licensed program that is growing fast, model the $10 billion crossing as an operating event, not a distant legal curiosity. Five days is a short notice window. Two hundred seventy days is not forever if capital analysis has to catch a moving reserve mix. Stopping net new issuance is a real option on the page. It is also a commercial decision with market consequences.
If you design reserves, treat uninsured bank cash as a priced exposure under this draft. Bills help credit risk. They do not staff a weekend desk. Custody maps need names, signatures, and fail-over paths. I have watched too many pitch decks skip the sentence that explains who can actually move the keys.
- Do not assume every fintech can file with the Fed.
- Do not treat 120 days as a promise attached to an incomplete letter.
- Do not budget only the $5 million floor and call the model done.
- Do not confuse exchange liquidity with issuer redemption.
- Do not assign a consortium to the Fed path without a control analysis.
Can any stablecoin company apply directly to the Fed? On this draft, no. The application route in the shorter notice is for an insured state member bank seeking approval for a subsidiary. Other potential issuers use the regulator that matches their legal form.
Does the bank or the subsidiary file? The insured state member bank files. The subsidiary would issue if approvals land.
Is approval automatic after 120 days? Only after a substantially complete application sits undecided through the statutory window. Completeness is a separate 30-day conversation.
Is $5 million enough for every issuer? No. It is a proposed initial floor. The higher calculated requirement, plus Board discretion in specified cases, can sit above it.
How do uninsured reserve deposits change the math? At the proposed 2 percent rate, a hypothetical $1 billion exposure contributes $20 million before other charges.
Can several banks share one filing? Maybe, if the issuer qualifies as a subsidiary of each. That is a facts-and-control question, not a courtesy.
What happens when a state issuer crosses $10 billion? A proposed 360-day transition, a halt-of-net-issuance option, and a possible waiver, with fast notice after the crossing.
Are the September 24 packages already binding? No. They are proposals. The comment deadline depends on Federal Register publication. This is educational analysis, not a recommendation to buy, sell, or hold anything.
Why The Legal Entity Still Decides The Story
I keep returning to entity design because it is the least glamorous and most decisive part of this week’s paper. A technology company can supply wallets. A bank can hold reserves. A consortium can publish a logo. Only some of those actors can be the applicant on the Fed’s insured-bank path. Mix them up and you get a narrative that sounds current and is still wrong about jurisdiction.
The same caution applies to capital theater. A fully reserved token can still need loss-absorbing resources because the issuer is a firm with operations, vendors, and credit exposures. A cheap-looking cash account at a commercial bank can become an expensive capital item. A bill-heavy portfolio can look conservative and still leave a redemption process that cannot pay on a holiday. Those are not contradictions. They are two ledgers.
Scale adds a third ledger. State permission can be a genuine route. It is not an infinite ceiling under this draft. Growth itself can change the supervisor. Firms that treat $10 billion as a branding target rather than a regulatory event will do the homework later, under a shorter clock, with less patience from counterparties.
None of this tells you which private name will win a license first. It does tell you how to read the next announcement. Look for the applicant, not the slogan. Look for completeness, not the date on the cover letter. Look for reserve mix, not only the claim of one-to-one backing. Look for control papers before you assign a consortium to a federal desk. And look for outstanding issuance before you assume a state program can stay state-supervised by habit.
The drafts are long because the questions are still live. Comment letters will argue about the 2 percent rate, the three-year runway, affiliate counting, and the edges of custody. That argument is useful. Treating the September 24 release as a finished gate is not. If you take one habit from this piece, make it this: separate the promise on the token from the firm that has to keep the promise when someone actually asks for dollars.
That is the qualification test hiding under the headlines. Not whether a team can buy bills. Whether the right legal person can file, fund the capital stack those bills do not replace, and still redeem on a timetable that survives contact with a real customer.
