Synapse and Evolve: What Failed and What Changed

Which of these are you?

• You want to know what happened. Read the timeline and the "which figure means what" table. The dollar figures in circulation disagree, and the table says why.
• You want to know whether the rules changed. Go to the tracker. It lists each item with its status and the date we checked it.
• You use a fintech app today. Read "What does this change for you", then do the lookup in chapter 3.
• You saw a large number and want to quote it. Use the CFPB figure and its definition, or do not quote a total at all.

What was Synapse, and what broke?

Synapse was a banking-as-a-service middleware company. It was not a bank. It kept the ledger that recorded which customer's money sat where, across fintech apps and the partner banks that held the deposits. The CFPB says Synapse failed to keep adequate records of where consumers' funds were located and to make sure those records matched the records of its partner banks.

When Synapse filed for Chapter 11 on April 22, 2024, the reconciliation exposed the gap: the partner banks found they held less in consumer funds than Synapse's records showed. The CFPB describes the result as a shortfall of between $60 million and $90 million, and says consumers did not have any access to their funds for weeks or months, with many not yet receiving their full balances.

The dated record, from primary sources:

The stipulated judgment includes injunctive relief, a prohibition on selling customer information, and a $1 civil money penalty, which is what allows the CFPB to reach its Civil Penalty Fund for consumer redress. The Federal Reserve Board's June 2024 order against Evolve cited deficiencies in anti-money-laundering controls, in oversight of its fintech partnerships and in consumer compliance, and was coordinated with the Arkansas State Bank Department. It came from the Federal Reserve Board, Evolve's primary regulator, not from the FDIC. It was a supervisory order, not a bank failure.

Why didn't FDIC insurance simply pay?

Because the event insurance responds to did not happen. The FDIC says deposit insurance does not protect against the insolvency or bankruptcy of a nonbank company, and that if a nonbank fails, recovery runs through court proceedings and may take considerable time. No insured bank was closed. What failed was the record of ownership and the company that kept it.

The CFPB's 2023 payment-app report anticipated the shape of this. It found that pass-through eligibility is extremely difficult to verify before a failure, because the FDIC or NCUA decides it only afterward. Chapter 3 covers the lookup and the arithmetic that apply to you before anything goes wrong.

Which dollar figure means what?

Four kinds of numbers circulate, and they measure different things. Use one, say what it measures, and do not add them together. The CFPB's range is the only one on this list that comes from a primary source we retrieved and that defines its own terms.

Two arithmetic checks keep the table honest. $265 million less $180 million is $85 million, which falls between $60 million and $90 million. The two allocations add to $55,214,058. Neither shows how much any one customer recovered, and we have no primary source for a per-customer or per-app recovery figure.

What did regulators change, and what is still pending?

Less has become final than most summaries suggest. Last checked: 2026-10-04. Re-check due: 2026-11-03. This tracker is the fastest-decaying page in the series, so each row carries its own verification date and source. Status wording is limited to what we could confirm on that date.

Reading the table correctly matters more than the table. The $5,000,000 in H.R. 8087 is a ceiling for a rule the FDIC would write, not a fixed new limit, and it applies to noninterest-bearing transaction accounts only. A proposed rule is not a rule: the custodial proposal has been open 732 days since publication, and for 626 days since its comments closed. A bill referred to committee has not been passed. We found no enacted change to the $250,000 standard amount.

We could not retrieve the congress.gov page directly at writing time, so the bill rows rest on the congress.gov data retrieved the same day by our verification pass. If any status here has moved by the time you read it, the date column tells you how old it is.

How does a ledger break in the first place?

Any structure with two sets of books has a reconciliation problem, and the safeguard is matching them often and exactly. In a pooled arrangement there are at least two: the middleware's ledger of each customer's share, and the bank's record of the pooled account's total balance. Every deposit, withdrawal, card swipe and fee has to be reflected in both. If the two drift apart, the gap is invisible while money keeps flowing in and out, and it becomes visible only when something forces a full count. A bankruptcy is such a forcing event.

That is a general description of the mechanism, not a finding about any one company. For Synapse specifically, the CFPB's statement is narrow and sourced: Synapse did not keep adequate records of where consumers' funds were, and did not ensure its records matched its partner banks' records. The FDIC's custodial proposal is aimed at the same weak point, which is why its central requirement is daily reconciliation for each individual owner. Whether that proposal becomes a rule is the open question in the tracker.

This is also why the structure matters to a depositor even when every bank is solid. FDIC rules make the bank's records, or records kept in good faith by someone who undertook to keep them, the basis for recognizing each owner's share (12 CFR 330.5). If the second set of books is wrong, the claim depends on reconstructing it.

What do the three reform options mean in practice?

The FDIC's May 1, 2023 report set out three paths: limited coverage, which keeps the current framework at a specified limit that could be higher than $250,000; unlimited coverage, which insures all deposits; and targeted coverage, which varies the limit by account type and gives business payment accounts substantially higher protection. The FDIC said targeted coverage best meets the objectives of financial stability and depositor protection relative to cost, and that the options would require action by Congress.

Mapped onto the 2026 bills, only the targeted idea shows up in the text we read. H.R. 8087 directs the FDIC to set a higher amount for noninterest-bearing transaction accounts, a category the options report ties to business payment accounts. A higher amount for those accounts would not change the limit on a savings account or a certificate of deposit, and it would not address a middleware ledger. If a bill like this became law, it would change the insured amount at a bank. It would not change who has to prove what after a nonbank fails.

How do you track these rules yourself?

You can re-check any row of the tracker in a few minutes, and a status you verify yourself is worth more than ours. Use these steps:

1. For the custodial proposal, open the Federal Register and search "Recordkeeping for Custodial Accounts", then open the reginfo.gov Unified Agenda entry for RIN 3064-AG07 and read the stage and the final-rule date.
2. For the bills, open the congress.gov page for H.R. 8087 and read the latest action and its date. A change from "referred to committee" to a markup, a committee report or a floor vote is real movement; a new cosponsor is not.
3. For Part 370, open the current eCFR text and check the covered-institution definition and any amendment note at the end of the section.
4. Write the date. Our tracker is dated 2026-10-04 and due for a re-check 2026-11-03, 30 days later. Do the same with any note you keep.

If a status has moved, change the row and its date together. A tracker with an old date and a new status is the kind of error this chapter exists to prevent.

What does this change for you?

Two things stay the same as before the collapse, and one is better understood. The insurance still attaches to the bank, not the app. The pass-through claim still depends on records you cannot verify in advance. What is better understood is the failure mode: the loss of access can come from a ledger problem with every bank open, and it can last weeks or months.

Five checks, kept short because chapter 3 has the working version:

1. Is your app a bank or a go-between? If it is a go-between, find the named bank.
2. Is the named bank on BankFind? Confirm the bank, not the app.
3. Is your balance pooled for the benefit of customers? Pooled balances rest on the ownership records.
4. How long could you go without this money? A balance you need next week does not belong only in a pooled account.
5. Do you hold at least one account where you are the named account holder at a chartered bank? That one does not depend on an intermediary's ledger.

The point is not to avoid apps. It is to size what you keep where, with the structure in view. Our account freeze guide covers a different cause of lost access, and the two business banking apps comparison names two companies and their banks.