If you still hold a balance with Cypher, you have two days left. On September 6, 2026 the provider's app, dApp and business platform go offline, and the withdrawal window closes with them. On its wind-down page the operator writes that funds are safe and can be withdrawn at any time until that date, free of charge. After that, the convenient route through the application is shut. Anyone who does nothing before then depends on their own backup, and whether that backup exists is decided long before September 6.
Cypher was a wallet application with its own payment card and its own token. In July 2026 the payments provider Nium acquired the company and sent the existing platform into wind-down. A four-stage timetable has been running since then, and three of those stages are already past. This article sorts the dates, explains what self-custody really means in this situation, and adds a measurement of our own taken this morning: what is a token still worth when its protocol ends in two days?
The operator lists the key dates on a dedicated wind-down page. In the order in which they take effect:
Three of those four dates are history. The fourth is the only one where anything can still change for you, and it is also the only one where doing nothing costs money.
Cypher belonged to a category of provider that has become common in recent years: a wallet app that holds stablecoins, a payment card that lets you spend that balance in everyday life, and an in-house token that rewards spending with cashback. The company said it was funded by Y Combinator and Coinbase Ventures, and it targeted an international audience.
The technical point that matters for today's situation sits in its own description: Cypher presented itself as a self-custody provider. Self-custody means the private keys to your balance are generated on your device and stay there, rather than in the provider's till. That is the difference between a wallet and an account: with an account the provider holds your money for you, with a wallet you hold it yourself and the provider supplies only the interface.
The distinction sounds academic until a company is sold. Then it is the whole difference.
Nium is a provider of cross-border payments infrastructure. The purchase was aimed at the technology and the team, not at keeping the existing application running. In its own announcement Cypher writes that it is already working on a next version built directly on the buyer's infrastructure. The old platform is being shut down in an orderly fashion to make room for it.
For users this is a familiar constellation. The firm does not vanish in a dispute, there is no payment default and no frozen accounts, but a timetable with deadlines. That is precisely what makes it treacherous. An orderly wind-down produces no headlines to startle you. What it produces is an email that gets lost among the marketing, and a date that falls due eight weeks later.
With a self-custody wallet your balance does not disappear when the app does. The assets sit on the blockchain, and access to them hangs on the recovery phrase, the sequence of words shown to you during setup. Anyone who has it can import the wallet into another application and carry on as if nothing had happened.
Anyone who does not have it has a problem that cannot be solved after the shutdown date. The word sequence can only be exported from within the running application. Once the application is offline there is no longer anywhere to request it, because the essence of self-custody is that the provider does not know it either.
Cypher therefore explicitly advises users to back up or export their recovery data before the shutdown and to test access with a compatible alternative wallet. The second half of that sentence is the more important one: an export you have never tried is an assumption, not a backup. How that test works and which formats fit together is something we described using the example of a change of manufacturer, in our piece on restoring a seed phrase with a different manufacturer.
CryptoSlate names one limitation, and it belongs here: not every type of balance can be carried over in this form. Balances tied to the card function or to the provider's rewards programme behave differently from plain stablecoin holdings in the wallet. Anyone who had both in the app should check them separately.
The order matters more than it looks. If you withdraw first and delete the app afterwards, you may have overlooked remainders you can no longer reach. If you export first and test the import in a second wallet, you can then withdraw at your leisure and keep a route open for anything left behind.
In practice that means: write the word sequence down on paper or stamp it into metal, keep it separate from the device, and do not upload it to a password vault in the cloud. Then import the same word sequence into a second, independent wallet and check whether the same addresses and the same holdings appear. Only when that matches is the backup a backup. Any widely used wallet that supports the same recovery format will do as the second destination; what matters is that you actually carry out the import rather than merely assume it would work.
The operator itself says two things about withdrawals: a transfer out is possible at any time before September 6, and the provider charges no fee for it. On routes and duration, the wind-down page was silent in the version available this morning. CryptoSlate reports that settlement takes place in USDC on the Ethereum layer 2 Base, and that between 24 and 48 hours may pass before the amount arrives in the destination wallet.
Take that range seriously if you are starting today. In the worst case two days lie between triggering a withdrawal and receiving it, and the window closes on September 6. An order you send on the evening of September 5 may fall into a settlement that no longer makes it past the deadline.
The second pitfall is the destination address. A withdrawal in USDC on Base belongs at an address that is served on Base. Send it to an address that exists only on another chain, or to the deposit address of an exchange that does not support this network, and recovering the amount will at best be laborious. Check before you send that the recipient explicitly names Base and USDC.
This analysis was carried out by cryptoticker.io itself on September 4, 2026. Method: a direct query of the token contract through a public Base node, plus a count of all trading pairs through a public market data interface, both at 03:53 UTC at Base block 50,852,326. Objects examined: one token contract, five trading pairs and five provider endpoints.
The contract sits on Base at address 0xD262A4c7108C8139b2B189758e8D17c3DFC91a38 and reports 18 decimals. Total supply is exactly one billion tokens, read straight from the contract. At the time of measurement the token trades in five pairs across three decentralised venues. And that is where it gets thin:
Sixteen trades in a day means roughly one trade every hour and a half. A single sale of a few hundred dollars would visibly move this market. Anyone still holding CYPR who wants to get rid of it is not selling into a market but against it.
Not verifiable and therefore not claimed here: the number of accounts affected, the German share of them, and the sum still sitting unwithdrawn in the application. Those figures are held only by the provider. Two subdomains that one would normally check alongside could not be resolved from our environment; nothing can be inferred from that about their state, and so nothing is said about it here.
The operator has announced that the token protocol will be permanently wound down on September 6 and that the token ecosystem ends with it. There is no compensation or buyback programme for holders.
Legally that is hardly surprising, and for those affected it is bitter all the same. A rewards token of this construction is not a share in the company. It carries no claim on part of the price a buyer pays for the firm, and it establishes no receivable that would be served in a wind-down. Whoever held it held a promise of utility inside an application, and that application is being switched off.
From that follows a question worth asking about any app-native token long before a shutdown is announced: what exactly is left of this token once the app disappears? For most of them the honest answer is: trading on a decentralised exchange, for as long as somebody leaves liquidity standing there. Our measurement above shows how much that amounts to when it matters.
Cypher did not issue the card under its own name alone. According to CryptoSlate, the Osmosis Pay card offering ran on the same foundation, and the operators there are said to have pointed their cardholders to the same two dates. That could not be confirmed on the public Osmosis Pay page this morning; the statement is therefore expressly attributed to the reporting outlet and is not a finding of our own.
If you use a crypto card whose issuer you have never questioned, this is the moment for that question. Between the brand on the card and the company that technically handles settlement, crypto cards often stack two or three layers. If one of those layers fails, the product fails, regardless of whose name is on the front. Which European providers work with which substructure, and who actually issues the card, is shown in our comparison of crypto credit cards.
Cypher targeted an international audience; how many users in Germany are affected is not publicly known and will not be estimated here. The check itself takes only a few minutes:
If at step three you find that you never backed up the word sequence, that is no reason to panic, but it is the reason to start today rather than on Saturday.
Cypher is not an isolated case this year. Only in late August another provider discontinued its wallet application and told users to export their keys; we covered that case under Cosmostation and the deadline for exporting keys. When an entire blockchain is switched off, the situation gets harder still: on top of the key you then need a network that can still move the assets at all.
The cases differ in their trigger and resemble one another in their sequence. First comes an announcement with comfortable lead time, then stages that switch off individual functions, and at the end a date behind which the convenient route is closed. With self-custody providers the inconvenient route through the recovery phrase stays open. With custodial providers who genuinely hold your money there is no second route; there, the deadline is the deadline.
For your own set-up one plain rule follows: for every application holding assets, the recovery phrase belongs outside that application, backed up and tested once. For amounts meant to sit still for the long term, a provider app is the wrong place anyway, because its continued existence hangs on a corporate decision.
For the CYPR token there is no third route: no buyback, no compensation, and a market that by our measurement this morning consists of $3,432.66 in depth and $63.62 in daily turnover. Anyone looking to sell should know that in this environment they move the price themselves.
The operator's wind-down page with all the dates is here in the original announcement, and the notice about the acquisition sits on the company blog.
(As of September 4, 2026. This article is not investment advice. Prices and fee structures change; check the terms with the provider before you buy.)
