Wednesday, a crypto card program stopped existing. Not because of a smart contract bug. Not because a token price collapsed. Kulipa, the card issuer behind Ready’s payment card, suddenly wound down. Ready — the wallet formerly known as Argent — and Solflare both woke up to the same reality: cards dead, zero notice. Founder Itamar Lesuisse said he found out at the same time as his users. That sentence is more damning than any exploit report.
Here is the anomaly the market missed: there was no price chart event. Bitcoin did not flinch. The ledger did not fork. Yet a piece of financial infrastructure that connected the self-custody world to the traditional card rails just vanished. The Defiant reported the shutdown. But the broader takeaway is not that one product failed; it’s that an entire route to fiat was built on a single counterparty that no one could verify.
Context: A Bridge, Not a Blockchain
Ready is a self-custodial smart-contract wallet. It holds keys to on-chain assets. Solflare is a Solana-native wallet with a similar philosophy. Neither project tries to be a bank. Instead, they outsourced the banking part to Kulipa, a card issuer that lets non-custodial wallets plug into Visa and Mastercard networks. The result was a product that felt like a bank card but worked like a self-custody keychain — until the bridge disappeared.
The Defiant’s report gives us the key facts: Kulipa suddenly wound down its card issuing operations, leaving Ready, Solflare, and likely other wallet projects without card service. Ready’s founder said there was no advance notice. The company stated that user funds were not affected. That last line is doing a lot of work. On-chain funds probably were safe. But card service was not, and service continuity is what card users actually pay for.
The architecture was never a single system. It was a split stack. On one side, smart contracts and self-custody. On the other, a regulated issuer with KYC, AML, bank sponsors, and card program licenses. The two halves are joined by connectors that no one on the crypto side can inspect. Users were told to trust the code, but the real trust was in an unlisted intermediary.
Core: A Single Point of Failure in the Fiat Layer
My first question when I see a black-swan event is not “what happened?” It is “who else is sharing the same exposure?” I learned this habit the hard way in 2017, when I audited Ethereum Classic’s code ahead of a network split. The question was not whether one contract was safe; it was whether a single faulty primitive could take down every dependent system. Kulipa is that kind of primitive.
Ready and Solflare both used Kulipa. That means the failure is not idiosyncratic. It is a correlated shock. A single issuer was acting as the fiat off-ramp for multiple, unrelated wallet ecosystems. When it collapsed, every dependent product fell at the same time. That is the classic definition of a single point of failure. No amount of decentralized ledger magic fixes a centralized card processor that wakes up dead.
Let’s be precise about what a card issuer does. Kulipa sat between blockchain wallets and the traditional payment network. It needed bank relationships to sponsor a BIN, an eight-digit identifier that allows a card to be recognized by merchants and ATMs. It needed card scheme approval from Visa or Mastercard. It needed compliance systems for sanctions screening, suspicious activity monitoring, and customer due diligence. All of that is confidential, contractual, and absolutely opaque to a wallet builder.
You can audit a smart contract line by line. You cannot audit a banking partner’s risk appetite from a block explorer. You can verify that your assets are safe on-chain. You cannot verify that a sponsor bank won’t pull its supporting relationship tomorrow. The asymmetry is structural. Where the code forks, we find the fold.
Now, the part that should worry every wallet builder: switching issuers is not a configuration change. A new issuer requires new KYC flows, new BIN allocations, new bank sponsors, new compliance due diligence. Even if Ready signs a new partner today, existing users will need to re-apply, receive new physical cards, and wait for card scheme approval. This is not a migration; it is a product relaunch. The fact that Ready did not even have a public warning suggests the team had no observable health indicator for Kulipa’s banking relationships. I have seen the same failure in software supply chains: teams monitor their own code but not the legal and financial stability of their critical dependencies.
The phrase “user funds were not affected” is also narrower than it sounds. It refers to the on-chain wallet balance. It does not necessarily mean that any fiat balance pre-loaded into the card program is fully protected. We do not know from the reporting whether card-linked balances existed, how they were held, or whether users will be made whole. If the card program was a pass-through ledger, there may be claims against a company that no longer operates. That is not a reason to panic, but it is a reason to separate what is verified from what is assumed.
Floor cracks reveal the foundation’s weight. The foundation here was not a blockchain. It was a company. And companies can fail in ways that code cannot.
Contrarian: The Wrong Lesson Is “Self-Custody Failed”
The crypto-skeptic takeaway will be simple: “self-custody still depends on trusted third parties.” That is true and useless. The event did not prove that self-custody is fake. It proved that self-custody solves one problem — asset control — while leaving another problem unsolved. The other problem is access to fiat payment rails. Those rails are not decentralized, and pretending they are is a recipe for exactly this kind of breakdown.
If the market reacts by abandoning self-custody cards and moving to custodial products from centralized exchanges, it will be trading one single point of failure for another. Binance and Crypto.com can issue cards through their own licensed entities, so those cards may be more stable at the payment layer. But the assets behind them sit under exchange control. You exchange a counterparty risk in the fiat layer for a counterparty risk in the custody layer. That is not a risk reduction; it is a risk relocation.
The correct response is boring: build redundant fiat rails. Multiple issuers, multiple card programs, standard interfaces for switching sponsor banks. That is how traditional fintech companies handle the same vulnerability. They do not rely on “trustless” magic; they rely on backup contracts and failover processes. Crypto wallets that treat card issuance as a strategic service need the same discipline. A wallet with no issuer fallback is not a product. It is a bet.
This is also a correlation event, not a one-off. In options trading, when correlations spike, everyone’s hedges fail at the same time. Kulipa’s collapse hit multiple wallets simultaneously. There may be other undisclosed clients of Kulipa that have not spoken yet. Watch the next few weeks. If a second issuer suddenly winds down, the market should treat the entire card infrastructure as systemically fragile. If no other shoe drops, we still learned the same old lesson: counterparty risk can nullify perfect code.
Takeaway: List Your Issuer Fallback
Volatility is the premium on uncertainty; service continuity is the premium on redundancy. Every wallet team that offers a card should have one answer ready when asked: “Who is your backup issuer?” If the answer is a blank stare, the product is not ready for real use.
Users should treat any self-custody card as a convenience, not a guarantee. Funds may be safe on-chain, but the plastic in your pocket is a lease on someone else’s banking relationship. That lease can expire without notice.
The next three to six months will separate serious projects from marketing demos. Watch for dual-issuer announcements, migration plans, and honest public post-mortems. Strategy is the shield; execution is the sword. The ledger remembers what the market forgets: on-chain assets stayed put, but the off-ramp just fell through. The question now is whether the foundations can hold a second floor before users stop asking.