Sponge builds a USDC-backed Visa card that developers hand to AI agents. For a person who needs to buy groceries, almost none of the ordinary questions have published answers.

Updated October 2026

Licensed

Available in US

Fast payouts

18+ Only

If you have never heard of Sponge and you are looking for something to pay for a coffee, a subscription or a train ticket, close this page: the product on offer is not built for that. Sponge is a tool for developers. It issues a Visa card number that is created through a command-line tool or an SDK, funded with USDC, and handed to an AI agent so the agent can spend on your behalf without being given access to a bank account. The card exists as a component in a piece of software. There is no consumer app, and the card’s own pages do not list the countries it serves, the fee schedule, the daily limits or the ATM situation.

I gave it 4.9 out of 10, and I want to be honest about what that number is made of: it is not a verdict on a scam. It is a verdict on a product that has published a great deal about its plumbing and almost nothing about the things you would need in order to decide whether to put your own money behind it.

The conditions described here are current as of October 28, 2026.

Who this is for, in one line: developers who are giving an AI agent a USDC-backed Visa card, and who already know what a BIN is. Who it is not for: anyone looking for a normal everyday payment card, a bank account, or a fallback when a normal card is refused.

Will they give you the card?

Short answer: the card’s public pages do not say. There is no country list anywhere on them. What the platform terms do tell you is a minimum age of 13 for the service — and that is the whole of the eligibility information published.

That sounds like a small thing, and it is not. Whether you can get a card at all depends on where you live, and two separate questions are involved: where you are resident, and what nationality you hold. Those are not the same, and they get screened differently. Nationality and country of birth are recorded and checked separately from residence, because sanctions screening and an issuer’s own risk policy are often applied to citizenship, not just to your address. Many regulators treat a person who cannot be screened as a person who cannot be served at all: EU rules require an institution to refrain from the transaction or relationship, and where necessary to end an existing one, which is why accounts get closed or frozen while documents are outstanding. That is the general pattern across the industry, and it is why “not stated” here is a real absence rather than a formatting gap.

So the honest position is this: you cannot tell from the published material whether you qualify. If your country matters to you — and if you are one of the people for whom an ordinary bank card is hard to get, which is who I assume you are reading this — then that uncertainty is the answer. You would be signing up on the strength of a documentation page, not a promise.

Two more things about who gets served. Prepaid and debit cards are generally available to adults aged 18 and over, and some issuers set higher limits, so the 13-year platform minimum sits well below the usual age floor for card issuing. And the tiers question has no answer either: a single Sponge Card is described, with no tier structure to compare.

What I would check first, before anything else, is the issuer’s own eligibility language once you are inside the sign-up flow, because that is where country and residence screening actually happens. But I want to be careful not to imply there is a trick to get around any of it. If the issuer declines you, that is the end of it, and anyone offering you a way around identity checks is selling you a problem, not a card.

How to get it

You do not apply for this card the way you apply for a bank card. There is no form, no photograph of your passport on a website, no waiting for a plastic card in the post. The whole process is a sequence of developer steps.

Here is the flow as described. You start a hosted KYC flow from the CLI or SDK, passing a redirect URI. Once that clears, an issuer application follows after approval. The application collects your occupation — or a SOC code — and a set of consent flags: E-Sign, the account-opening privacy notice, the Sponge Card terms, accuracy, and non-solicitation. Then creating the card itself needs a billing address in four fields (line1, city, region, postal_code, country_code), plus your email and phone.

Notice the shape of that billing address. It is four structured fields and a two-letter country code, and it goes straight into the data a merchant’s fraud system reads. I will come back to why that matters, because address verification is the quiet reason a lot of virtual cards get declined — and it is the reason your billing country and your issuer country can end up in a fight.

What is not stated here is as important as what is. The documents accepted and the age proof required are not published. There is a single hosted KYC flow and a single issuer application — no separate KYC tiers, no second level of verification for larger balances, nothing that would tell you at what size of transaction your identity checks get heavier. If you are the kind of person who wants to know whether you will be asked for a passport, a driving licence, a selfie, or a proof of address before you ever spend anything, you will not find that out in advance.

One more thing worth saying plainly: this is not a bank account. Sponge Inc. is the company in the footer, described as “Sponge (YC W26)” — a very young company. The card itself is issued by Rain, which is described as a Visa Principal Member; the issuing bank behind Rain is not named on the pages I read, and the terms are governed by US law. Licences are not stated. When you read about e-money institutions later in this review, hold that thought: in most of these programmes the crypto brand is only a programme manager or a front-end, and the licence that matters belongs to someone else.

What it really costs

This is the section where I have least to tell you, and I would rather say that plainly than dress it up. Issuance is not stated. The fee schedule is not stated. ATM terms are not stated. Daily and monthly limits are not stated. The only guidance on limits is the principle that collateral determines spending power, and there is an endpoint — getSpongeCardStatus — that returns the spending power figure for a card.

There is no tier structure either: one Sponge Card, no Basic/Silver/Gold ladder to compare, which means the comparison shoppers usually do first simply cannot be done here.

Let me show you what that absence costs you in practice, because “no published fees” is not the same as “no fees” and it is definitely not the same as “free”. Take three ordinary payments.

A coffee in a different country. You would expect this to involve a conversion from USDC into whatever the terminal charges, plus a foreign-exchange margin, and you would expect to find both on the card’s page. Neither is there. So you are spending an unknown amount more than the sticker price, by an unknown percentage, and the spread belongs to a company whose bank you do not know the name of.

A monthly subscription. Card subscriptions are usually processed in the card’s own currency and often in the merchant’s billing currency. Card currency is not stated, so you cannot tell whether a renewal three months from now will be converted, at what rate, or at what fee — and there is no way to predict the amount on your statement.

An ATM withdrawal. Unknown. Whether it works at all, what it costs, whether it is offered on a virtual card created for an agent in the first place.

What I can tell you about the general shape of these costs across the category, since Sponge publishes none of it: on-chain network fees for your deposit are charged by the blockchain, not by the card provider — so that cost exists and lands on you regardless of what any issuer says. Beyond that, every crypto card in this space charges something for foreign exchange, and the spread between them is often the largest number on the schedule. Some competitors publish 0% FX on their top tier and “no hidden spread” on stablecoin spending; others publish tiered FX fees. Sponge publishes nothing, which puts it at neither end of that range as far as a reader can tell.

There is also a fee you might expect to find and will not: none is stated for funding, and none is stated for withdrawing collateral back to the wallet. Whether that means the operations are free, or means the fees are simply unpublished, I genuinely cannot tell you. That is the honest state of it.

How to top it up

Top-up is a command, not a payment screen. You move USDC collateral from your Sponge wallet with card fund; the example chain given is base, and the SDK mentions base or solana as an optional chain. That is the entire funding model described. No other coin is supported.

Notice what “top up options” means here. In most of this category, top-up means depositing from an exchange or an on-chain address into a card balance, and the number of coins and networks you can choose from is one of the main ways cards are compared. Sponge has one coin and two named networks. The scoring reflects that exactly, and I think the score is fair rather than harsh: this is a deliberately narrow funding rail, and it is narrow in a way that makes sense for a card whose entire job is to spend a stablecoin balance on behalf of an agent.

The minimum deposit is not stated, and neither is the fee for the top-up.

Getting money back out works the same way round: card withdraw, or withdrawSpongeCard, returns USDC collateral to your wallet, with an amount given to up to two decimal places. Fees are not stated there either.

Two cautions about the mechanics, both general to the category rather than specific to Sponge. First, USDC is not natively issued on every network, and Circle lists it as natively supported on 38 networks as of 16 September 2026, including Ethereum, Solana, Base, Polygon PoS, Avalanche, Arbitrum and Stellar. Sending USDC over a network the provider does not support, or to an address belonging to another network, can lose the funds outright or need manual recovery with no guarantee, because blockchain transfers are irreversible. Getting the network right is not a formality here. Second, stablecoins are not risk-free instruments: USDC is US-dollar-pegged and Circle describes it as redeemable 1:1 for dollars and fully backed by cash and cash equivalents, mostly in an SEC-registered government money market fund, with monthly reserve attestations — but “despite the name, stablecoins are not necessarily stable,” and USDC itself temporarily lost its peg in March 2023. Tether, which is a stablecoin too, says it disables tokens in wallets on the OFAC sanctions list and has frozen funds tied to illicit activity. If your agent is spending real money, your collateral sits in an instrument that has wobbled before and lives inside a system that can freeze.

Paying with it every day

Everyday use is where this card scores lowest, and the reason is structural rather than a matter of taste. There is no consumer app — none was found, and the thing is used via the CLI (npx spongewallet), the SDK and MCP, so store ratings do not apply. There is no Apple Pay, no Google Pay, no ATM story, none of it disclosed.

The physical question is worth answering plainly: the documentation describes a virtual card only, created by createSpongeCard. Physical delivery is not stated. So if you were hoping for something in the post, there is nothing to wait for.

Now, do not read my earlier section on Apple Pay and tokenisation as a promise — I want to be explicit about that, because I can see how a reader could take it the wrong way. The general mechanics are worth knowing for your own protection, not as a workaround. When a card number is replaced by a device-specific number stored in the Secure Element rather than on a vendor’s servers, each payment uses a one-time dynamic code, and an issuer can restrict a tokenised number so it cannot be used on a magnetic stripe, on a phone or on a website. Whether Sponge’s card number can be added to a wallet is simply not disclosed, and I am not going to suggest a route to adding it that the issuer has not offered. If that matters to you, it is a question to ask Sponge before you fund anything.

The everyday-use picture also lacks the parts you would use to judge reliability. Which wallets are supported is not stated. Rewards are not stated. Dispute handling is not stated — and that last one is the one that would matter most if something went wrong with a purchase.

Understand why merchant declines happen, because if you do hit them you will want to know whether it is you or the card. A payment card number starts with a six- or eight-digit issuer identification number, and a merchant’s system can look that up to see which institution and which country issued it. If the issuer country differs from your billing address, some merchants apply extra scrutiny. Some decline prepaid or foreign-issued cards outright for exactly this reason. So a virtual card created through a developer tool, with a billing address of four fields and a country code, is running into a structural disadvantage in the physical world — which is another way of saying that “accepted wherever Visa is accepted”, which is how the homepage describes the card, is a statement about the network and not a promise about individual checkouts.

If you are reading this as someone whose ordinary card gets declined, that structural disadvantage deserves more weight than usual, and my recommendation at the end reflects it.

What can go wrong

Nothing has gone wrong in public. The official record shows no published incidents, and a user-report and news search on 2026-10-21 found nothing. There is no Trustpilot page and no reviews to read. I want to be careful here: a clean record on a card this young and this small is weak evidence of safety, not strong evidence of it. It mostly tells you the card is young.

The company picture is thin in a specific way. Sponge Inc. is behind it, “Sponge (YC W26)” — a YC Winter 2026 company, which places the whole operation in its first years. The card is issued by Rain, “a Visa Principal Member”, which is not the same as naming the bank that actually holds your money. The issuing bank is not named, and the licences are not stated, even though the terms are governed by US law. If you want to know whether your collateral sits behind an e-money licence with safeguarding obligations, or behind a bank deposit with deposit guarantee, the published pages will not tell you — and those are two very different protections.

The security documentation, by contrast, is unusually good, and I want to give credit where it is due. The card number and CVC are returned encrypted with a one-time secret key, and the docs instruct you to treat that key as secret and not to log or persist the decrypted values. That is a sensible design and a sensible instruction. But the custody model itself is not stated, and neither is two-factor authentication nor card freeze — so the parts of security that are about recovering access or stopping a card in an emergency are simply absent from what I could read.

Two general risks that apply to anything in this category, framed so you can weigh them yourself. Control of crypto belongs to whoever holds the private key; with a custodial wallet the provider holds the credentials and you must trust them completely. And a seed phrase — typically 12 to 24 words — is an unencrypted private key, so anyone who obtains it can restore the wallet and spend its funds. Transfers are typically irreversible and a stolen key has no built-in recovery: the network cannot freeze or return what has gone. The history of custodial platforms is not encouraging: Mt. Gox with about US$350 million reported stolen across 2011 to 2014, Coincheck at about US$400 million in 2018, and the Ronin Network at US$625 million in 2022. Holding funds with a custodian carries counterparty and hack risk.

And then there is the risk that has nothing to do with security at all: the programme simply ending. Visa and Mastercard do not issue crypto cards to consumers themselves; they license issuers and can suspend a programme. In November 2022 Visa said it had terminated its global agreements with FTX, and the FTX US debit card programme was wound down. WaveCrest, a Visa issuer that supplied crypto card brands, was suspended by Visa in Europe in 2018, closing every WaveCrest-issued Visa programme. Closures typically run in a predictable order — a notice period, then spending and loading stop, then users are asked to withdraw remaining balances. I am not predicting anything for Sponge. I am telling you that this is a category with a demonstrated shutdown rate, and that your exit route depends on card withdraw continuing to work.

If support matters to you, the contact options are a contact link on the homepage, an account on X at @paysponge, and a Discord.

What to take instead

If the reason you are here is that an ordinary card is hard for you to get, then the honest comparison is not “Sponge versus nothing” — it is Sponge versus other crypto cards that publish their eligibility rules, their fees and their limits. Here is how the alternatives look on exactly the dimensions Sponge leaves blank. These are not endorsements; they are the disclosures you can check.

Three things stand out from that table. First, every alternative publishes something Sponge does not: ether.fi names its jurisdictions and says plainly what happens if yours is not on the list; Anodos publishes a full country list including a long exclusion list; Xplace publishes an annual fee ladder from free to $999. Reading a competitor’s exclusion list is far more useful than reading nothing, because you can check yourself against it in ten seconds.

Second, ether.fi’s warning is the one I would underline for you: “if your jurisdiction is not listed, you will not be able to apply.” That is a sentence I would want to have before signing up anywhere, and Sponge does not offer its equivalent.

Third, on rewards: two of the three publish cashback rates and one publishes nothing. If you are choosing on economics, that is a bigger differentiator than any of the card logos.

Who I would recommend it to

I would recommend this card to exactly one kind of person: a developer who is building an AI agent that needs to pay for things on its own, who holds USDC, who is comfortable with a CLI and an SDK, and who treats the card as one line item in a larger system they already trust. For that person it is a coherent design — a narrow funding rail, a virtual card created programmatically, a documented encryption scheme for the card credentials, and spending power tied to collateral so it cannot overdraw.

I would not recommend it to you if you are trying to solve a banking access problem. And I want to be direct about why, because you are the reader I most expect to be here. You came looking for something that works when an ordinary card is refused, and what Sponge offers is a card that carries the same structural disadvantages as every other crypto card — a foreign or programme BIN that merchants can look up and decline, an address verification system that compares your typed address against issuer records and is generally unavailable for foreign-issued cards, declining authorisations that can leave holds on your funds for up to seven days — with none of the compensating information. You would be taking on all of the risk of this category and none of its disclosures.

The pattern of the missing information is itself the finding. Countries, fees, limits, ATM, card currency, dispute handling, custody model, two-factor authentication, card freeze: not stated, one after another. That is not a card that is hiding something necessarily; it is a young developer product that has documented its software surface thoroughly and its financial surface hardly at all. But you are the one funding the collateral, and a lack of published limits means a limit you discover at the checkout rather than in advance.

If you want to take the general lessons from this category rather than this particular card, they are these. Check the issuer’s jurisdiction list against your own country and citizenship before you apply to anything. Assume a foreign-exchange fee exists, because it does. Keep collateral sized to what you could lose if the programme were suspended, since shutdowns in this industry are documented, not hypothetical. And if you hold a seed phrase, treat it as the money itself, because that is exactly what it is.

Frequently asked questions

What is the Sponge card?

The Sponge card is a Visa payment card issued by Rain, described as a Visa Principal Member, and created through a command-line tool or SDK rather than a website application. It is backed by USDC collateral held in your Sponge wallet, and it is described on the homepage as a credit card accepted wherever Visa is accepted. The intended use is to hand the card number to an AI agent so the agent can spend on your behalf without access to a bank account. It is virtual only in the documentation; physical delivery is not stated.

How does Sponge card work?

You fund USDC from your Sponge wallet with card fund on base or solana, and that collateral sets your spending power — the docs say collateral determines spending power, and getSpongeCardStatus returns the figure. You then create a virtual card with createSpongeCard. Spending draws down that balance. Collateral comes back to the wallet with card withdraw, to two decimal places. Card currency, fees and limits are all not stated on the card’s pages.

Is Sponge a bank?

No, and this is worth being clear about. Sponge Inc. is the company behind the product, described as “Sponge (YC W26)”. The card itself is issued by Rain, “a Visa Principal Member”, and the issuing bank behind Rain is not named. Licences are not stated. The terms are governed by US law. A crypto brand is often only a programme manager or front-end while a licensed issuer holds the actual card programme, so the licence that matters may not be Sponge’s to hold.

Who owns Sponge?

Sponge Inc. owns the platform, and it is a YC Winter 2026 company — so it is a young organisation. The card programme is not owned by Sponge: Rain issues it, and Rain is a Visa Principal Member, which means it licenses from the Visa network rather than being Visa. The underlying bank is not named on the pages I read, and no licences are stated.

Which countries can use the Sponge card?

The card’s public pages do not list any countries. There is no country list at all, so you cannot check your own eligibility in advance. The only eligibility information published is a minimum age of 13 for the service, which is notably below the 18-plus norm for card issuing. Residence and citizenship get screened separately, so “no country list” leaves both questions open.

How much does the Sponge card cost?

Nobody can tell you from the published material, and I mean that literally: cost, monthly fee, conversion fee, FX fee, ATM charges, daily limit and monthly limit are all listed as not stated. There is no tier structure either — a single Sponge Card is described. What is published is the principle that collateral determines spending power, and the endpoint that returns it. Any number I gave you here would be my invention, not Sponge’s.

Is Sponge card safe?

The security documentation is the strongest part of the product: the card number and CVC come back encrypted with a one-time secret key, and the docs tell you to treat it as secret and never log or persist the decrypted values. Against that, the custody model is not stated, two-factor authentication is not stated, card freeze is not stated, and there is no dispute process described. No official incidents are published and no complaints were found, which for a card this young is weak evidence.

Can I add the Sponge card to Apple Wallet or Google Pay?

Not disclosed, and I would not try to work around the answer. Apple Pay and Google Pay are both listed as not stated for this card. In general, tokenised wallet payments replace the card number with a device-specific number held in the Secure Element rather than on the vendor’s servers, and an issuer can restrict that number by channel. Whether Sponge supports it is a question for Sponge before you fund anything.

How do I withdraw money from Sponge?

With card withdraw, or withdrawSpongeCard, which returns USDC collateral to your wallet with an amount given to up to two decimal places. Fees for that operation are not stated. The general caution applies to the way out as much as the way in: blockchain transfers are irreversible, so sending on a network the provider does not support, or to an address belonging to another network, can lose funds or need manual recovery with no guarantee.

Is the Sponge card worth it?

For a developer wiring a USDC-funded card into an AI agent, plausibly yes — the funding rail is narrow on purpose and the credential handling is documented well. For a person trying to pay for ordinary things, I would say no on the evidence available: no country list, no fees, no limits, no physical card, no consumer app, no dispute process. The information needed to judge it is simply not published, and that absence is the answer.

Content created by the «cryptocardtop» team