Sponge card, reviewed: a virtual Visa funded by USDC, built for developers and their agents

Updated October 2026

Licensed

Available in US

Fast payouts

18+ Only

The Sponge card is a credit card issued by Rain, a Visa Principal Member, that you can only create through a command-line tool or an SDK. The company behind it, Sponge Inc. (a Y Combinator W26 startup), describes the card as a way to give a software agent a Visa number backed by USDC collateral sitting in a Sponge wallet. There is no phone app, no physical card mentioned in the documentation, and very little on the public pages about countries, fees or limits. That last point is the one that matters most if you are trying to decide whether to use it, and it is the reason this review exists.

I went through Sponge’s homepage, the CLI and SDK reference, the platform terms and the issuer language, and compared what I found against a handful of alternatives that occupy roughly the same niche: Xplace, ether.fi and the Anodos Card. The picture that came together is of a product that is very interesting to a small audience of developers running AI agents, and confusing or off-putting for almost everyone else. Below, I walk through who Sponge will and will not work for, how you actually get the card, what it costs (or rather, what Sponge does not say it costs), how funding works, what daily life with the card looks like, the risks worth weighing, and what other cards make more sense for most readers.

Conditions current as of October 28, 2026.

Verdict

I am giving the Sponge card 4.9 out of 10. The score reflects a product that does one narrow thing well — give an AI agent a USDC-backed Visa number — but is closed about the basic information a normal cardholder expects: which countries are supported, what the fees are, what the limits are, and whether the card works in Apple Pay or at an ATM. The score table below explains where the marks came from.

The audience note from my research is worth saying out loud: the card is built for developers giving AI agents a USDC-backed Visa. If that is what you are doing, the missing information may not matter to you, because you will be spending programmatically and the agent will not need a contactless tap or a cash withdrawal. If you are an ordinary person looking for a crypto card to use at a coffee shop, the gaps in the public information will probably be a deal-breaker.

In short, the card suits a developer who already lives in a terminal and is comfortable funding an agent with USDC. It does not suit someone who wants a phone app, a physical card, clear fees or a published list of countries where the card works.

Who will they give you the card

Here is the most uncomfortable fact in this review, and the reason it sits at the top: Sponge’s public pages do not state which countries the card is available in. The card pages do not include a country list, and the platform terms do not set residency or citizenship rules. The only rule I could find anywhere is a minimum age of 13 for the service, and even that is the platform’s age floor, not a card-issuance rule. Everything else about who qualifies — which documents, which addresses, which nationalities — is not stated.

What this means for you in practice: until you start the sign-up flow, you will not know whether you qualify. The only public signal is that the card is issued by Rain, which is a Visa Principal Member, and that the terms are governed by US law. Neither of those tells you whether your country is on a list.

For comparison, a typical crypto card in 2026 publishes either a long list of supported countries or a short list of excluded ones. ether.fi, for example, lists the jurisdictions it issues in and the 31 US states where it can ship a physical card, and says outright that if your jurisdiction is not on the list, you will not be able to apply. The Anodos Card publishes both a “150+ countries” headline and a list of residence-based exclusions. Sponge publishes neither.

If you live in a country that other crypto cards routinely exclude — Belarus, Cuba, Iran, North Korea, Russia, Syria, Venezuela, and so on — you should assume Sponge’s list is at least as restrictive. The card pages simply do not give you the information to rule yourself in or out, so the burden falls on you to start the application and find out.

How to get it

Getting a Sponge card is unlike getting any other card I have reviewed. There is no app to download, no sign-up form on a website, and no application you can fill in on a phone. The whole flow runs from a terminal.

You start by installing the Sponge wallet CLI (npx spongewallet) or pulling the SDK into your own code, then call into the Sponge APIs. The card-creation flow has two distinct steps. First, you trigger a hosted KYC flow from the CLI or SDK, with a redirect URI that takes the user (or the agent’s user) to Sponge’s identity check. The form collects an occupation or SOC code, an email and a phone number, plus the four standard consent flags: E-Sign, the account-opening privacy notice, the Sponge Card terms, and confirmations of accuracy and non-solicitation. Second, once KYC is approved, the same CLI or SDK submits an issuer application, and the card is minted with a billing address (line1, city, region, postal_code, country_code), the email and the phone collected earlier.

A few practical points follow from this:

• The documents Sponge accepts, and any minimum or maximum age, are not stated in the public pages.
• There is a single KYC level, not a tiered model, so the same flow applies to everyone.
• The card you get is a virtual card only, produced by a createSpongeCard call. The homepage describes it as a credit card issued by Rain, accepted wherever Visa is accepted, but it does not describe a physical card. Physical delivery is not stated.
• The card’s currency is not stated.

For an AI-agent use case, this is exactly right. You write a script, the script creates the card, and the agent now has a Visa number. For a human user, the friction is real: you need a computer, a wallet already funded with USDC, and the patience to set up a developer environment just to spend money in a shop. If the agent is paying on your behalf, you still need to do the KYC once, in a browser, with a redirect from the CLI.

Fees

This is the section I would most like to be able to write, and the one where Sponge gives the least to work with. The public pages do not state the issuance fee, the monthly fee, the conversion fee, the FX fee, the ATM fee, the daily limit or the monthly limit. The only number-adjacent fact I can report is that “collateral determines spending power”: the SDK exposes a getSpongeCardStatus call that returns the card’s current spending power, which is set by how much USDC you have moved into the card. There is no published tier system — a single Sponge Card is described, and that is it.

The table below records exactly that: the values that are not stated, against the values that are.

A few practical consequences, because the absence of published numbers is itself a fact you have to deal with:

• A coffee in another country. You buy a coffee in, say, London, paying in pounds. The card number says nothing about how the USDC gets converted into pounds, and Sponge does not publish a conversion fee or an FX margin. If the card currency is USD, you can be sure there is at least a Visa cross-border fee in there, plus whatever margin Rain takes on the conversion. You will not see the rate in the docs, and you will only learn it from your statement.
• A cash withdrawal at an ATM. ATM fees are not stated, and there is no public confirmation that ATM withdrawals are supported at all. Because the card is documented as virtual only, and there is no mention of a physical card, you should assume the safest case: ATM cash is not part of the product.
• A monthly subscription. There is no monthly fee mentioned. If there is one, you will only find out when you see it on your statement or in the SDK’s status output. There is no tier system published, so a single card is the only option.

I would compare this to a card like Anodos, which publishes an FX fee of 1.50% on its Free tier, 1.25% on Pro and 1.00% on Elite, and a clear monthly or annual fee per tier, or Xplace, which publishes a tiered annual fee ($0 / $99 / $249 / $999) and a “0% FX on Platinum” headline. ether.fi publishes a per-tier FX margin (0.5% / 0.25% / 0% / 0%). Sponge publishes none of that. If clear pricing matters to you, the comparison is already over.

Top up

Funding the Sponge card is the most clearly documented part of the product, because the funding flow is the same as the agent’s payment flow: it is just a USDC transfer. The Sponge SDK exposes a card fund command that moves USDC from the user’s Sponge wallet into the card’s collateral position. The example in the docs uses the Base network; the SDK also lists Solana as an optional chain. No other coins are accepted, and no other networks are listed.

To reverse the flow, the same SDK exposes card withdraw (also written withdrawSpongeCard), which returns USDC from the card’s collateral back to the wallet. The amount can be specified up to two decimal places. Fees on either direction are not stated.

A few consequences worth saying out loud:

• The card is single-collateral. If you hold BTC, ETH or anything other than USDC in your Sponge wallet, you cannot fund the card with it. You would need to convert to USDC first, on whatever venue you use, and pay that venue’s fees. Sponge does not offer in-wallet conversion.
• The card is single-network. The funding is USDC, but only on Base or Solana in the SDK examples. USDC exists on many other networks, and if you send it on the wrong one — say Polygon or Avalanche — the funds can be lost. Stablecoin transfers are irreversible, and Circle’s policy is to recover such losses only with manual support intervention, which is not guaranteed.
• The minimum top-up is not stated. The maximum is set by the size of the USDC balance in your wallet. The daily and monthly limits are not stated either, so there is no published ceiling to plan around.

In practical terms, topping up is a developer action, not a consumer action. You write a script that calls card fund with an amount, the wallet checks that you have that much USDC, and the transfer goes through. If you are running an agent, this is the right shape: a programmatic call that takes a number and returns a funded card. If you are a person, the friction is that there is no “Add money” button on a phone, and no “buy USDC” button inside Sponge — you have to bring USDC in from somewhere else.

For context, Xplace accepts BTC (cbBTC), ETH (wETH), SOL, JitoSOL, USDC, USDT and xStocks on Solana, and the Anodos Card accepts a list of “Supported Blockchains” (Ethereum, Polygon, Optimism, Arbitrum, and others) but does not list the coins. ether.fi’s collateral is “any supported token” deposited into the user’s Safe. None of those is as narrow as Sponge’s USDC-on-Base-or-Solana, and all of them are at least as clear about what they accept.

Paying with it every day

The daily-use story is the section where the card scores lowest, and once you read the documentation it is easy to see why. The card is virtual, it is created by code, and the documentation is silent on the consumer channels that most readers will care about: Apple Pay, Google Pay, contactless tap, online checkout with 3-D Secure, recurring subscriptions, and cash at an ATM.

What I can tell you, with the same caveats as above:

• Online checkout. The card number and CVC are returned by the SDK encrypted with a one-time secret key, and the docs tell you to treat the decrypted values as secret and not to log or persist them. So in principle, you can use the card on a website by decrypting the number and typing it in. In practice, there is no consumer flow for this — the intended pattern is that an agent types the number in, on your behalf.
• In-store payment. There is no mention of a physical card, a tokenised mobile wallet or a contactless tap. You can type the card number into a payment terminal that supports manual entry, but that is rare at a checkout.
• Recurring payments and subscriptions. Nothing in the docs addresses them directly. Because the card is prepaid against USDC collateral, it cannot overdraw, which means a subscription that prices higher than the balance will simply be declined. There is no published grace period.
• Cashback and rewards. The public pages do not state a rewards programme. There is no mention of cashback, points or any per-tier benefit.
• Customer support. The homepage links to a contact form, and the company is on X as @paysponge and on Discord. The docs do not promise a phone line, an in-app chat or a response-time guarantee.
• Disputes and chargebacks. Not stated. Prepaid cards, as a rule, have weaker dispute rights than credit cards, so do not assume the same protections.
• A consumer app. None was found. The product is the CLI (npx spongewallet), the SDK, and an MCP integration. There are no app-store ratings because there is no app.

If you are an AI agent or a developer building one, none of this is a problem: the card is a programmatic number, and the agent does not need a contactless tap. If you are a person, the absence of mobile-wallet support is a serious limitation, because an increasing share of in-store payment runs through Apple Pay or Google Pay, and a virtual card that you cannot tokenise into your phone is hard to use in a shop.

For comparison, Xplace runs as a consumer app with a virtual and physical card, ether.fi ships both virtual and physical cards in many countries, and Anodos issues physical and virtual cards with a published tiered programme. All three have at least the option of a physical card and a mobile-wallet path. Sponge, at the time of writing, has neither.

What can go wrong

A crypto card has a standard set of risks, and Sponge adds a few of its own. I would group them as follows.

Counterparty risk on the issuer. The card is “issued by Rain, a Visa Principal Member”, per the homepage. The issuing bank is not named in the public materials, and the company’s licences are not stated. Rain is a real Visa Principal Member, and Visa Principal Membership is a higher bar than a sponsored programme, but the absence of a named bank and licence list means the regulatory safety net under your balance is not visible from the outside. If Rain’s programme were to be suspended by Visa, you would have to rely on Sponge’s ability to wind down the card and return your USDC.

Counterparty risk on Sponge itself. The company is Sponge Inc., associated with Y Combinator’s W26 batch. That is a young company, and the documentation does not describe how the USDC collateral is held, whether it is segregated, or what happens to it in the event of Sponge’s insolvency. The custody model, two-factor authentication and a card-freeze feature are not stated. With a self-custody-linked card, the funds normally stay in your own wallet until the moment of payment, but the docs do not explicitly say that is what Sponge does.

Stablecoin risk. The card is collateralised by USDC. USDC is, as Circle describes it, redeemable 1:1 for USD and backed mostly by cash and short-dated US Treasuries, with monthly attestations. It is also not immune to depegs: in March 2023, USDC briefly lost its peg during the US banking crisis, and a card that settles in USDC collateral would have felt that. Tether’s USDT has its own reserve history. Stablecoins are not bank deposits, and a deposit guarantee does not apply.

Network and transfer risk. The only documented funding chains are Base and Solana. If you send USDC on the wrong network, the funds can be lost. Stablecoin transfers are irreversible, and Circle’s policy is to assist with recovery only through manual support, with no guarantee. Sponge does not publish a “wrong network” recovery procedure.

Disclosure risk. The card number and CVC are returned encrypted with a one-time secret key. That is good practice, but it puts the burden on you (or your agent) to handle the secret correctly. If the decrypted values are logged or leaked, the card can be used by anyone with the number and CVC. There is no public mention of a card-freeze control.

Programme-closure risk. Card programmes are shut down from time to time. In November 2022, Visa terminated its global agreements with FTX and the FTX US debit card was wound down. WaveCrest, a Visa issuer that supplied several crypto card brands, was suspended by Visa in Europe in 2018, closing all WaveCrest-issued Visa programmes. A 2026 closure reported by a secondary source said new loads ended 8 July 2026 and spending stopped 8 August 2026 (single source, unconfirmed). None of these events involved Sponge, but they are the kind of event a USDC-backed Visa programme can be exposed to.

Complaints and incidents. As of 2026-10-21, I could not find user reports or news incidents involving Sponge. There is no Trustpilot page I could locate. That is partly because the product is small, partly because there is no public cardholder app where users would leave reviews, and partly because the audience is developers rather than consumers.

None of this means Sponge is unsafe in any absolute sense. It means the safety net is not as visible as on cards that name their issuing bank, list their licences, publish their custody model and ship a consumer app with 2FA and a freeze button.

What to take instead

If the gaps above are deal-breakers — and for most readers, they probably are — here is how the alternatives stack up. I picked three cards that occupy roughly the same niche: a Visa issued against crypto or stablecoin collateral, with a published country or exclusion list, a published fee or tier structure, and some form of consumer app or portal. None of them is built for AI agents in the way Sponge is, but that is the point: they are built for people.

Xplace — 7.7 / 10. Xplace issues Visa cards in 160+ countries, with virtual cards available instantly and physical cards shipping in 7-14 business days (Silver is live; Gold and Platinum are announced). It has a Cash mode that spends a stablecoin balance and a Credit mode that borrows against holdings. The annual fee is published: $0 on Basic, $99 on Silver, $249 on Gold, $999 on Platinum. “0% FX” is published on Platinum, and stablecoin spending is described as having “no hidden spread”. Rewards are 0.5% / 1% / 1.5% / 2% cashback by tier, paid in USDC. Funding accepts BTC (cbBTC), ETH (wETH), SOL, JitoSOL, USDC, USDT and xStocks on Solana. The card currency is not stated, and the FX fee for non-Platinum tiers is not published. For a developer who also wants a consumer-grade card, Xplace is a reasonable answer.

ether.fi — 6.7 / 10. ether.fi is the most jurisdictionally transparent of the three. It lists the countries and 31 US states it issues in (Andorra, Argentina, Australia, Austria, Brazil, Canada, Cyprus, Czech Republic, France, Germany, Hong Kong, Hungary, Italy, Japan, Mexico, Poland, Portugal, Singapore, Spain, Switzerland, UAE, UK, plus the US states), with Belgium and Ireland eligible only for virtual cards. It says outright that if your jurisdiction is not listed, you cannot apply, and blocks ATM withdrawals in North Korea, Iran and Myanmar. The cost structure is per tier: Core normally gets one free virtual card, with a $10 fee for the first virtual card in a list of regional-pricing countries (Argentina, Angola, Bangladesh, Bolivia, Colombia, Ecuador, Indonesia, Malaysia, Mexico, Nigeria, Peru, Philippines, South Africa, Thailand). The FX margin is per tier: 0.5% on Core, 0.25% on Luxe, 0% on Pinnacle and VIP, with a 0% FX article for EUR purchases. The conversion fee is not stated. Cashback is paid in ETHFI (locked 7 days), with 3% up to monthly caps and lower ladders for EUR. ATM, cash advances, P2P and gambling MCCs are excluded. Funding accepts “any supported token” into a user Safe. If your country is on ether.fi’s list, the pricing transparency and the credit-mode structure are notable.

Anodos Card — 7.2 / 10. Anodos publishes a residence-based exclusion list (Belarus, China Mainland, Cuba, India, Iran, Iraq, Israel, Myanmar, Nepal, Nicaragua, North Korea, Russia, Syria, Turkey, Ukraine, Venezuela, Vietnam), says Chinese and Hong Kong passports are not eligible, and applies “enhanced review + explicit approval” to Nigeria, Sri Lanka and South Africa. Spending is blocked in Cuba, Iran, North Korea, Russia, Syria, Ukraine and Venezuela. There is a separate US programme in 30 states. The fee tiers are: Free ($0), Pro ($9.99/mo or $99.90/yr) and Elite ($29.99/mo or $299.90/yr). The FX fee is 1.50% / 1.25% / 1.00% by tier, and the non-card currency-exchange fee is 1.00% / 0.75% / 0.50%. Interest is 0% APR on purchases currently, with a late-payment fee up to $40 and a $29 returned-payment fee. The first virtual card is free, with extra virtual cards priced at $20 / $10 / $10 by tier. Funding accepts collateral on Ethereum, Polygon, Optimism, Arbitrum and others. Rewards are not stated. For a consumer who wants a published fee ladder and a clear exclusion list, Anodos is the most readable of the three.

If you came to Sponge specifically because you are building an agent and need a programmatic Visa number funded by USDC, the alternatives are not direct replacements. If you came to Sponge for any other reason, one of the three above is likely to fit your situation better, because they all publish the information Sponge does not.

Who I would recommend it to

Pulling the threads together, I would recommend the Sponge card to a narrow audience and only that audience. If you are a developer building an AI agent that needs to spend USDC over a Visa rail, and you are comfortable running a CLI or wiring an SDK, Sponge is genuinely useful: the API shape, the USDC-on-Base-or-Solana funding, the encrypted card-number delivery and the issuer relationship with Rain are all sensible for that use case. The score reflects that the product does its intended job, even though it is not for most people.

I would not recommend it to a reader who:

• wants a physical card, or a card in Apple Pay or Google Pay;
• needs a published list of supported countries before they apply;
• cares about a clear fee schedule, an FX margin or an ATM fee;
• wants a consumer app with 2FA, a card-freeze button and a rewards programme;
• needs to use cash at an ATM or pay in a shop with a contactless tap;
• holds assets other than USDC, or wants to top up from a chain other than Base or Solana.

For any of those readers, the missing information is itself the answer: a card that does not publish its countries, fees, limits, custody model or licensing is a card that does not give you what you need to plan around it. Sponge is a developer tool that happens to be a card. If that is the shape of product you are looking for, you will probably already know it. If it is not, one of the alternatives in the section above is a better fit.

Frequently asked questions

What is the Sponge card?

The Sponge card is a credit card issued by Rain, a Visa Principal Member, that you can only create through a command-line tool or an SDK. It is backed by USDC collateral sitting in a Sponge wallet, and the company behind it, Sponge Inc., describes it as a way to give a software agent a Visa number. The card is virtual only, and the public pages do not publish a country list, fees or limits.

How does the Sponge card work?

You install the Sponge wallet CLI or pull the SDK, run a hosted KYC flow with a redirect URI, then submit an issuer application from the same CLI or SDK. Once approved, the card is minted and the SDK exposes card fund to move USDC from the wallet into the card and card withdraw to move it back. Spending is debited against the USDC collateral, and the SDK’s getSpongeCardStatus returns the card’s current spending power.

Is Sponge a bank?

No, Sponge is not a bank. Sponge Inc. is a Y Combinator W26 startup that markets and operates the card, and the card itself is issued by Rain, a Visa Principal Member. The issuing bank and Sponge’s licences are not stated in the public materials. Sponge’s terms are governed by US law.

Who owns Sponge?

Sponge Inc. owns the Sponge product. The company is associated with Y Combinator’s W26 batch, and the company’s footer identifies it as “Sponge (YC W26)”. The card is issued by Rain, a Visa Principal Member, under Visa’s rules. The ownership structure, investors and any parent companies are not stated in the public materials I reviewed.

Is the Sponge card available in my country?

The card’s public pages do not publish a list of supported or excluded countries. Until you start the sign-up flow, the only public signal is that the card is issued by Rain, a Visa Principal Member, and that the terms are governed by US law. If your country is one that other crypto cards commonly exclude, you should assume Sponge’s list is at least as restrictive. The platform’s terms only set a minimum age of 13 for the service.

How do I get a Sponge card?

You get a Sponge card by installing the CLI (npx spongewallet) or pulling the SDK, then calling the hosted KYC flow with a redirect URI. After KYC is approved, the same CLI or SDK submits an issuer application with a billing address (line1, city, region, postal_code, country_code), an email and a phone number. Once the issuer approves, the virtual card is created and the SDK returns the encrypted card number and CVC.

Does the Sponge card require KYC?

Yes. The card-creation flow starts with a hosted KYC step, where the form collects an occupation or SOC code, an email, a phone number and the standard consent flags (E-Sign, account-opening privacy notice, Sponge Card terms, accuracy, non-solicitation). After KYC is approved, the same CLI or SDK submits an issuer application. There is a single KYC level, not a tiered model. The documents accepted and the age requirement are not stated.

How do I top up the Sponge card?

You top up the Sponge card by calling the SDK’s card fund command, which moves USDC from your Sponge wallet into the card. The example in the documentation uses the Base network, and the SDK also lists Solana as an optional chain. No other coins or networks are documented. To reverse the flow, card withdraw returns USDC from the card’s collateral back to the wallet, in amounts up to two decimal places. Fees are not stated.

Is the Sponge card free?

The public pages do not state the issuance fee, the monthly fee, the conversion fee, the FX fee or the ATM fee. There is no published tier system, and the only number-adjacent fact I can report is that “collateral determines spending power”: the SDK’s getSpongeCardStatus returns the card’s spending power, which is set by how much USDC you have moved into the card. Without a published fee schedule, you cannot tell from the outside what the card actually costs to hold and use.

Is the Sponge card safe?

The card is virtual only, and the SDK returns the card number and CVC encrypted with a one-time secret key, with instructions to treat the decrypted values as secret and not to log them. That is good practice. The public pages do not state the issuing bank, Sponge’s licences, the custody model, whether two-factor authentication is available, or whether a card-freeze control exists. As of 2026-10-21, I could not find user complaints or news incidents involving Sponge, and there is no Trustpilot page. The safety net is less visible than on cards that name their issuer, list their licences and ship a consumer app with 2FA and a freeze button.

Content created by the «cryptocardanswers» team