Why AI Companies Sell Through a Merchant of Record?

An AI product launches on a Tuesday and has paying customers in forty countries by Friday. That is not an exaggeration of the category, it is the normal distribution pattern for software sold through an API or a web app.

It also creates a tax and compliance problem no other business model generates this quickly. So why do AI companies turn to a Merchant of Record?

Four characteristics of the category explain it.

Global demand from day one. There is no domestic launch phase. The first hundred customers sit across a dozen jurisdictions, each with its own registration thresholds and filing obligations.

Digital services tax treatment. AI subscriptions and API credits fall under digital services rules in the European Union, the United Kingdom and a growing list of countries. VAT applies at the customer’s rate, which means tracking rates across every market you sell into.

Usage-based billing. Metered pricing and credit top-ups multiply transaction volume, and every transaction carries its own tax determination rather than a single monthly event.

Elevated dispute ratios. Consumer AI subscriptions attract the pattern that generates chargebacks: trial conversions, forgotten renewals and statement descriptors nobody recognises.

That last point ends more payment relationships than any other. Card schemes monitor dispute ratios continuously, and crossing a threshold puts your processing at risk regardless of your revenue.

A Merchant of Record absorbs all four by becoming the legal seller: tax obligations, chargeback liability and acquirer relationships move off your side entirely.

At InflowPay, that transfer runs non-custodially, which means funds never touch our balance sheet and settlement completes in four days across 80+ countries.

In this article we break down each reason.

The Tax Problem AI Companies Hit Immediately

Most software businesses discover tax obligations gradually. AI companies hit them in the first month, and the reason is distribution.

A product launched on Product Hunt or circulated on X reaches a global audience instantly. There is no domestic phase during which you learn your home jurisdiction before expanding.

Three obligations arrive together.

EU VAT on digital services. Selling to European consumers means charging VAT at the customer’s rate, not yours. Twenty-seven member states, twenty-seven rates, and a filing obligation through the OSS scheme that begins once you cross the threshold.

The rate applied depends on where your customer sits, which means your billing system needs to determine location reliably and apply the correct rate per transaction.

UK VAT operates separately since Brexit, with its own registration and filing regime.

US sales tax nexus builds state by state. Economic nexus thresholds vary, several states now tax SaaS explicitly, and crossing a threshold triggers registration and filing in that state specifically.

Why usage-based billing makes this harder?

Subscription businesses process one transaction per customer per month. AI companies with metered pricing or credit top-ups process many.

Every one of those carries its own tax determination. A customer buying credits three times in a week generates three taxable events, each requiring correct rate application and correct reporting.

The compliance surface scales with transaction count rather than with customer count, which is the opposite of what most finance teams model when forecasting.

The timing problem

Registration takes weeks. Thresholds are crossed in days.

By the time a fast-growing AI company recognises it has obligations in a jurisdiction, it has frequently been accruing them for a quarter.

Why Chargeback Ratios Threaten AI Subscriptions?

Consumer AI products combine four characteristics that generate disputes faster than almost any other software category.

Trial conversions. A free tier converting to a paid plan produces a charge the customer did not actively initiate that day. Trials that auto-convert without a clear reminder generate disputes from customers who simply forgot.

Impulse signup patterns. AI tools are frequently purchased in a moment of curiosity rather than after evaluation. The gap between buying intent and the first renewal is where regret lives.

Unrecognised descriptors. A customer who subscribed to a product by one name and sees a parent company, a legal entity or an abbreviation on their statement disputes the charge. This is the single largest preventable cause in the category.

Rapid product iteration. Features change, limits adjust, pricing evolves. Customers who signed up for one thing and encounter another have a legitimate grievance that arrives as a chargeback.

Why the ratio matters more than the losses?

The disputed amounts are rarely the problem. A twenty dollar subscription reversed costs twenty dollars plus a dispute fee.

The ratio is what ends businesses.

Card schemes monitor disputes as a percentage of transactions, continuously. Crossing a threshold triggers a monitoring programme with monthly fees, remediation requirements and a timeline to reduce the number.

Fail to reduce it and the merchant account is terminated. Termination can bring a MATCH listing, a scheme database consulted during underwriting that makes obtaining a new account considerably harder for years.

The scaling trap

Growth makes this worse rather than better.

Dispute ratios tend to rise during rapid acquisition, because new customers dispute at higher rates than established ones. The period when an AI company is scaling fastest is precisely when its ratio is most exposed.

What a Merchant of Record Changes for AI Companies?

The transfer is legal rather than technical, which is why it addresses problems tooling cannot.

Tax obligations move entirely

VAT, sales tax, OSS, IOSS and GST attach to the seller of record. When a Merchant of Record occupies that role, registrations and filings become its responsibility across every jurisdiction where your customers sit.

This is not outsourced compliance. The obligation does not exist on your side at all.

For a company whose customer base spans forty countries by month three, that removes a surface that was growing faster than the team could register for it.

Chargeback liability moves

Disputes land on the MoR’s account rather than yours, which means the ratio attaches to them.

This is the change that matters most for consumer AI. The mechanism that terminates merchant accounts during rapid growth no longer points at your business.

Prevention remains yours, and any provider will raise it with you if your rate climbs. But a bad quarter stops being an existential event.

Acquirer relationships move

AI companies sit in categories acquirers frequently treat as elevated risk: new, fast-scaling, consumer subscriptions, occasionally content-generating.

A single acquirer declining your category stops your processing entirely. A Merchant of Record routing across multiple acquirers turns that into an acceptance rate change rather than an outage.

Launch speed changes

Obtaining merchant accounts across regions takes months. An AI company reaching global distribution in weeks cannot wait for that.

At InflowPay, onboarding completes in under 24 hours, with settlement in four days across 80+ countries and 90 to 95% acceptance through multi-acquirer routing.

What to Check Before Choosing a Merchant of Record?

Six checks separate providers more reliably than published rates. Most teams run two.

1. Establish where funds sit before settlement
2. Confirm the settlement period and its exceptions
3. Verify whether chargeback absorption is contractual
4. Request the prohibited categories list in writing
5. Check usage-based billing support
6. Read the exit terms before you need them

The first check determines everything else. Where does my revenue sit between the sale and my payout?

Most Merchants of Record collect into their own accounts before settling. During that window, a risk review or compliance hold on their side interrupts your cash flow exactly as a processor freeze would.

You moved the exposure rather than removing it. A non-custodial structure answers this differently, because nothing is held.

The third check requires precise wording. Is liability absorbed contractually or at your discretion? Those are very different commitments described in similar language, and the distinction only surfaces when you need it.

The fourth check matters specifically for AI companies. Request the prohibited categories list in writing, along with how policy changes are communicated.

AI products sit in a category several providers treat cautiously, and policies here are evolving faster than in established verticals. A provider unwilling to supply the list has answered a different question usefully.

The fifth check is category-specific. Usage-based billing and credit top-ups are not universally supported, and discovering a limitation after integration is expensive. Confirm metered pricing, prepaid credits and overage handling explicitly.

The sixth check is the one nobody runs. Read the exit terms while you are happy. Notice period, funds in transit, reserve duration after termination, and whether you can export transaction history.

Exit terms are negotiated at signature and discovered at departure, which is the wrong order.

Conclusion: The Model Fits the Category

AI companies adopt the Merchant of Record model earlier than most software businesses, and the reason is structural rather than preferential.

Distribution is global from day one. There is no domestic phase during which you learn one tax regime before expanding. The first hundred customers sit across a dozen jurisdictions, each with its own thresholds and filing calendar.

Usage-based billing multiplies the surface. Metered pricing and credit top-ups generate many transactions per customer, and each one carries its own tax determination. Compliance scales with transaction count rather than customer count, which is the opposite of what most finance models assume.

Dispute ratios rise precisely during growth. Trial conversions, impulse signups and unrecognised descriptors produce the pattern card schemes monitor. Crossing a threshold triggers a programme that ends in termination, and a MATCH listing follows the business for years.

A Merchant of Record addresses all three by becoming the legal seller. Tax obligations do not exist on your side. Dispute ratios attach to their account. Acquirer relationships are theirs, routed across several rather than resting on one.

Before choosing a provider, three questions matter more than the rate.

Where do funds sit between the sale and your payout? Most Merchants of Record hold your revenue on their balance sheet before settling, which means you relocated the exposure rather than removing it.

Is chargeback absorption contractual or discretionary? The distinction only surfaces when you need it.

Does the provider support usage-based billing and credit top-ups? Not all do, and discovering it after integration is expensive.

At InflowPay, funds never pass through our balance sheet, which makes a freeze structurally impossible rather than merely unlikely. 4% + $0.35 all-in, four-day settlement, 80+ countries, and onboarding in under 24 hours.

Learn more at inflowpay.com.