Every payment processor in the world charges a fee. Stripe takes 2.9% plus 30 cents. PayPal takes roughly the same. Square, Adyen, Braintree, every name you have heard takes a cut of every transaction they touch.
HMD Pay charges zero. Nothing. Not as a promotion, not as a loss leader, not as a "free for the first six months" marketing exercise. The model is zero fees, permanently.
The first thing anyone says when they hear this is: "That cannot possibly work."
Three years in, it works.
To understand why zero fees make sense, you need to understand what payment processors actually sell.
The obvious answer is: they process payments. Merchant accepts a card, processor handles the communication between the card network, the acquiring bank, and the issuing bank. Money moves from customer to merchant. The processor takes a percentage for making that happen.
But the actual cost of processing a transaction is tiny. The infrastructure is automated. The card networks charge interchange fees that the processor passes through. The processor's own cost per transaction, the compute, the bandwidth, the fraud checks, comes to fractions of a penny. The 2.9% that Stripe charges is not the cost of processing. It is the cost of processing, plus the cost of everything else Stripe does, plus profit.
That margin, the gap between cost and price, is where the entire industry lives.
Our question was: what if we collapsed that margin to zero?
The conventional wisdom says this is suicide. If your revenue is a percentage of transactions and you set that percentage to zero, your revenue is zero. Simple maths.
But the conventional wisdom assumes that per-transaction fees are the only possible revenue source. They are not.
Payment processing involves holding money. When a customer pays, the money does not teleport instantly to the merchant. It sits in the processor's accounts, briefly, during settlement. At low volume this is irrelevant. At high volume, even a brief holding period on aggregate funds generates meaningful interest income. This is not a novel idea. Banks have operated on this principle for centuries. No payment processor talks about it, because admitting that float income exists undermines the narrative that the per-transaction fee is paying for the processing.
Zero fees also attract volume faster than any marketing campaign. When you charge nothing, the integration decision becomes trivial for merchants. There is no fee comparison to run, no margin impact to model, no finance team approval to seek. The path from "should we try this?" to "yes" shortens to nearly nothing. Volume compounds quickly.
And volume itself generates data. Transaction patterns, conversion rates, average basket sizes across industries and regions. At sufficient scale, aggregate data has its own value.
Three revenue sources keep HMD Pay running.
Float income covers baseline operations. The interest rate environment affects how much, but the principle works whenever rates are above zero. At the transaction volumes we process, the numbers are meaningful.
Premium services sit on top of the free tier. Basic processing is free. Detailed analytics, custom reporting, priority settlement, and dedicated support are paid features. The free tier brings merchants in. The premium tier monetises the ones who need more.
Cross-division value is the third and least obvious source. HMD Corporation's other products use HMD Pay internally. No third-party processing fees on any transaction across any HMD property. This does not generate revenue for the payment division specifically, but it reduces costs across the organisation. The holding company captures the value.
Running a zero-fee processor forces a kind of engineering discipline that fee-charging processors can afford to ignore.
When your revenue per transaction is zero, every penny of infrastructure cost matters. There is no transaction fee to absorb inefficiency. If your servers are over-provisioned, it comes out of your margin. If your fraud detection generates too many false positives requiring manual review, it comes out of your margin. If anything at all is wasteful, you feel it immediately.
This constraint produced some of our best engineering decisions. Transactions route dynamically to the cheapest processing rail based on type, currency, and destination. Cryptocurrency, IBAN, and card payments each have different cost profiles, and the routing engine optimises in real-time. KYC, fraud detection, and compliance checks are automated as far as regulation permits. Manual review is reserved for genuine edge cases. Infrastructure autoscales aggressively. We pay for compute at the moment of use, not for idle capacity.
These are things every payment processor should do. Zero fees made them mandatory.
If this model works, why does nobody else do it?
Stripe generates billions from transaction fees. Switching to zero-fee would destroy that revenue overnight, with the volume and float benefits appearing gradually over months. No board would approve that transition. The incentive structure of an established company makes it impossible.
The model also requires significant volume before the alternative revenue sources cover costs. There is a gap after launch, before scale, where the business burns cash. You need capital and patience to survive that period. We had both, because HMD Corporation's holding structure could absorb early losses while internal products generated baseline volume.
Cultural inertia plays a role too. The payment industry thinks in basis points. Suggesting that you can build a payment processor without charging basis points is like telling a restaurant it can survive without charging for food. The analogy is imperfect, but the reaction is the same: incredulity, followed by a list of reasons it cannot work, followed by a quiet change of subject when you show the numbers.
HMD Pay has processed real transactions, at zero fees, for over three years. The model is not theoretical. It is running.
It is not the right model for every payment company. It requires a specific set of circumstances: a holding structure to absorb early losses, internal products to generate baseline volume, and no legacy revenue to protect. We had all three.
HMD Pay is not a consumer-facing product. It is B2B infrastructure, an API that powers transactions behind the scenes for HMD Corporation's own products and for partner businesses that integrate it. End users never see HMD Pay directly, the same way most people never see the payment processor behind their favourite checkout page
But it is proof of something that I think matters: the assumption that payment processing must cost money is just that. An assumption. Not a law of physics. Not an engineering constraint. A business model choice that everyone made, and that nobody questioned, until someone did.