How is a non-custodial gateway different from just sharing my wallet address?

Posting one address gives you no way to tell which customer paid, no per-invoice matching, no confirmation tracking, and a public record tying every sale to one address. A gateway adds unique per-payment addresses, invoice matching, finality tracking, underpayment handling, and webhooks, while keeping settlement exactly as direct as the bare address was.

Can a non-custodial gateway steal or lose my funds?

It cannot spend from addresses derived from your public key material, and well-designed forwarding contracts fix your wallet as the only possible destination. What a bad or failed gateway can do is stop watching, which delays detection, not possession. The residual risks worth checking are the forwarding-contract design and what recovery tooling exists if the platform disappears.

Why do non-custodial gateways usually skip merchant KYC?

Identity-collection duties in most payment regulation attach to businesses that hold or transmit other people's funds. A gateway that only watches chains and derives addresses from merchant-supplied material holds nothing, which in many jurisdictions leaves it outside those duties. The merchant's own legal obligations to customers are unchanged either way.

Do customers experience anything different at checkout?

Usually nothing visible. They see a hosted payment page with an address, a QR code, and a live status, the same as custodial checkouts. The difference is invisible: the address they pay belongs to the merchant rather than the platform, and there is no platform account or identity step in front of the payment on a well-built non-custodial flow.

What is a non-custodial payment gateway?

A non-custodial payment gateway is a service that lets a merchant accept cryptocurrency payments which settle directly to wallets the merchant controls, while the gateway itself only watches blockchains, matches incoming transactions to invoices, and notifies the merchant's store. The defining property is negative: at no point in the payment's life does the gateway hold, pool, or control the funds. That one property is where everything else about the model follows from, so this page walks through what such a gateway actually does, how it keeps addresses under merchant control, what it costs, and what it honestly cannot do.

What the gateway actually does

Strip custody out of a payment processor and three jobs remain. First, address management: turning the merchant's wallet material into a unique, correct payment address for every invoice, so payments match themselves to orders. Second, watching: following each chain's blocks, detecting the payment, tracking confirmations against that chain's finality rules, and handling the awkward cases, underpayments, late payments, reorganized blocks. Third, telling the merchant: a signed webhook or status feed the store can trust, so fulfilment can proceed without anyone reading a block explorer. Merchants pay for those three jobs, not for holding money.

How addresses stay under merchant control

The mechanism differs by chain, and the differences are worth knowing because they are what you would verify if you doubted a gateway's claims. On Bitcoin-family chains the merchant supplies an extended public key, from which the gateway can derive unlimited fresh receiving addresses without ever being able to spend from them: derivation needs only the public half, spending needs the private half the merchant kept. On account-based chains such as Ethereum, TRON, or Solana, each payment gets a small forwarding contract whose destination is fixed to the merchant's own wallet when it is created, so even the gateway's operator keys can only move funds to that one place. On chains with native payment references, XRP, Stellar, and TON, customers pay the merchant's own account and a per-invoice tag, memo, or comment does the matching. And Monero acceptance works through a private view key the merchant shares, which can observe incoming payments to the merchant's own wallet but cannot spend a coin of them.

How the gateway gets paid

A custodial platform can deduct its fee from the money passing through it. A non-custodial one cannot, because nothing passes through it. The common solution is a prepaid fee balance: the merchant tops up a small credit, and the gateway deducts its percentage from that credit each time a payment confirms, leaving the customer's transfer untouched. Some gateways bill monthly instead. Either way, a merchant evaluating pricing should look past the headline percentage to the network-cost question: when the gateway broadcasts on the merchant's behalf, sweeping a forwarding contract or relaying a gasless payment, does it pass the chain's real cost through transparently, or estimate it in its own favor?

What the model cannot do

Honest limits, because the model has them. No fiat settlement: a gateway that never holds crypto cannot convert it to euros and wire it to a bank, so merchants who want money in a bank account need a custodial processor or their own exchange relationship. No push-button refunds: the gateway holds nothing to refund from, so a refund is a transfer the merchant sends from their own wallet. No key recovery: if the merchant loses their own wallet, the gateway cannot restore it, which is why serious non-custodial gateways ship recovery tooling that re-derives payment addresses from the merchant's recorded material, and why merchants should treat wallet backups as part of the setup. And volatility is the merchant's: funds arrive as the crypto the customer paid, not as a stabilized balance, unless the merchant accepts stablecoins in the first place.

Common questions

How is a non-custodial gateway different from just sharing my wallet address?: Posting one address gives you no way to tell which customer paid, no per-invoice matching, no confirmation tracking, and a public record tying every sale to one address. A gateway adds unique per-payment addresses, invoice matching, finality tracking, underpayment handling, and webhooks, while keeping settlement exactly as direct as the bare address was.
Can a non-custodial gateway steal or lose my funds?: It cannot spend from addresses derived from your public key material, and well-designed forwarding contracts fix your wallet as the only possible destination. What a bad or failed gateway can do is stop watching, which delays detection, not possession. The residual risks worth checking are the forwarding-contract design and what recovery tooling exists if the platform disappears.
Why do non-custodial gateways usually skip merchant KYC?: Identity-collection duties in most payment regulation attach to businesses that hold or transmit other people's funds. A gateway that only watches chains and derives addresses from merchant-supplied material holds nothing, which in many jurisdictions leaves it outside those duties. The merchant's own legal obligations to customers are unchanged either way.
Do customers experience anything different at checkout?: Usually nothing visible. They see a hosted payment page with an address, a QR code, and a live status, the same as custodial checkouts. The difference is invisible: the address they pay belongs to the merchant rather than the platform, and there is no platform account or identity step in front of the payment on a well-built non-custodial flow.

Where this site stands: this article describes the model NoHoldPay implements, and the product page documents the specifics per chain. For the two-model comparison, see custodial vs non-custodial.