BTCPay Server review

Why zero here means zero#

BTCPay charges nothing because there is nobody to charge. The FAQ is explicit that the project is not a company and that payments settle straight into the merchant's own wallet, so the only money leaving a transaction is the network fee the merchant would have paid anyway.

Every other card in this index carries a percentage. This one carries a labour cost instead: somebody has to run the node, keep it patched, watch the Lightning channels and answer the phone when a customer's payment sits unconfirmed. That work is real and it is not on a pricing page, because there is no pricing page.

The custody question is settled rather than negotiated#

Non-custodial is the default and the only mode. The software builds invoices against an extended public key the merchant supplies, and the private key never reaches the server, so nothing in this stack can freeze a balance for the simple reason that nothing in it holds one.

That closes the largest structural risk in this category and opens a smaller, sharper one. A lost seed phrase is unrecoverable and no support desk has a reset procedure. Read the custody guide before concluding this is automatically the safer option.

An integration list nobody else matches#

Twenty-three named platforms appear on the project's own page, running from WooCommerce and Shopify down to Ghost, Pretix and Invoice Ninja. Commercial processors maintain five or six because each plugin costs a salary; a community maintains twenty-three because each one is somebody's own shop.

The variance that comes with that is the trade-off worth checking. A widely used plugin is maintained continuously, while a niche one may lag the platform it targets by a release or two. Look at the commit history of the specific plugin you need, not at the length of the list.

What this card cannot tell you#

The asset count is blank because the project publishes no count. Bitcoin on-chain and Lightning are named in the FAQ; altcoin support arrives through separate plugins whose status moves independently of the core release, so any number printed here would be stale within a quarter.

Mass payouts, white label and recurring billing are similarly unstated on the pages we read. Some of them exist in plugin form. This card records what the project publishes about itself, and what it publishes is a feature tour rather than a specification.

Who should run this#

A merchant with an engineer on staff, revenue they are content to hold in bitcoin, and a reason to care that no third party can observe or halt the flow. Where those three conditions hold this is the strongest card in the index, and the saved percentage is not the main reason.

An organisation taking donations is the other clean fit. There is no acquirer in the path to deplatform a cause, which is a structural property rather than a policy promise, and policy promises are most of what the nonprofit page has to weigh elsewhere.

Who should not#

A business that needs euros in its bank account on Tuesday. Nothing here converts or settles fiat: the merchant receives bitcoin and arranges everything after that, which means a second relationship and a second set of fees this card does not show.

A team with no operational headroom. Self-hosting a payment endpoint is a commitment measured in years, and a processor charging one percent that answers tickets is cheaper than an unmaintained node that drops invoices during a launch week.

How this card scores where it does#

Cost takes the maximum and onboarding does too, because there is no verification step to pass and no application to be refused. Coverage is the weakest line by a distance: two networks against fifty elsewhere in the index, and no settlement currency at all.

Support scores low for a reason that is not quite a criticism. The channels are a chat room and a public tracker, both genuinely good, and neither carries a published response commitment — and the method counts only what a reader can check.

Where BTCPay Server sits in the index#

It is the only card here that is not a business, and reading it as one produces the wrong answer. The comparison that matters is against running no processor at all, not against the percentage on the card below it.

Every figure above came off the project's own pages on 2026-09-02. Setting the cost weight to zero on the methodology page is the quickest way to see what the rest of this card is actually worth to you.