Paying with Bitcoin: On-chain, Lightning, Liquid, Spark and Cashu Compared

Why there is more than one way

Because payments are not all alike. Someone holding a meaningful amount for years, or moving it between their own wallets, needs something different from someone buying a coffee. A merchant needs different properties again from someone sending small amounts privately.

The underlying reason is a property of Bitcoin itself: block space is limited. A new block arrives roughly every ten minutes, and the fee depends on how many others want into the same block. How a Bitcoin transaction works walks through that in detail. For billions of small everyday payments it is therefore neither necessary nor sensible to settle each one directly on the blockchain.

On-chain: the settlement layer

On-chain means the transaction goes straight to the Bitcoin network and is confirmed in a block. Real bitcoin UTXOs move from one script to the next. This is the layer where everything ultimately settles.

If you hold your own keys, you build the transaction yourself and broadcast it. No payment provider, no federation, no mint and no open channel that has to keep working.

• Where the bitcoin sit. In a UTXO that only your key can move.
• Who has to keep cooperating. Nobody. Only the consensus rules apply.
• What it costs. Waiting for confirmation, and a fee that tracks demand for block space.

That makes on-chain the right choice for long-term holding, larger amounts, moving funds between your own wallets, and for opening and closing every other layer. Everything built on top anchors its ends here.

Lightning: payment channels

Lightning moves the individual payment off the blockchain and into a channel between two parties. Inside that channel the balance can shift as often as you like without a new block. Through a network of connected channels a payment also reaches people you have no direct channel with. The Lightning Network explained covers the mechanics.

For everyday use this is a mature answer: seconds instead of minutes, very small amounts become practical, and there is a large ecosystem of wallets, nodes and payment providers.

• Where the bitcoin sit. In an on-chain UTXO backing the channel. With a self-custodial wallet, the claim on it stays yours.
• Who has to keep cooperating. With self-custody, nobody permanently, but you do need to be able to respond if a counterparty publishes an outdated state. With a custodial wallet you are trusting the provider outright.
• What it costs. Liquidity has to be there, and channels have to be managed.

Inbound and outbound liquidity, opening and closing channels, routing: that is the technical reality underneath. Modern wallets hide a great deal of it. The complexity has not disappeared, it has moved somewhere else.

Liquid: a federated sidechain

Liquid is a separate blockchain alongside Bitcoin. Bitcoin are moved in through a peg and used there as L-BTC. According to the Liquid Network technical overview, a block is produced roughly every minute, and a peg-in requires 102 confirmations on the Bitcoin network before the funds can be claimed on Liquid.

One distinctive feature is Confidential Transactions: the amount and the asset type are hidden by default, and according to the documentation not even federation members can see them.

• Where the bitcoin sit. The actual bitcoin are held by the federation. You hold L-BTC, a representation of them on Liquid.
• Who has to keep cooperating. The federation. Its members are geographically distributed, largely exchanges and financial companies, and blocks are signed by at least two thirds of the block signers.
• What it costs. A different security model from Bitcoin itself.

Spark: transfers without your own channels

Spark is the newest of the approaches here and works on a statechain model. Put simply: a payment does not move a UTXO on the blockchain each time. Instead, control over it is handed on off-chain.

The practically interesting part is how it meets Lightning. A Spark user can send and receive Lightning payments without running channels at all. That targets precisely the hurdle where self-custodial Lightning often stalls in practice.

• Where the bitcoin sit. In a UTXO on Bitcoin L1. The Spark documentation describes funds as always non-custodial and recoverable on L1 through a unilateral exit that depends on Bitcoin and nobody else.
• Who has to keep cooperating. The Spark Operators, but within limits: the documentation states they cannot move funds without the user, who is a required participant in any transfer. The security claim rests on at least one operator behaving honestly and deleting its key share afterwards, or on a configured threshold of them, depending on the setup.
• What it costs. One added assumption compared with plain on-chain custody, and a young ecosystem.

How the model holds up over years, and how widely the operator set ends up distributed, cannot be settled today. That is not an argument against Spark. It is an argument for matching your amounts and expectations to it.

Cashu: ecash issued by a mint

Cashu is different in kind from everything above. It is a Chaumian ecash protocol: a mint issues digital tokens that users pass between each other. Because of blind signatures the mint does not necessarily know which deposit connects to which later payment. That gives it privacy properties none of the other layers offer in the same way.

In practice it usually runs through Lightning: you pay the mint and receive tokens, which are then very easy to hand on, for example as text or a QR code. A received token is only safe once the recipient has swapped it at the mint.

• Where the bitcoin sit. With the mint. You hold a claim, not bitcoin.
• Who has to keep cooperating. The mint operator, completely. The Cashu documentation says so plainly: mints act as a custodian for your satoshis, and users must trust the mint to redeem their ecash.
• What it costs. The full failure risk of that mint.

The five layers side by side

The same four questions, answered five times. The order follows distance from settlement: strongest control at the top, more convenience further down, and with it a growing number of assumptions.

Which layer fits when

The useful question is not which layer is fastest, but which loss you would be willing to absorb in the worst case. Almost everything else follows from that.

When the wallet picks the layer for you

For users, something else may end up mattering more: perhaps they will not need to know which route a payment takes. A wallet could check in the background whether Lightning has the liquidity, whether Spark fits, or whether an on-chain transaction makes more sense. All that would be visible is one button: pay.

We know this pattern from elsewhere. Nobody sending a message thinks about which protocol carries it. There is little reason bitcoin payments should not head the same way.

One difference remains, though, and it is an important one. The interface may get simpler. The differences in security must not become invisible along the way. A UTXO under your own keys is not the same as L-BTC inside a federation, and an ecash token is a claim against a mint. Precisely when wallets start mixing these layers behind the scenes, showing people where their bitcoin currently sit becomes more important, not less.