There is a good chance that the "Lightning" wallet on your phone is not talking to the Lightning Network the way you think it is. Increasingly, it talks to Spark, and the design goal is that you never notice.
Spark is a Bitcoin Layer 2 built by Lightspark, the company led by David Marcus (formerly PayPal and Meta). It does not use Lightning's payment channels. It uses a statechain model: ownership of an on-chain UTXO is passed around off-chain, coordinated by a small set of operators called the Spark Entity. Today that entity consists of two companies, Lightspark and Flashnet. Every transfer inside Spark is co-signed, and therefore seen, by that entity.
Keep that last sentence in mind. Everything below follows from it.
The quiet consolidation
Spark solved a real problem. Building a self-custodial Lightning wallet is hard: channels, liquidity, routing, node management. Spark abstracts all of that away. According to Lightspark, spinning up a Lightning wallet on Spark takes about six lines of code. So wallets came.
Wallet of Satoshi, one of the largest Lightning wallets in the world, now runs its self-custodial mode on Spark. Cake Wallet rolled out Lightning to more than a million users through the Breez SDK, which is built on Spark. The Breez SDK itself defaults to Spark. Piece by piece, a large share of the wallets people call "Lightning" now funnel through the same two-operator entity.
Lightspark markets the invisibility as a feature: users will not even notice Spark powering their wallet. From a pure UX standpoint, that is fair. Seen from the other side, it means millions of people believe they are transacting on an open, permissionless network while their transfers route through a coordinator that sees every one of them. By design, they were never told.
What "sees every transfer" means
When you move money inside Spark, the operators do more than facilitate. They hold one half of the 2-of-2 signature on every transfer, so they necessarily observe who sends to whom, how much, and when. Spark's own documentation states that operators can observe transfer metadata. The comparison pages on their site say the same thing in plainer words: Spark exposes transaction metadata to its operators.
Lightning proper works differently. Individual Lightning payments never touch the blockchain, and onion routing means each hop only learns the node directly before and after it. No single party is supposed to hold the full picture. In Spark, the full picture lives, by construction, inside a two-company federation.
Spark's defenders call this an acceptable trade for usability, and they point out, correctly, that visibility is not custody. That deserves a fair hearing.
The honest part: it is not theft
Spark is self-custodial in the way that matters most. The operators hold only one key in the 2-of-2, so they cannot move your funds on their own. When you receive Bitcoin on Spark, you also receive a pre-signed unilateral exit transaction. Even if the operators disappear or refuse to cooperate, you can broadcast it after a timelock and recover your Bitcoin on Layer 1. That is a real guarantee, and it is more than a custodial wallet gives you. Any honest critique of Spark has to start by conceding it.
The trust model is 1-of-n: as long as a single operator behaves honestly, your funds stay safe. Operators also delete old key material after each handoff, so a later compromise cannot reverse a past transfer. On the custody axis, Spark is a thoughtful design.
The problem is not your funds, it is your footprint
Privacy and custody are different axes, and Spark quietly conflates them. Not being able to steal your Bitcoin is not the same as not being able to see your Bitcoin. On the seeing axis, Spark comes close to a worst case: a named, regulated, two-company entity with a complete metadata ledger of everything on its network. A growing share of "Lightning" users are on that network without knowing it.
It does not stop at Spark's own users. Suppose you refuse Spark entirely and run your own sovereign node. If the person you pay uses Wallet of Satoshi or Cake, your payment lands on Spark's side anyway, and the operator records the receive: recipient, amount, time. You never chose Spark. You never enrolled. You paid someone who did, and a company you will never interact with now holds a line item of your spending.
As Spark absorbs more endpoints, a larger fraction of everyone's counterparties sits behind that one observer. Privacy on Bitcoin was supposed to be something you could choose for yourself. Spark makes a piece of it something your counterparties choose for you, and something an infrastructure company decides on behalf of both of you.
What to take away
This is not a claim that Spark is a scam or that your coins are at risk. The claim is narrower and, I think, more important: a large and growing share of Bitcoin's instant payments layer now flows through infrastructure that is invisible to the people using it and fully visible to the two companies running it.
Maybe smoother UX in exchange for a surveillable middleman on ordinary payments is a trade you want to make. You can only make it if you know it is being offered. Right now, for most users, it is not. The first honest step is naming what is under the hood.
Next: what happens to that metadata once your wallet is linked to a public identity, and why "we don't log" is a promise, not a protection.
