I have run Lightning nodes since June 2018. I have opened channels, closed channels, debugged failed routes at three in the morning, paid for things with Lightning across three countries, watched node operators bankrupt themselves on liquidity, and built Bitcoin infrastructure as a professional for the better part of a decade. I have wanted Lightning to win the entire time.

This article is me, finally, facing what actually happened.

The tech worked. The UX didn’t. The market moved on.

All three of those statements are true, and the version of this story you usually read picks one of them and dunks on the other two. The honest version says they fit together. Lightning is a beautiful piece of engineering that solved a problem most users do not have, in a way most users could not figure out, and the actual payments market in 2026 is dominated by something else entirely.

This piece walks through what Lightning is, what it solved, why it could not become what its early advocates promised, what it did actually become, and what the whole experience tells us about a deeper question. The deeper question is the one I keep coming back to: how much do users actually value sovereignty and security versus a good user experience? Lightning at ten is the cleanest natural experiment we have on that trade-off, and the answer is not the one Bitcoin’s culture wants.

A short history

The Lightning Network was sketched out in a 2015 whitepaper by Joseph Poon and Thaddeus Dryja. The idea was elegant. Bitcoin’s base layer can settle only a small number of transactions per second globally, which makes it unsuitable for the kind of high-volume, low-value payments that real-world commerce needs. Lightning lifts payment activity off the base chain into a network of bilateral channels that settle to bitcoin only when they open and close. Inside a channel, two parties can exchange unlimited payments at zero marginal cost and near-instant speed. Connect enough channels together and you have a network where any user can pay any other user through a chain of intermediaries, with cryptographic guarantees that nobody along the path can steal the money.

Mainnet implementations shipped in 2018. Three independent codebases, LND, Core Lightning (then called c-lightning), and Eclair, gave the network a healthy genetic diversity from day one. Early adopters were node operators, exchanges, and a self-selecting Bitcoin developer community. Twitter integrated tipping via Lightning in 2019. The first Lightning-native businesses appeared. Capacity climbed from essentially zero to roughly 1,000 bitcoin held in public channels by the end of 2020.

The first inflection point came in September 2021, when El Salvador adopted Bitcoin as legal tender and the Bukele government rolled out the Chivo wallet to push citizens onto Lightning rails. Public capacity tripled in the following twelve months. The narrative said this was the beginning of a global rollout. Venture capital arrived in force. Lightning Labs and Strike raised at large valuations. A generation of Bitcoin-native payment startups was born.

The second inflection point came quietly. Wallet of Satoshi, the most-used Lightning wallet in the world, withdrew from the United States in November 2023, citing regulatory pressure on its custodial model. Public capacity peaked the same year, drifted lower through 2024 and 2025, and recovered only in 2026 as institutional players returned. El Salvador rescinded Bitcoin’s legal-tender status in February 2025 as a condition of a $1.4 billion IMF loan. The Chivo wallet was put up for sale. By the time you read this, the most-funded and most-state-backed Lightning adoption push in history is being unwound under treasury department oversight.

That is the shape of the curve. The technology is mature. The user base is not.

What Lightning actually is, briefly

The mechanism is worth saying out loud because it is genuinely beautiful.

Two parties open a Lightning channel by funding a multisig output on the Bitcoin base chain. Inside the channel, they exchange a series of signed transactions that update their respective balances, with each new transaction implicitly revoking the previous one through a cryptographic dance involving revocation keys, so that if either party tries to broadcast an old state to the base chain, the other party can punish them by sweeping the entire channel. The result is a bilateral, off-chain payment relationship with strong cryptographic guarantees and no third-party trust required.

The network part comes from chaining channels together. If I have a channel with Alice and Alice has a channel with Bob, I can route a payment to Bob through Alice. Alice’s honesty along the path is enforced cryptographically by hash time-locked contracts, and onion routing borrowed from Tor obscures the full path from each individual hop. Routing nodes earn small fees for forwarding payments. In principle, anyone can be a routing node. In practice, a small number of well-capitalised nodes route most of the traffic.

The performance characteristics are extraordinary. Lightning payments settle in hundreds of milliseconds, fees are typically a fraction of a cent, and the network can in principle scale to millions of transactions per second. The cryptography is rigorous, the implementations are mature, and the protocol has been running in production for almost eight years without a major security incident.

If the only thing that mattered was the tech, Lightning would already have won.

The custody trap

It is not.

Lightning’s design assumes that each user operates their own node, manages their own channels, and takes custody of their own keys. This is the Bitcoin maximalist’s dream architecture: every user is a sovereign node, nobody can freeze your money, nobody knows your balances, the network is unstoppable. In practice, running a Lightning node is hard. Managing channels is harder. Being online when payments arrive is non-negotiable. Backing up channel state correctly is a software problem most consumer apps still solve imperfectly.

The market reacted to this difficulty in the obvious way. The Lightning wallets that achieved real adoption were custodial. Wallet of Satoshi, Cash App, Strike, Blink, and a half-dozen others took the same approach: the user opens the app, sees a balance, sends and receives Lightning payments instantly, and never touches a channel or a key. Under the hood, the service operates the actual node and the user holds a database row claiming a balance. Functionally identical, ideologically heretical. Bitcoin’s culture coined the slur “shitcoin” for custodial Lightning wallets, on the grounds that they reduce Bitcoin to a backend settlement layer rather than a user-controlled monetary system. The criticism is correct. The custodial wallets are the only ones that actually work for normal users.

This is the trap. Lightning’s distinctive promise, trustless P2P payments, requires non-custodial wallets. Non-custodial wallets are too hard for normal users to operate reliably. Therefore the version of Lightning that achieved any adoption is the custodial version, which by construction discards the property that made Lightning ideologically interesting in the first place. The wallet that most successfully popularised Lightning, Wallet of Satoshi, was so far from the original vision that it was eventually pushed out of the US by Travel Rule compliance pressure, which it could not satisfy precisely because it was a regulated centralised service.

You can call this success or failure depending on what you think Lightning was supposed to be. Either way, the version the market actually used and the version the original designers proposed have very little in common.

Why the non-custodial UX is so hard

When I tell normal users to run a non-custodial Lightning wallet, here is the list of things they hit in the first week.

Inbound liquidity. To receive a Lightning payment you need a channel with remaining capacity from your side. A fresh wallet has none. To receive your first payment somebody else has to open a channel to you, which costs them an on-chain fee. Phoenix and Breez and Mutiny solve this by auto-opening a channel on first use and deducting a fee from the first incoming payment. It is the right answer, and it surprises every new user, because no previous money app has ever charged them a fee to receive money.

Channel closures. Closing a channel settles funds on the base chain and costs an on-chain fee anywhere from a dollar to thirty dollars depending on mempool state. Uncooperative closures cost more. For someone whose Lightning balance is twenty dollars, a thirty-dollar exit fee is the entire balance. Nobody warns them.

Watchtowers. The cryptographic punishment mechanism that keeps your counterparty honest only works if you are online to enforce it. The fix is a watchtower service that monitors the chain on your behalf. Most consumer wallets hide this configuration, which means most consumer wallets are running a security model that the protocol does not actually deliver.

Backups. Channel state changes every time you transact. A seed phrase only restores the on-chain keys, not the channels. Channel state must be backed up separately and continuously, and if your phone dies between backups you can lose money. Phoenix solved this with encrypted server-side backups; again, useful, and again, a step away from pure self-custody.

Routing failures. Payments fail because there is no route with enough liquidity, because a routing node is offline, because a competing payment locked the same liquidity in another channel. The failure messages are technical and the retry behavior is opaque. Most users try once, fail, and conclude the network is broken.

Synchronisation. Mobile wallets sync on cold start. It takes seconds. For a power user this is invisible. For somebody trying to pay for coffee, “wait fifteen seconds while my wallet syncs” is the moment they reach for the dollar app.

Each is small in isolation. Together they are an enormous gap between Lightning and any of the consumer payment apps it was supposed to replace.

The merchant side

The other half of any payment network is the merchant. Lightning’s merchant story is even less rosy than the user story.

Most businesses that say “we accept Bitcoin” still mean on-chain Bitcoin, not Lightning. The Lightning-accepting merchants are concentrated in a few categories: cryptocurrency exchanges, Bitcoin-adjacent SaaS and hosting, a handful of cafés and bars in Bitcoin-friendly cities, and a long tail of online merchants who use a processor like Strike or OpenNode or BTCPay Server to convert Lightning payments to fiat at the till. The BTCPay Server project has done excellent work building open-source merchant infrastructure, but its installed base is small compared to Stripe or Adyen.

The Bitcoin-native physical payment story is more poignant. BoltCards are NFC plastic cards that hold a Lightning identifier and let a merchant tap-to-receive a payment. They are technically beautiful, work reliably, and are used almost nowhere. Lightning-accepting POS terminals exist in El Salvador, in a small Lightning-positive corner of Lugano, and in the occasional pop-up event. Compared to the global rollout of contactless cards and mobile-wallet QR payments, Lightning’s physical presence is invisible.

The structural reason is straightforward. A merchant who already accepts cards has very little to gain from also accepting Lightning. Card processing fees are 1.5 to 3 percent in most markets, which is similar to Lightning’s all-in cost after liquidity and exit fees are accounted for. Card payments are already instant from the merchant’s perspective. The customer-side reach of cards is approximately everyone. Adding Lightning is more work for an audience that is functionally zero. The merchants who do accept it do so for reasons that are not purely commercial.

The adoption numbers

What does Lightning’s actual adoption look like in 2026?

The most-cited number is the public network capacity, which sat at roughly 5,600 BTC in May 2026, about $400 to $500 million in dollar terms depending on the bitcoin price on the day. That is up from a 2025 low of around 4,100 BTC but only modestly above the 2022 peak. Capacity has been more or less flat in dollar terms for four years. Public channel count is around 50,000 and active node count is around 13,000 to 15,000.

The transaction volume numbers are messier. River Financial, one of the better-instrumented Lightning operators, reports its own Lightning volume in the range of tens of millions of dollars per month, with the network-wide volume estimated by industry analysts at roughly $1.1 billion in monthly volume across 5 million transactions, in late 2025 and early 2026. The headline “$1 billion monthly volume” was passed around by Lightning advocates as a milestone, and it is real, but it deserves three asterisks.

First, the volume figures conflate genuine Lightning peer-to-peer payments with custodial-to-custodial transfers inside large Lightning-using services. Most of Strike’s volume, for instance, is users buying bitcoin with dollars and remitting it through Strike’s Lightning rails. Functionally this is a remittance product on Lightning infrastructure, not Lightning being used as money.

Second, the volume is dominated by a small number of institutional operators. The same River report estimates that the top ten Lightning-using businesses account for the great majority of activity. Lightning is increasingly the back-end of fintech apps and exchanges, not a peer-to-peer payment network in the original sense.

Third, the comparison numbers are humbling. Stablecoins settled roughly $1.7 trillion in transaction volume in 2024 according to Visa’s on-chain analytics, growing through 2025 and 2026. Pix in Brazil settled approximately $4.4 trillion in 2024, or roughly $370 billion per month. Visa global card volume runs in the tens of trillions per year. Lightning’s $1.1 billion per month is, by these standards, a rounding error. It is not nothing. It is not the global payment rail Lightning’s advocates have been promising for ten years.

The El Salvador autopsy

El Salvador was the closest thing to a controlled experiment we will get on whether state-backed adoption could close the gap.

In September 2021 the Bukele government declared bitcoin legal tender and rolled out the Chivo wallet to citizens with a $30 sign-up bonus. The infrastructure was real. The promotion was relentless. The international press coverage was enormous. If anywhere was going to demonstrate that ordinary people would adopt Lightning given the right push, it was here.

The follow-up data is unambiguous. The percentage of Salvadorans who reported using bitcoin for transactions fell from 25.7 percent in 2021 to 21 percent in 2022, to 12 percent in 2023, to 8.1 percent in 2024. Twenty percent of people who downloaded the Chivo app never used their $30 bonus. Sixty-one percent of those who used the bonus stopped using the app afterwards. The Yale School of Management called it a near-quantitative failure of the adoption hypothesis, and Yale was being polite.

In December 2024, as a condition of a $1.4 billion IMF Extended Fund Facility, El Salvador agreed to remove the mandatory acceptance of bitcoin by merchants, stop accepting bitcoin for tax payments, and wind down the Chivo wallet. Legal tender status was rescinded in February 2025. The Chivo wallet is being privatised. The country’s central holdings, which Bukele still publicly celebrates, have continued largely through accounting reshuffles rather than market purchases. The IMF disclosed in mid-2025 that El Salvador had made no new market purchases since February 2025, contradicting Bukele’s “one bitcoin per day” Twitter narrative.

You can spin this however you want. The honest reading is that the most-funded, most-state-backed, most-aggressively-marketed Lightning adoption push in history produced an active-use rate in the single digits, then collapsed under the first serious external pressure. If state-backed adoption could not move the needle here, it is hard to construct a case for it moving the needle anywhere.

The structural problem

The deeper question this whole story is asking is one Bitcoin’s culture has been refusing to answer for a decade. How much do users actually value sovereignty and security versus a good user experience?

The Lightning experiment is a clean test. Lightning’s distinctive product is sovereign, trustless, P2P payments. The custodial Lightning wallets sacrifice sovereignty and trustlessness in exchange for usability. The market overwhelmingly chose the latter. When forced to pick, users picked easy over sovereign by a ratio that was not even close.

This is not a Lightning-specific result. It is the same answer in every market where the trade-off has been offered. Most Bitcoin is held on exchanges, not in self-custody. Most DeFi volume runs through centralised front-ends and aggregators. Privacy coins have not won. Hardware wallets are used by a tiny fraction of crypto holders. The pattern is consistent and it has been consistent for a long time.

There is a more comfortable version of this story Bitcoin’s culture likes to tell itself. The version is: users do not understand the value of sovereignty yet; we need better education; once they understand what they are giving up, they will choose differently. I have spent a decade in this industry and I no longer believe this. The user research is in. The natural experiments have run. People who deeply care about sovereignty are a small minority. People who deeply care about easy, instant, cheap payments are everyone. Stablecoins on Tron and Solana figured out how to deliver the second want with 10 percent of the technical complexity of Lightning, and they ate the market that Lightning was built for.

This is the strategic position Lightning is now stuck in. It is too complex for non-believers, who have stablecoins for the use case that matters to them. It is too custodial in practice for believers, who would rather hold bitcoin and use on-chain for the rare cases when they need to move it. The middle ground, where Lightning was supposed to live, did not turn out to contain very many people.

What Lightning actually became

Lightning did not become a global P2P payment network. It became something else, and it is worth being honest about what.

Lightning is the dominant settlement rail for cryptocurrency exchanges moving bitcoin between each other. Kraken, Bitfinex and a long list of smaller venues run Lightning nodes and settle inter-exchange flows over the network in seconds. It accounts for a meaningful fraction of on-network volume.

Lightning is the back-end for Strike, the most successful Bitcoin-USD remittance product, which lets users send dollars from one country to another by buying bitcoin, routing it through Lightning, and selling it on the other side. The user does not see the bitcoin or the Lightning channel; they see a remittance that arrives in seconds at a lower fee than Western Union. Lightning is what makes the product possible.

Lightning is the payment rail for the Nostr decentralised social protocol, where “zaps” let users tip each other in satoshis. The dollar volume is modest, but Nostr is one of the few places where Lightning is being used in the spirit of the original design: small, fast, peer-to-peer, ideologically committed.

Lightning is the playground for the Bitcoin power-user community. The people who run their own nodes and use Lightning for everyday personal payments are a small, committed group. They are not, by any honest count, a market.

Add it all up and Lightning is a successful niche payments infrastructure with a few thousand serious users and a handful of institutional operators. That is not what the marketing promised. It is also not nothing.

What I have to admit

I have wanted Lightning to win for ten years. I built Bitcoin Layer-2 infrastructure for a living. I ran nodes through every major upgrade. I have used Lightning to pay for things in three countries, and I have used it through every consumer wallet that existed.

What I have to admit is that the version of Bitcoin payments I wanted was not the version most people wanted, and that the trade-off the Lightning protocol made, sovereignty over usability, is not the trade-off that the actual market in payments cared about. I can keep believing that sovereignty is important. I do believe that. But I cannot keep believing that the market shares my preferences. The data is too clear.

Two possible futures keep me open to the idea that Lightning is not done. The first is regulatory. If stablecoins get squeezed hard in some jurisdictions over the next few years, the demand for a sovereign payment rail might rise sharply, and Lightning is one of the few mature options waiting in the wings. The second is technical. If somebody, somewhere, finally produces a non-custodial Lightning UX that is as easy as Cash App, the dynamic changes overnight. The smartest people in Bitcoin have been trying to produce that UX for seven years. The fact that they have not is information, but it is not proof that they will not.

In the meantime, I run my nodes, I keep my channels open, and I treat Lightning as what it has actually become: a useful piece of infrastructure for a small set of use cases, not the future of money. The future of money, for now, is being settled on rails I do not love but cannot honestly deny are winning. That is what a market is for.

References

1. Poon, J., & Dryja, T. (2016). The Bitcoin Lightning Network: Scalable Off-Chain Instant Payments. https://lightning.network/lightning-network-paper.pdf
2. Bitcoin Visuals. Lightning Network Capacity. https://bitcoinvisuals.com/ln-capacity
3. 1ML. Lightning Network Statistics, Bitcoin mainnet. https://1ml.com/statistics
4. CryptoSlate (2023). Lightning Network app Wallet of Satoshi ends support for U.S. customers. https://cryptoslate.com/lightning-network-app-wallet-of-satoshi-ends-support-for-u-s-customers/
5. The Block (2024). Major Bitcoin Lightning wallet provider quits US market. https://www.theblock.co/post/264585/wallet-of-satoshi-bitcoin-lightning
6. Yale Insights. El Salvador Adopted Bitcoin as an Official Currency; Salvadorans Mostly Shrugged. https://insights.som.yale.edu/insights/el-salvador-adopted-bitcoin-as-an-official-currency-salvadorans-mostly-shrugged
7. CoinDesk (2025). IMF Says ‘Efforts Will Continue’ to Ensure El Salvador Doesn’t Accumulate More BTC. https://www.coindesk.com/policy/2025/05/27/imf-says-efforts-will-continue-to-ensure-el-salvador-doesn-t-accumulate-more-btc
8. The Block (2025). El Salvador hasn’t bought Bitcoin since February, finance chiefs tell IMF. https://www.theblock.co/post/363483/el-salvador-hasnt-bought-bitcoin-since-february-finance-chiefs-tell-imf-contradicting-bukele-administration
9. Visa On-Chain Analytics. Stablecoin transaction volume dashboards. https://usa.visa.com/solutions/crypto/onchain-analytics-dashboard.html
10. Banco Central do Brasil. Pix Statistics. https://www.bcb.gov.br/en/financialstability/pix
11. River Financial. Lightning Network reports, 2024 and 2025 editions. https://river.com/learn/
12. BTCPay Server. Self-hosted, open-source Bitcoin and Lightning payment processor. https://btcpayserver.org/