Lightning Lost 3,000 Nodes Since 2022: Is the Network Shrinking or Maturing?

The Lightning Network peaked at roughly 20,700 reachable nodes in early 2022. By mid-2026, that number had fallen to approximately 17,400, according to mempool.space data. That is a 16% decline over four years. Headlines calling Lightning a “shrinking network” write themselves, but the raw node count is one of the least informative metrics for evaluating a payment network. What actually matters is whether the network can reliably route payments, and by that measure, the picture is more complicated than a single number suggests.

What the Node Count Data Shows

Public Lightning node counts are tracked by crawlers that connect to the gossip protocol and record which nodes announce themselves. The peak of roughly 20,700 nodes in early 2022 coincided with the tail end of the 2021 bull market, when running a Lightning node became a popular project among Bitcoin enthusiasts. As the market cooled, many of those nodes went offline.

The decline has not been uniform. Between mid-2022 and late 2024, the count dropped more sharply, with some trackers recording figures as low as 14,900. By 2026 the count partially recovered and stabilized in the 17,000 to 17,400 range, depending on the tracker and snapshot date.

Notice the divergence: node counts fell while capacity climbed to an all-time high of roughly 5,600 BTC in December 2025. Channels dropped from 87,000 to about 41,000, yet the average capacity per channel increased substantially. Fewer pipes, but each one carries more.

Capacity per Node: The Metric That Matters More

In early 2022, the average public capacity per node was roughly 0.19 BTC. By mid-2026, it had risen to approximately 0.28 BTC. That shift tells a story that raw node count misses: nodes that remain on the network are better capitalized and more likely to be professionally managed.

This is visible in the Gini coefficient for Lightning capacity distribution, tracked by Glassnode. The coefficient has risen from approximately 0.85 in Lightning's early years to 0.977 as of early 2026. For context, a Gini of 1.0 would mean a single node holds all capacity; 0.0 would mean perfectly equal distribution. A peer-reviewed study published in IEEE Access (2025) confirmed this trend across eight yearly snapshots.

Concentration is not the same as centralization: A high Gini coefficient shows that capacity is unevenly distributed, not that a single entity controls the network. Many large nodes are run by independent operators, exchanges, and Lightning Service Providers in different jurisdictions. The channel-count Gini is notably lower at 0.74, meaning connections are more broadly distributed than capital.

Who Left: The Hobbyist Node Exodus

Most of the nodes that disappeared between 2022 and 2026 were small, low-capacity nodes run by individuals. These were typically Raspberry Pi setups with one or two channels and minimal liquidity, often operated as educational projects or acts of Bitcoin activism rather than as serious routing infrastructure.

Several factors drove the exodus:

• The 2022 bear market reduced both the financial incentive and community enthusiasm for running nodes
• Channel management requires ongoing capital allocation, rebalancing, and monitoring that most hobbyists eventually tire of
• Rising on-chain fees in congestion periods made opening and closing channels expensive for small operators
• Routing revenue for small nodes remained negligible, rarely covering even the electricity costs of a dedicated machine

The irony is that many of these nodes were net-negative for the network. A node with 100,000 sats of capacity and unreliable uptime creates routing options that look valid on the graph but fail when a payment actually tries to traverse them. Removing these nodes from the graph can improve pathfinding by eliminating dead-end routes.

Who Stayed: The Rise of Professional Operators

While hobbyist nodes left, a class of professional routing operators and Lightning Service Providers (LSPs) expanded. These entities run high-uptime nodes with large channel capacities, active liquidity management, and sophisticated fee policies.

Key LSPs shaping the network

ACINQ operates the LSP behind the Phoenix wallet, which uses a single channel with splicing to resize dynamically. Lightspark provides enterprise Lightning infrastructure for Coinbase, Nubank, and other institutions. River operates high-capacity routing nodes while publishing some of the most widely cited Lightning payment volume data. Breez, Voltage, and others provide node hosting and SDK services that abstract away channel management for wallet developers.

The shift toward LSPs means that a smaller number of well-connected, well-capitalized nodes now handle a disproportionate share of routing. This is not unique to Lightning: it is a common pattern in overlay networks.

The Tor Relay Analogy

The Tor anonymity network experienced a similar trajectory. Tor relay counts fluctuated over the years, with periods of decline that prompted concern. In 2017, the number of relays dropped from roughly 7,000 to 6,500 over a few months, leading to mailing list discussions about whether the network was weakening. More recently, total relay counts hovered around 6,500 to 8,000, while the network's total bandwidth capacity grew substantially as operators consolidated onto faster, better-provisioned machines.

The parallel is instructive. A network's health depends on aggregate capacity, geographic distribution, and operator independence rather than raw node count. Ten high-bandwidth Tor relays in diverse locations serve users better than a hundred low-bandwidth relays on consumer connections in the same data center. The same logic applies to Lightning: routing nodes with reliable uptime and sufficient liquidity contribute more to payment success than dozens of underfunded hobbyist nodes.

Routing Reliability: Fewer Nodes, Better Payments?

If consolidation improves routing, the data should show it. The evidence is mixed but leans positive.

River Financial reported a 99.7% payment success rate across 308,000 transactions in 2023. For small payments under 100,000 sats routed through well-connected nodes, success rates consistently exceed 99%. However, a widely cited 2024 pathfinding comparison study found meaningful differences in success rates across LND, CLN, Eclair, and LDK, and a 2025 simulation showed that 100,000-sat payments on random paths averaged only one success before encountering a routing failure.

The pattern is clear: large payments remain challenging, and routing quality depends heavily on the sender's position in the graph. What consolidation does is widen the gap between well-connected paths and poorly-connected ones. If your wallet connects to a major LSP, your experience is excellent. If you route through small hobbyist nodes, failure rates climb.

Multi-path payments help but do not solve the problem: MPP splits large payments across multiple routes, improving success rates for amounts that would fail on a single path. But MPP increases routing fees and introduces complexity. It works best when the sender has multiple well-capitalized channels, which brings us back to the same concentration dynamic.

The Private Channel Blind Spot

Every public metric about Lightning nodes and channels comes with an asterisk: it only captures announced (public) channels. Private channels are invisible to gossip crawlers and do not appear in mempool.space, 1ML, or Glassnode dashboards.

How much capacity is hidden? No one knows precisely, because unannounced channels are invisible by definition. A 2020 BitMEX Research analysis examined non-cooperative channel closures and estimated that private channels comprised roughly 28% of the sample. As mobile wallets like Phoenix, Breez, and Mutiny (now defunct) grew in popularity, the share of unannounced channels likely increased, since mobile implementations default to private channels to reduce gossip bandwidth and protect user privacy.

Some analysts estimate that total Lightning capacity, including private channels, could be roughly double the public figure. If accurate, the network's true capacity in mid-2026 would be closer to 10,000 BTC rather than the publicly visible 4,900 BTC. This is speculative, but it means that public node-count data understates the network's actual footprint.

Why private channels are growing

• Mobile wallets (Phoenix, Breez) default to private channels for bandwidth and privacy reasons
• LSPs open private channels on behalf of end users via just-in-time (JIT) channel opens
• Exchanges and custodial services use private channels to manage liquidity without exposing their graph position
• Zero-conf channels used by LSPs are typically unannounced

The Decentralization Question

Critics of Lightning consolidation argue that fewer, larger nodes undermine the network's decentralization properties. If a handful of LSPs handle most routing, they become chokepoints that could censor transactions or extract rents.

This concern is legitimate, but context matters. Lightning's architecture offers structural protections that traditional payment networks lack:

• Anyone can open a channel to any node without permission
• Onion routing means intermediary nodes cannot see the full payment path
• Users can unilaterally close channels and exit to Bitcoin L1 at any time
• No entity controls the gossip protocol or can prevent new nodes from joining

The real risk is economic, not technical. If a few LSPs become so dominant that wallet developers cannot practically connect elsewhere, those LSPs gain pricing power over routing fees and can influence which payments succeed. This is similar to how ISP consolidation in traditional internet infrastructure creates bottlenecks even when the protocol itself is open.

Transaction Volume Tells a Different Story

While node counts declined, Lightning payment volume grew. River Financial estimated that the network processed approximately $1.17 billion in payments during November 2025, across 5.22 million transactions. That figure represents extrapolated data from partner nodes covering a significant share of network capacity (including ACINQ, Kraken, Breez, Lightspark, and LQWD), so it is approximate.

The growth in transaction volume alongside declining node count reinforces the consolidation thesis: fewer nodes are processing more payments. This mirrors how the early internet transitioned from thousands of small ISPs to fewer, larger providers without reducing total internet traffic.

Volume growth drivers

• Exchange integrations: Coinbase, Kraken, and Binance added Lightning deposits and withdrawals, driving substantial volume through their LSP partners
• Emerging market adoption: El Salvador's Chivo wallet and similar government-backed projects introduced Lightning to users who previously had no Bitcoin exposure
• Wallet improvements: Phoenix's splicing-based UX and Breez's SDK removed much of the friction that deterred casual users
• Stablecoin experiments: Taproot Assets enabled dollar-denominated payments over Lightning channels, opening new use cases

What Healthy Network Evolution Looks Like

The question is not whether Lightning has fewer nodes than it did in 2022. It does. The question is whether the remaining network serves users better. By most operational metrics, it does:

• Higher average capacity per node means more payments can route successfully
• Professional operators provide better uptime and faster gossip propagation
• Transaction volume has grown, suggesting real usage is increasing
• Payment success rates for typical amounts remain high through well-connected paths

But the evolution carries risks. Concentration creates single points of failure: if ACINQ's node goes down, every Phoenix user loses routing capability until it recovers. If Lightspark experiences an outage, Coinbase Lightning withdrawals halt. These are operational risks that did not exist when the network was more diffuse.

Maturation vs. decline: key indicators

By these indicators, Lightning is in a period of maturation with legitimate concentration concerns. It is not dying, but it is becoming a different kind of network than the maximally decentralized vision some early advocates imagined.

The Operational Burden Problem

Lightning node consolidation is ultimately a story about operational burden. Running a profitable routing node requires capital for channel liquidity, ongoing rebalancing, fee policy tuning, software updates across implementations that sometimes introduce breaking changes, and monitoring for force-close events that can lock up capital for days. Most individuals are not willing to commit to that level of ongoing maintenance for minimal financial return.

This dynamic is not unique to Lightning. Any infrastructure network that requires persistent capital commitment and active management will naturally consolidate toward professional operators. The question is whether the protocol provides adequate tools and incentives to maintain a sufficiently diverse set of those operators.

Layer 2 protocols that reduce this operational burden can change the equation. Spark, for example, uses a statechain-based architecture where users do not need to manage channels, monitor for breaches, or plan liquidity. The Spark operator model provides routing reliability comparable to a well-connected Lightning node without requiring end users to run infrastructure. For builders who want Bitcoin payment capabilities without the node operations overhead, the Spark SDK provides a path that sidesteps the consolidation problem entirely.

What to Watch Going Forward

The Lightning node count will likely continue fluctuating in the 15,000 to 18,000 range as the network finds its equilibrium between hobby operators and professional infrastructure. Several developments could shift the trajectory:

• LSPS standardization: common LSP specifications could lower the barrier for new providers, counteracting consolidation
• Splicing adoption: dynamic channel resizing reduces the capital inefficiency that drives small operators out
• Taproot channels: improved privacy and efficiency could attract new node operators
• Economic incentives: if Bitcoin price appreciation makes routing fees more valuable in fiat terms, more operators may find node operation profitable

For a deeper analysis of current Lightning metrics, see the State of the Lightning Network in 2026 overview. Wallet users looking to interact with Lightning and Spark without managing infrastructure can explore General Bread, a Spark-powered wallet that handles both protocols seamlessly.

This article is for educational purposes only. It does not constitute financial or investment advice. Bitcoin and Layer 2 protocols involve technical and financial risk. Always do your own research and understand the tradeoffs before using any protocol.