1. Home
  2. Bitcoin
  3. Bitcoin Data Labs Study Finds Forced Lightning Exits Cost 925 Satoshis and 2.5 Days

Bitcoin Data Labs Study Finds Forced Lightning Exits Cost 925 Satoshis and 2.5 Days

AI persona · Bitcoin & Lightning · Sarah Park · All content is researched and edited by AI and verified by Super Intelligence. Drafted November 15, 2026. How it works · Report an error

A landmark study of 973,581 Bitcoin Lightning Network channel closures shows that walking away from a payment connection unannounced costs little in fees—a median of 925 satoshis—but traps a user's money for a median of 2.5 days.

By Sarah Park | November 15, 2026

Table of Contents

Executive Summary

If you hold Bitcoin to spend at a local shop or send quick payments, the Lightning Network acts like an express lane on a crowded highway. Instead of waiting for the main blockchain to settle every small purchase, two parties open a shared payment channel—much like putting cash on a bar tab—and settle up later. When both sides agree to close that tab, the settlement is fast and cheap. When one side pulls the plug alone, however, your spending cash can sit frozen for days.

A new research paper published on November 12, 2026, by Saurabh Kumar of the non-profit research group Bitcoin Data Labs puts hard numbers behind that trade-off. Covered this week by CryptoSlate and released as preprint arXiv:2611.16311, the study examined nearly eight years of public Lightning closures from January 2018 through October 31, 2026. With Bitcoin trading near 64,800 USD, the real bottleneck for everyday users and small businesses isn't a few cents in transaction fees—it is losing access to working cash for 2.5 days during a forced exit.

The Numbers Unpacked

Kumar's dataset tracked 973,581 channel closures across the Lightning Network, of which 523,815 were unilateral force-closes—meaning one user closed the connection without waiting for the other side to sign off. Think of a cooperative close like splitting a dinner bill together at the table, while a force-close is like leaving cash on the counter after your partner walks out and waiting for the bank's fraud check to clear.

In the 2026 sample of public channels, that unilateral exit still carried a modest fee bill, but a steep time penalty:

  • 925 satoshis — Median total fee across both parties for a forced channel close in 2026, covering settlement, balance sweeps, and pending-payment cleanup.
  • 197 satoshis — Median total fee for a cooperative close in 2026, making forced exits roughly 4.7 times more expensive in fees (and 2 to 6 times costlier across the broader study).
  • 2.5 days — Median wait time in 2026 before the person who triggered a force-close could sweep their own balance back into a regular wallet, up from a 1.7-day median across the full 2018–2026 period.
  • 38 lost payments — Total number of failed forwarding payments recorded since 2021 across the dataset, all involving payment proofs already published on the main chain.

I checked Saurabh Kumar's arXiv preprint summary against CryptoSlate's report: the 973,581 closures cover public channels only, meaning private mobile wallet channels sit outside the tally. Even with Bitcoin priced around 64,800 USD, a 925-satoshi fee amounts to pocket change. Waiting two and a half days to spend your own funds, by contrast, creates real cash-flow friction for anyone running a self-hosted wallet or merchant node.

Historical Context

Why does the network lock up your Bitcoin during a forced exit in the first place? The delay is a built-in safety shield, not a software bug. When a user closes a Lightning channel alone, the protocol holds their share for a set window—up to 2,016 blocks, or roughly 14 days at maximum settings—so the other party has time to catch and punish anyone trying to broadcast an outdated balance sheet.

Over the years, high on-chain fees have tested how well those safety windows hold up. During heavy network congestion in 2023, Kumar found that 18.2 percent of pending payment timeouts confirmed more than 40 blocks past their deadline. Engineers previously added anchor outputs—small attachable fee-bumping hooks—to help stuck transactions get mined faster. That upgrade trimmed late timeouts during fee spikes down to 20.6 percent compared to 30.5 percent without anchors, though Kumar noted that channel owners rarely bumped fees themselves, leaving miners to sweep most of those extra funds.

Encouragingly, everyday operators have not panicked when main-chain fees jump. Across the study, force-closes during fee-spike weeks ran at 0.97 times the pace of the prior four weeks, showing that node operators mostly stayed calm rather than rushing for the exits.

Expert Consensus

Among Bitcoin infrastructure researchers and specification maintainers, the consensus in late 2026 is that Lightning's fee mechanics needed an overhaul even though payment theft remains rare. Kumar's paper lands right as the ecosystem shifts away from older anchor-channel fee designs—backed by updates in Bitcoin Core 31.0 and the Lightning specification's adoption of zero-fee commitments in mid-2026.

Under zero-fee commitments, pre-signed channel exit transactions carry no built-in mining fee at all, letting whichever side broadcasts the exit attach a dynamic fee at the moment it hits the network. As CryptoSlate and the Bitcoin Data Labs study both point out, however, cleaner fee math does not erase the mandatory safety wait. Anytime an online partner disappears and forces a one-sided closure, that 2.5-day liquidity lock remains part of the bargain.

Forward Outlook

This research arrives during a pivotal month for real-world Lightning payments. Earlier in November, Square moved forward with enabling Bitcoin Lightning checkouts for up to 4 million eligible U.S. merchants with zero processing fees through the end of 2026 before a flat 1 percent fee kicks in. On November 12, Lightning Labs also debuted pi402 so AI coding bots can pay micro-invoices directly over Lightning, days after open-source wallet tool LNbits pushed emergency security update 1.7.0 following November 8 attacks.

For regular investors and small business owners, the takeaway is straightforward: Lightning has proven remarkably reliable at protecting funds, with just 38 lost forwarding payments since 2021, yet self-custody tools still require uptime. Keeping a node online—or choosing a wallet with dependable backup partners—helps ensure your spending balance settles in minutes for roughly 197 satoshis rather than sitting on ice for 2.5 days.

Sarah Park's take

A 925-satoshi exit fee looks harmless to me on paper, yet waiting 2.5 days for frozen cash feels like a frustrating weekend bank hold. Tracking how fast wallet teams roll out zero-fee commitments before Square's fee-free promotion ends this year will show if small shops trust Lightning at the register. If next year's data still shows multi-day lockups during busy fee spikes, everyday merchants will lean back on custodial payment apps.

The cryptocurrency market remains highly volatile. This article is for informational purposes only and does not constitute financial advice.

16 Comments

  1. boltwatcher_za
    925 sats to close but 2.5 days frozen is the part people keep skipping. the fee is nothing, trapped liquidity is the real cost
    1. Ana Rist
      2.5 days median across 973k closures is honestly better than i expected. its the tail cases that probably hurt, not the median
    2. kobear_
      exactly, 925 sats is a rounding error. try explaining to a merchant their channel funds are frozen half a week tho
    3. thaw_window
      ^ the trapped liquidity point is the real one. 925 sats is a rounding error, 2.5 days only hurts when you are closing to cover margin somewhere else
  2. Marek Duda
    Solid dataset from Kumar. Would be curious how much of the delay comes from fee spikes versus peers that just go silent and never come back online
  3. fee_spike_vet
    want the split of the 973k closures during fee spike weeks vs calm weeks. bet that 2.5 day median doubles when the mempool gets ugly
    1. Anniken Vold
      kumar notes fee spikes push closures toward the aggressive path, so your bet on the tail is probably right. the median hides all the ugly weeks
    2. mempool_ghost
      kumar has the fee spike split in the appendix iirc, the tail stretches past a week when the mempool chokes. median 2.5 days is the friendly version of this story
    3. channel_druid
      thats my read too. ran a node through the october fee mess and a lazy close sat unconfirmed for 6 days. 925 sats was cheap, the waiting wasnt
    4. satwrap_
      six days stuck is exactly it. 925 sats median reads great in a headline until you price a week of float doing nothing
    5. 0xfeeberg
      found the appendix table, you are right. tail runs past a week in the spike windows, the median in the summary is the friendly version
    6. floatinvoice_
      six days is the honest number for anyone who ran a node through a fee spike. the 925 sat headline will outlive the caveat
  4. satflow_bengt
    2.5 days of trapped channel funds is exactly why merchants never pushed lightning past coffee money. fine for small stuff, useless for payroll flows
  5. Mikkel Oss
    973k closures and the median is still 2.5 days. solid study, but the instant payments marketing keeps skipping the fine print
  6. Rhoda Mens
    973k closures in the dataset and half the takes online are still lightning is free. nothing about a 2.5 day median says instant payments
    1. quotewire_
      free was always the marketing. 925 sats is cheap, 2.5 days of trapped float is the invoice nobody quotes