Lightning-Funded Bitcoin Cards

4 cards on this page take bitcoin, and only 4 of them fund over Lightning - but they are the reason this page exists. Card loads are small and frequent by nature, which is exactly the profile on-chain fees punish and Lightning was built for. A $10 on-chain load can lose more to the flat miner fee than to any operator margin, and it waits for a confirmation; the same $10 over Lightning settles before you lower the phone, at rail fees that round to cents. That is why Lightning-native cards are the only crypto cards where the minimum-load row is a non-issue - and the only ones where "top up at the register" is a normal gesture rather than a planned event.

The trade for that ergonomics is the layer below the rail: every program on this page credits your card balance as a claim against the operator, settled in milliseconds on a network that never asks who you are. The rail removes the cost problem and keeps the trust problem intact. This page works both.

The rail, precisely

The mechanics are short because the protocol is [1][2]: the card side publishes a BOLT 11 invoice or a static LNURL-pay code, your wallet routes the payment, the program's node settles and the balance is spendable. Three properties matter for card use specifically:

• Finality is instant and atomic. A Lightning payment either completes or returns intact - there is no pending state to reconcile against the card balance, no "pending top-up" limbo, no wrong-chain loss mode. Ten loads a week means ten clean settlements, not ten confirmation waits.
• The static QR (LNURL-pay) is the ergonomics unlock [2]. A fresh invoice per load is fine for a one-off; a permanent Lightning Address like Freedomia's [6] turns funding into one scan from any wallet - the difference between a rail you plan around and one you use reflexively.
• Failure is free. An unroutable payment costs nothing and moves nothing. Compare a stuck on-chain deposit to a card program's support queue and the value of that property becomes obvious.

Who actually ships a Lightning card

4 of the 4 live cards on this page take Lightning funding, and they differ in what the rail connects to. The table is the live population; the honest detail is in what the table cannot show - Moon is watch-listed, Blink is app-gated, and Cash App's Lightning is a wallet rail more than a card rail.

Live values from our database - details and caveats in each review. "-" = not published by the operator.

The consumer-scale implementation is Cash App: free Lightning transfers between compatible wallets, wrapped in a US banking stack with Sutton Bank issuing the prepaid card [8]. The card itself spends USD; the Lightning rail feeds the wallet that fills it. That two-step is not a limitation to hide - it is the model every regulated operator converges on, because the card network settles in fiat and someone must own the conversion moment.

What a Lightning balance is - and is not

The invoice settles in satoshis; the card spends dollars. Between those two facts sits the entire trust model. When the program's node accepts your payment, it owes you - the balance is a ledger entry, not a UTXO you control. Three documented variants:

• Convert on the way in (Freedomia): the balance is USD, exposure ends at the rail, each load is one disposal for tax purposes.
• Hold until authorization (Blink): the agreement states conversion "at the prevailing rate at the time of the Transaction" [7] - exposure lasts to the terminal, and a falling balance can decline a payment.
• Wrap the wallet, not the card (Cash App): the card spends USD, the wallet behind it holds BTC and speaks Lightning [8].

The self-custody hub covers the genuinely non-custodial end of the market; pure Lightning-native cards with no custodial conversion step do not exist in the live population at check time. Anyone marketing one is redefining the word.

Fee math: the rail versus everything around it

Lightning's reputation for "free" comes from P2P transfers, not from card programs. The rail fee is real but small: Freedomia documents 0.25-0.50% through providers such as Satora or Boltz [6]. The larger costs live around the rail - conversion spread at load or spend, the program's purchase fee, the monthly subscription, and ATM margins if you withdraw. On-chain for comparison: base fees sat at 1 sat/vB across all confirmation tiers on 2026-10-05 [4], so the network fee itself is cents - the on-chain penalty for small loads is the flat fee's proportion, not its size, plus the wait. Run your actual top-up pattern through the load planner before choosing a rail; the answer flips at roughly the $50-100 load size depending on the operator's fixed fees.

Field notes

• Static beats dynamic. A program with a permanent Lightning Address (Freedomia [6]) makes weekly $20 loads a two-second gesture. Per-load invoices make the same habit a chore - and habits decide whether a small-load strategy survives.
• Test with $5, not $500. Lightning routing is liquidity-dependent: a large load can fail on capacity where small ones sail through. First contact with a new program should cost you a coffee.
• Watch the conversion layer, not the rail. A 0.25% Lightning hop next to a 2.5% FX fee [6] is a rounding error next to the spread. The spread guide is the companion read.
• Mind the refund path. Moon refunds in on-chain BTC [9] - a refund to a Lightning-funded card becomes an on-chain deposit, with its own fee and confirmation wait. Small print with real consequences if you shop refund-heavy merchants.
• Screenshots before funding. Balance, mode (custodial or not), limits - capture them at first load. When a program's terms change under you, the frozen-balance guide explains your actual position; none of it improves without evidence.

Warnings and hard limits

• Moon is watch-listed, not recommended. The API docs claim continuity [9]; the operator's November 2025 mass card-blocking, frozen payout complaints and forced re-verification are documented in our Bitcoin-Check for Moon. Working infrastructure with broken trust is the exact shape a watch status exists for.
• Blink is app-gated and verify-statused. The cardholder agreement is public and current (v2.3) [7], but eligibility "as reflected in the Blink App" means the jurisdiction list is whatever the app shows you at signup - there is no public country matrix to check before installing.
• Capacity is a real ceiling. 4 programs rely on a network with ~2,698 BTC of public capacity [3]. Consumer loads fit easily; a single four-figure load may not route without MPP, and liquidity varies by hour. Do not schedule a rent payment over a rail you have not tested at that size.
• No KYC is not no law. Freedomia's no-KYC tier lives inside published limits and a sanctioned-country blocklist [6] - and operators that outgrow their regulator tolerance have historically added KYC retroactively, freezing balances that assumed the rules were permanent.

The on-chain vs Lightning guide covers the funding decision in depth; the spend hub covers what happens after any rail lands your balance. The rail is the cheap part - the program layer is where every Lightning card on this page earns its score.

Sources

1. BOLT 11 - the invoice format every card top-up starts from - accessed 2026-10-05
2. LNURL LUD-06 - static QR payRequest flow behind "scan to load" - accessed 2026-10-05
3. 1ml.com live statistics - 5,909 nodes, 19,815 channels, 2,697.80 BTC capacity - accessed 2026-10-05
4. mempool.space fee API - 1 sat/vB across all confirmation tiers (2026-10-05) - accessed 2026-10-05
5. Bitcoin wiki - dust limits, 294-546 sat, policy not consensus - accessed 2026-10-05
6. Freedomia FAQ - Lightning Address funding, 0.25-0.50% rail fee via Satora/Boltz - accessed 2026-10-05
7. Blink cardholder agreement v2.3 - balance conversion at transaction time, custody modes - accessed 2026-10-05
8. Cash App on the App Store - free Lightning Network transfers, Block Inc. - accessed 2026-10-05
9. Moon developer docs - API v1/v2 "live and supported", changelog 2026-08-18 - accessed 2026-10-05