How Bitcoin stays alive when banks and card networks go down

In 2019, Rodolfo Novak sent a Bitcoin transaction from Toronto to Michigan without internet or satellite. He used a ham radio, the 40-meter band, and the ionosphere as his relay.

Nick Szabo called it “Bitcoin sent over national border without internet or satellite, just nature’s ionosphere.” The transaction was tiny, the setup finicky, and the use case borderline absurd.

Yet, it proved something: the protocol doesn’t care what carries its packets.

That experiment sits at one end of a decade-long stress test the Bitcoin community runs quietly in the background, a distributed R&D program testing whether the network can function when the usual infrastructure fails.

Satellites broadcast blocks to dishes across continents. Mesh radios relay transactions across neighborhoods without the need for ISPs. Tor routes traffic around censors. Ham operators tap out hexadecimal over shortwave.

These aren’t production systems. They’re fire drills for scenarios most payment networks treat as edge cases.

The question driving it all: if the internet fragments, how fast can Bitcoin come back online?

Satellites give Bitcoin an independent clock

Blockstream Satellite broadcasts the full Bitcoin blockchain 24/7 via four geostationary satellites covering most populated regions.

A node with an inexpensive dish and a Ku-band receiver can sync blocks and stay in consensus even if local ISPs go dark.

The system is one-way and low-bandwidth, but it solves a specific problem: during regional blackouts or censorship, nodes need an independent source of truth for the ledger state.

The satellite API extends this further. Anyone can uplink arbitrary data, including signed transactions, from ground stations for global broadcast. goTenna partnered with Blockstream to let users compose transactions on offline Android phones, relay them via local mesh, then hand them to a satellite uplink that broadcasts without touching the wider internet.

The bandwidth is terrible, but the independence is absolute.

This matters because satellites provide an “out-of-band” channel. When regular routing fails, nodes scattered across different continents can still receive the same chain tip from space, providing a shared reference point for rebuilding consensus once terrestrial links return.

Mesh and LoRa build Bitcoin backhaul at human scale

Mesh networks take a different approach: instead of broadcasting from orbit, they relay packets device-to-device across short hops until one node with internet access rebroadcasts to the broader network. TxTenna, built by goTenna, demonstrated this in 2019.

Users send signed transactions over a mesh network from offline phones, hopping node to node until reaching an exit point. Coin Center documented the architecture: each hop extends reach without requiring any participant to have direct internet access.

Long-range LoRa mesh pushes this concept further. Locha Mesh, started by Bitcoin Venezuela, builds radio nodes that form an IPv6 mesh over license-free bands.

The hardware, Turpial and Harpia devices, can carry messages, Bitcoin transactions, and even block sync over several kilometers without an internet connection.

Tests in disaster zones proved successful crypto transactions across multi-hop networks where cellular and fiber were both down.

Darkwire fragments raw Bitcoin transactions into small packets and relays them hop-by-hop over LoRa radios. Each node reaches roughly 10 kilometers of line of sight, turning a neighborhood of hobbyist radios into ad hoc Bitcoin infrastructure.

Urban range drops to a 3 to 5 kilometers range, but that’s enough to route around localized outages or censorship chokepoints.

Academic projects like LNMesh extended this logic to Lightning Network payments, demonstrating offline micropayments over local wireless mesh during power outages.

The volumes are small and the setups fragile, but they establish the principle: Bitcoin’s physical layer is fungible. As long as there exists a path between the nodes, the protocol functions.

Tor and ham radio fill the gaps

Tor represents the middle ground between the regular internet and exotic radio. Since Bitcoin Core 0.12, nodes automatically start a hidden service if a local Tor daemon is running, accepting connections via .onion addresses even when ISPs block known Bitcoin ports.

The Bitcoin Wiki and Jameson Lopp’s setup guides document dual-stack configurations in which nodes route traffic over both clearnet and Tor simultaneously, complicating efforts to censor Bitcoin traffic at the ISP level.

Experts warn against running nodes exclusively over Tor due to eclipse-attack risks, but using it as one routing option among several substantially raises the cost of blocking Bitcoin infrastructure.

Ham radio sits at the far end of the spectrum. Beyond Novak’s ionosphere experiment, operators have relayed Lightning payments via amateur radio frequencies.

These tests involve manually encoding transactions, transmitting them over HF bands using protocols like JS8Call, then decoding and rebroadcasting on the other side.

The throughput is laughable by modern standards, but the point isn’t efficiency. The point is demonstrating that Bitcoin can move across any medium capable of carrying small data packets, including ones that predate the internet by decades.

What a global partition actually looks like

Recent modeling explores what happens during a prolonged global internet outage.

One scenario splits the network into three regions, Americas, Asia-Pacific, and Europe-Africa, with roughly 45%, 35%, and 20% of hash rate, respectively.

Each partition’s miners continue producing blocks while adjusting the difficulty independently. Local exchanges build their own fee markets and order books on diverging chains.

Within each partition, Bitcoin continues working. Transactions confirmed, balances updated, local commerce proceeds, but only within that island. Cross-border trade freezes. When connectivity returns, nodes face multiple valid chains.

The consensus rule is deterministic: follow the chain with the most cumulative proof of work. Weaker partitions are reorganized, and some recent transactions are removed from global history.

If the outage lasts hours to a day and the hash distribution isn’t wildly skewed, the result is temporary chaos followed by convergence as bandwidth returns and blocks propagate.

Prolonged outages create the risk that social coordination will override protocol rules, exchanges, or that large miners will choose their preferred history. Still, even that remains visible and rule-bound in ways that traditional financial reconciliation is not.

Banks don’t have fire drills for this

Contrast that with what happens when payment infrastructure breaks. TARGET2’s 10-hour outage in October 2020 delayed SEPA files and forced central banks to manage liquidity and collateral manually.

Visa’s Europe-wide failure in June 2018 saw 2.4 million UK card transactions fail outright and ATMs run dry within hours after a single data center switch died.

The ECB’s TARGET system suffered another major outage in February 2025, prompting external audits after backup systems failed to activate.

IMF and BIS documentation on CBDC and RTGS resilience explicitly warns that large-scale power or network outages can simultaneously hit primary and backup data centers, and that centralized payment systems require complex business-continuity planning to avoid systemic disruption.

The architectural difference matters. Every Bitcoin node holds a full copy of the ledger and validation rules.

After any outage, as soon as it can communicate with other nodes, via satellite, Tor, mesh, or restored ISP, it simply asks: what’s the heaviest valid chain?

The protocol defines the resolution mechanism. No central operator reconciles competing databases.

Banks depend on a layered, centralized infrastructure comprising core banking ledgers, RTGS systems such as Fedwire and TARGET, card networks, ACH, and clearinghouses.

Recovery involves replaying queued transactions, reconciling mismatched snapshots, sometimes manually adjusting balances, then bringing hundreds of intermediaries back into sync.

Visa’s 2018 outage took hours to diagnose despite a full-time operations team. The ECB’s TARGET incidents required external reviews and multi-month remediation plans.

Bitcoin practices for worst-case scenarios

So, in a crisis, a plausible scenario emerges: a subset of miners and nodes stays synchronized via satellite and radio, maintaining an authoritative chain tip even as fiber and mobile networks fail.

As connectivity returns in patches, local nodes pull missing blocks and reorganize to that chain within minutes to hours.

Meanwhile, banks figure out which payment batches settled, reschedule missed ACH files, and wait for RTGS systems to complete end-of-day reconciliation before reopening fully.

This doesn’t mean Bitcoin “wins” instantly. Card rails and cash still matter for consumers. But as a global settlement layer, it might reach a consistent state faster than a patchwork of national payment systems, precisely because it’s been running continuous fire drills for world-scale failure modes.

The ham operators tapping out transactions over shortwave, the Venezuelan mesh nodes routing sats across blackout neighborhoods, the satellites broadcasting blocks to dishes pointed at the sky, and these aren’t production infrastructure.

They’re proof that when the usual pipes break, Bitcoin has a Plan B. And a Plan C. And a Plan D that involves the ionosphere.

The banking system still treats infrastructure failures as rare edge cases. Bitcoin is treating it as a design constraint.