Bitcoin nodes around the world

I have been running a Bitcoin full node at home since 2017. The current setup is a Raspberry Pi 5 with a 2 TB SSD on my fiber connection in Bangkok. It draws about seven watts and has been quietly verifying every block miners produced for the past several years. It does not earn me anything — does not stake, mine, or yield — and it is the single most useful piece of software I have ever installed.

The world map on this page is full of dots. Each dot is one of those quiet machines somewhere on the planet — Berlin, Virginia, Tokyo, my apartment in Bangkok — running the same software, enforcing the same rules, refusing to be lied to. This article is about what those dots do, why their distribution matters, and why you should add one more to the map.

What a node actually is

A Bitcoin node is software that downloads, verifies, stores, and relays the entire Bitcoin blockchain. The reference implementation is Bitcoin Core, used by the overwhelming majority of operators. A handful of alternatives — btcd (Go), libbitcoin (C++), and a few others — exist at the margins.

A node is not a wallet, miner, or exchange. It is a referee. It checks every transaction and block against the consensus rules — the rules Satoshi wrote in 2009, plus soft-fork upgrades since — and rejects anything that does not conform.

Full vs pruned vs SPV

Not all “nodes” are equal. There are three flavors people commonly mix up:

• Full archival: stores every block since the genesis block in January 2009. As of 2026 the chain is around 600 GB. This is what block explorers and Lightning routing nodes typically run.
• Pruned: verifies the entire chain on the way down, then deletes old blocks once they are buried. Minimum disk requirement is about 20 GB. You still get full validation — you just cannot serve historical blocks to new peers.
• SPV (Simplified Payment Verification): only downloads block headers (~80 bytes per block, currently ~70 MB total) and trusts full nodes to tell the truth about transaction bodies. Most mobile wallets use SPV or a slimmer variant.

When the Bitnodes crawler talks about “nodes,” it means full or pruned nodes — software that fully verifies. SPV clients do not count.

What “running a node” looks like in practice

The first time you start Bitcoin Core, it spends 12–24 hours in initial block download (IBD). Your machine connects to peers, fetches every block since 2009, and replays the entire history of the Bitcoin economy — every transaction, every signature, every UTXO movement — against the rules.

Once IBD finishes, your node enters steady state. It listens for new transactions, gossips them to peers, and when a miner finds a block it validates and either accepts or rejects within a second. Mundane and continuous: sync, verify, reject the invalid, relay the valid. Forever.

Reachable vs unreachable

The map only shows reachable nodes. A reachable node accepts inbound TCP connections on port 8333 (or 8334 for Tor). Most home nodes — including mine, for years — sit behind a NAT or firewall and only make outbound connections. They are still real, fully validating nodes; they just cannot be discovered by a crawler.

Bitnodes typically counts 16,000–20,000 reachable nodes. The actual count is multiples higher — surveys suggest 50,000–100,000+ total when you include the unreachable majority. The map is the tip of an iceberg.

Why nodes matter, in one sentence

Users with full nodes cannot be lied to about the supply, the rules, or the validity of their own transactions.

There are 21 million coins and never one more — not because Coinbase says so, but because tens of thousands of machines refuse to accept any block that violates the schedule. Miners can spend megawatts producing blocks that print extra coins; nodes will throw those blocks straight in the trash.

The decentralization budget is simple arithmetic: more independent nodes in more independent jurisdictions equals more cost to anyone who wants to coerce the network into changing.

What the world map shows

The distribution of reachable nodes on this page is roughly:

• ~30% Europe (Germany leads — Hetzner, OVH, strong home-node culture)
• ~30% North America (USA dominant, Canada non-trivial)
• ~10% Asia (Japan, Singapore, South Korea; mainland China minimal post-2021)
• ~10% Tor / I2P (jurisdiction-less)
• The rest distributed across Latin America, Africa, Oceania, and the rest of Asia

No single country comes close to a majority. That is the property that matters. If 60% of nodes ran inside one jurisdiction, that government would have meaningful coercive leverage. As it stands, you would have to coordinate raids in dozens of countries to meaningfully reduce the count.

The Tor share is a feature, not a bug

Roughly 15–25% of reachable nodes connect over Tor. Tor hides geographic location from crawlers — and, more importantly, from ISPs and governments. It is resistant to ISP port blocking, country-level firewall censorship (China’s GFW), and targeted deanonymization of known operators.

If your country one day decides to make running a Bitcoin node uncomfortable, Tor is the escape hatch. A high Tor share is a sign that the network has built that escape hatch in advance.

User agents — implementation diversity

The Bitnodes user-agent breakdown shows roughly 95% of reachable nodes running Bitcoin Core, spread across many versions (/Satoshi:26.0.0/, etc.). The rest are btcd, libbitcoin, Knots, and a handful of others.

Implementation diversity is a small but real defense. If a consensus-critical bug ever shipped in Bitcoin Core — and it has happened, see CVE-2018-17144 — having even a few percent of the network running a different codebase reduces the chance of a single bug taking down everything.

Why YOU should run a node

Running a node is, in my opinion, the single most important thing any individual can do for Bitcoin. Reasons:

1. You verify every payment you receive yourself — no trusting a block explorer, an exchange, or a wallet provider’s backend.
2. Every additional node makes coercion harder. One more independent vote that the consensus rules will be enforced.
3. Privacy. Your wallet talks to your node, not to some company that logs every address you query. SPV wallets leak metadata constantly. Your own node fixes that.
4. You stop being a customer. You become a participant.

Mining is expensive and largely industrial now. Running a node is the most powerful thing left that any individual can do without capital or permission.

What it actually takes

• Hardware: Raspberry Pi 5 + 1 TB SSD (~$200 all-in). Or any spare laptop — a ten-year-old ThinkPad will do the job.
• Software: Bitcoin Core — free, open source. Download, verify the GPG signatures, run. The official full-node guide walks through it step by step.
• Time: 12–24 hours initial sync (less on a fast connection).
• Bandwidth: ~5–50 GB/month, depending on peers served. Most home internet plans absorb this without noticing.

Plug-and-play options

If you want a friendlier UX, several teams package Bitcoin Core + Lightning + a web dashboard onto pre-built hardware: Umbrel, Start9, and MyNode. Expect ~$400–600 for a complete kit. Worth it if you want one-click and Lightning out of the box. I prefer bare Bitcoin Core — fewer moving parts, fewer attack surfaces — but the boxes are a perfectly good way in.

The Thailand angle

Thailand has roughly 50–100 reachable nodes in any given snapshot. That is not many. The map is dense in Frankfurt and Ashburn; it is sparse from Bangkok to Jakarta to Manila. Running a node here contributes to resilience in a genuinely underrepresented region.

I run mine on residential fiber, ~99% uptime for years. Electricity is negligible — the Pi 5 idles at ~7 W. There is no regulatory ambiguity about running node software in Thailand: you are not custodying anyone’s money, you are running a verifier. If even a few hundred more Thai readers ran nodes, the Asia share on the map would visibly shift.

Your node, your rules

“Your node, your rules” sounds like a slogan, but it is mechanically true. When you choose which version of Bitcoin Core to run, you choose which consensus rules to enforce. If a future soft fork is proposed and you disagree with it, you can decline to upgrade. Your node will keep enforcing the old rules.

BIP-9 and later BIP-8 define version-bit signaling, but the harder fact is that a soft fork only sticks if economically meaningful nodes adopt it. Miners cannot impose changes by themselves.

The decentralization stress test

2017 — SegWit and UASF. Miners dragged their feet on SegWit. Node operators ran BIP-148 — the User-Activated Soft Fork — which threatened to orphan any block that did not signal SegWit support. Miners capitulated within weeks. Nodes won.

2024 — ordinals and runes. A loud debate over filtering inscriptions at the mempool level. Node operators voted with their software (Core, Knots, custom mempool policies), and that distribution shaped the Bitcoin Core team’s stance more than any single statement could.

Nodes are the constitution. Miners are the executive. The constitution wins when the executive overreaches.

Run a node

Look at the map. Find your country. Notice how many dots are there, and how many are not. Then go to bitcoincore.org, download the software, plug in an SSD, and add yourself to it.

It is the cheapest, most powerful, most underrated political act available to a person who cares about Bitcoin.