Finding · free in full
The Sleeping Fleet
Every Satoshi-era mining reward, classified one coin at a time.
In Bitcoin's first years, mining rewards were paid straight to a raw public key — a format called P2PK that fell out of use around 2013. These are the oldest coins in existence: the ones people mean when they say "Satoshi-era".
We took every single one of them from our own copy of every block, and asked one question of each: is this exact coin still sitting untouched today?
Still asleep — 34,185 mining rewards, out of 196,479 ever paid this way, that have never once been spent.
Outputs examined
196,479
Never moved
34,185
Block range
0 – 955,609
Departures dated
157,766
The result
Two thirds of Bitcoin's first year never spent a coin
Sorted by the year each reward was mined, the pattern is stark and it only points one way. Of everything mined in 2009, 66.13% by value has still never moved. By 2011 that has collapsed to 1.52%. For every coin mined this way between 2014 and 2017, the figure is zero — the professional era left nothing on the table.
| Mined | Outputs | BTC mined | Still asleep | BTC asleep | % BTC asleep |
|---|---|---|---|---|---|
| 2009 | 32,490 | 1,624,502.87 | 21,487 | 1,074,352.49 | 66.13% |
| 2010 | 67,920 | 3,395,543.99 | 11,473 | 573,547.42 | 16.89% |
| 2011 | 58,477 | 2,871,586.36 | 888 | 43,716.11 | 1.52% |
| 2012 | 31,082 | 1,460,941.34 | 200 | 9,443.30 | 0.65% |
| 2013 | 5,049 | 110,676.43 | 118 | 2,962.71 | 2.68% |
| 2014 | 1,091 | 15,099.53 | 0 | 0.00 | 0.00% |
| 2015 | 230 | 5.30 | 0 | 0.00 | 0.00% |
| 2016 | 46 | 0.98 | 0 | 0.00 | 0.00% |
| 2017 | 70 | 0.30 | 0 | 0.00 | 0.00% |
| 2018 | 12 | 0.09 | 7 | 0.08 | 92.42% |
| 2019 | 5 | 0.54 | 5 | 0.54 | 100.00% |
| 2020 | 1 | 0.02 | 1 | 0.02 | 100.00% |
| 2022 | 2 | 0.00 | 2 | 0.00 | 100.00% |
| 2023 | 1 | 0.00 | 1 | 0.00 | 100.00% |
| 2024 | 1 | 0.00 | 1 | 0.00 | 100.00% |
| 2025 | 1 | 0.00 | 1 | 0.00 | 100.00% |
| 2026 | 1 | 0.00 | 1 | 0.00 | 100.00% |
The rows from 2018 onward are the last P2PK curiosities — a handful of anachronistic raw-pubkey coinbases, worth almost nothing and nearly all untouched.
The other half of the question
When the rest woke up
Knowing which coins slept is only half a finding. We also dated every departure — the exact block at which each of these keys was finally spent. 157,766 awakenings, from our own address index, no external source consulted.
| Woke | Keys | BTC |
|---|---|---|
| 2009 | 2,723 | 136,150.13 |
| 2010 | 46,407 | 2,320,043.91 |
| 2011 | 67,735 | 3,346,128.25 |
| 2012 | 30,696 | 1,500,271.18 |
| 2013 | 7,052 | 256,758.75 |
| 2014 | 624 | 30,632.67 |
| 2015 | 166 | 8,102.88 |
| 2016 | 272 | 13,129.14 |
| 2017 | 930 | 46,430.43 |
| 2018 | 151 | 7,526.31 |
| 2019 | 87 | 4,325.51 |
| 2020 | 132 | 6,600.01 |
| 2021 | 246 | 11,778.74 |
| 2022 | 28 | 1,400.36 |
| 2023 | 41 | 1,850.03 |
| 2024 | 264 | 11,720.69 |
| 2025 | 85 | 2,375.34 |
| 2026 YTD | 57 | 2,450.04 |
Read the shape of it. The great cash-out ran from 2010 to 2012 — roughly 7.2 million BTC of movement, in the years when a price first existed at all. After that, a long tail that pulses with every bull market: the 2017 spike, 2020–21, and a distinct revival across 2024–2026.
And the oldest coins in Bitcoin are still waking up right now. Fifty-seven keys have moved for the first time so far this year.
Method
The trap this registry refuses
Block 1's coinbase — Satoshi's first mining reward after the genesis block — shows blockchain activity in 2024 and 2025. Read at the level of the script, it looks alive. A careless analysis headlines it: "Satoshi's coins moved."
They did not. Bitcoin Core's own gettxout confirms the
50 BTC sits unspent with around 960,000 confirmations. What the history actually
shows is strangers sending homage deposits to Satoshi's public key — money going
in, nothing coming out.
So this registry never classifies at the script level. It asks, for one exact coin at a time, whether that specific outpoint is still present in the unspent set — a set which Bitcoin Core itself certified against our copy, to the satoshi, on the day of the measurement.
Departures are dated only for keys with unambiguous fund-then-spend histories. Keys that were reused — 70 of 157,766 — are recorded as deferred rather than guessed at.
Controls
Before any figure above was allowed out, the following had to hold. Each is a case where we already know the true answer, so a wrong result would prove the instrument broken rather than the finding interesting:
- Block 1 classifies as sleeping, confirmed independently by
gettxout— the homage-deposit trap above. - Block 170 classifies as sleeping.
- Block 9 classifies as awoken, spent at height 170. That is the Satoshi → Hal Finney transaction: the first bitcoin ever sent to another person, read straight off our own disks.
- The sleeping totals match the certified unspent set's P2PK coinbase rows to the satoshi on both sides — 34,185 outputs and 170,402,266,597,442 sats.
Reproduce it
The registry lives in six tables. The headline figure is one query:
SELECT count(*) AS outputs_total,
sum(sleeping::int) AS outputs_asleep,
sum(CASE WHEN sleeping THEN value_sats END) / 1e8 AS btc_asleep
FROM fleet_outputs;Which returns, on the sealed dataset: 196479 | 34185 | 1704022.66597442.
The full build runs end to end from raw block files and a live node — no external data
source is consulted at any point.
This covers P2PK coinbase outputs — mining rewards paid to a raw public key. It is not a census of all early Bitcoin, and not every P2PK coin is Satoshi's; the "Satoshi-era" label describes the period, not the owner. We make no claim about who controls any key here.
Data as of block height 959,975 (unspent-set snapshot) and 959,976 (chain parse),
2026-07-28. The unspent set was certified bit-exact against Bitcoin Core's own
coinstatsindex the same day.
The dataset
Everything above, as files
The finding is free and complete — everything needed to check it is on this page. The underlying registry, all six tables, coin by coin and departure by departure, is listed with its full schema here.