Market cap $2.14 T −3.58 % over 24 hours24h volume $74.99 BBTC dominance 57.9 %ETH 9.9 %Fear & Greed 10 Extreme fearCoins 21,366

Bitcoin quantum exposure census counts 7.16 million BTC behind keys already visible on-chain

A Bitcoin quantum exposure census counts 7.16 million BTC, 35.6% of unspent supply, behind keys or scripts already on-chain. It measures visibility.

10 min read

Abstract stacks of coins watched by an eye above blocks and a pie chart
Table of contents
  1. What “exposed” means in a Bitcoin quantum exposure count
  2. 73% of the 7.16 million BTC is counted by hash matching
  3. Address reuse, the part that depends on the holder
  4. The limits the author sets on the count
  5. How the figure compares with Glassnode and earlier estimates
  6. What the sources say on timing, and what they leave open
  7. Three draft BIPs, none with an activation date

A count posted on the Bitcoin developer forum delvingbitcoin.org on 18 November 2026 finds that 7,164,754 BTC, or 35.6% of the 20,109,207 BTC held in unspent outputs at block 975,019, sit in outputs whose public key or script can already be read on the blockchain. The author, who posts as duncan0k, published the figures, the code and the lookup tables so that others can rerun them. The post is a census of Bitcoin’s quantum exposure. It describes no new algorithm and no new machine.

The total itself is close to what was already known. Glassnode counted 7.12 million BTC, 35.4% of issued supply, in an update dated 10 November, according to Brave New Coin, and the text of draft BIP 361 says that “over 34%” of all bitcoin had revealed a public key by 1 March 2026. What the new post adds is a split by output type, a count of how often the same key was used under two address formats, and a list of what the method cannot see.

The census’s own method page puts the limit in one sentence: “A public key on chain does not let anyone take the coins today.”

What “exposed” means in a Bitcoin quantum exposure count

Bitcoin is spent by signing with a private key, and the network checks the signature against the matching public key. Working back from a public key to its private key is not practical for today’s computers. A large enough quantum computer running Shor’s algorithm could do it, which is the threat that draft BIP 360 describes. Such an attacker needs one input, the public key. Our guide to how private and public keys work covers the basics.

Most Bitcoin addresses do not show that key. They carry a hash of it, and the key itself is written on-chain only when the owner spends. Until then there is nothing for Shor’s algorithm to work on.

BIP 360 separates two situations. In a “long exposure” attack the key has been visible for longer than a transaction takes to confirm, so the attacker has all the time needed. In a “short exposure” attack the key appears only while a payment waits in the mempool, and the attacker must beat the confirmation. The census counts the first kind only: coins whose key or script was already public when the snapshot was taken, on 9 November 2026 with Bitcoin Core v28.1.0.

73% of the 7.16 million BTC is counted by hash matching

The count is built from two very different blocks. The first, 1,937,232 BTC, is exposed by format alone. Pay-to-public-key (P2PK) outputs, the format of Bitcoin’s first years, write the key straight into the output: 44,972 of them still hold 1,714,619 BTC, 8.5% of everything unspent. Taproot (P2TR) outputs also carry a key in the output script. There are 53 million of them, yet together they hold 222,545 BTC, about 1.1% of the total. Old-style bare multisig adds 68 BTC.

The second block is 5,227,522 BTC, about 73% of the count. These coins sit in hash-type outputs (P2PKH, P2WPKH, P2SH, P2WSH), which hide the key or script by design. They are counted only because their hash was found in tables that duncan0k built from the public keys and scripts the scan found on-chain up to block 975,019. The largest of those tables holds 1.28 billion entries. In practice a match means an address that has already been spent from, or whose key was shown elsewhere, and that holds coins today. About 3.30 million BTC of this block are in key-hash outputs and 1.93 million BTC in script-hash outputs.

Output typeUnspent BTCKey or script public, BTCShare of type
P2PKH (legacy address)4,502,293.811,162,510.9125.8%
P2WPKH (native SegWit)8,310,066.712,137,831.7225.7%
P2SH (includes nested P2WPKH)3,678,167.101,177,704.6932.0%
P2WSH1,678,829.15749,474.3444.6%
P2TR (Taproot)222,545.28222,545.28100%
P2PK1,714,618.811,714,618.81100%
Bare multisig68.3668.36100%
Other2,618.1500%
Total20,109,207.367,164,754.1235.6%

Source: duncan0k, delvingbitcoin.org, 18 November 2026; unspent outputs at block 975,019, rounded to 0.01 BTC.

Abstract key under a magnifying glass beside a bar chart and a document
About 73% of the 7.16 million BTC is counted because a hash matched a key or script revealed earlier on-chain.

Read the other way, 12.94 million BTC, 64.4% of the unspent supply, are not counted. Native SegWit (P2WPKH) is the largest type with 8.31 million BTC, and about three quarters of it sits behind keys the scan did not find on-chain.

Address reuse, the part that depends on the holder

The post answers a question raised in the review of the draft: how often is one key used under several formats? The scan found 40,385 keys paid both as P2PK and as P2PKH, 189,857 keys both as P2PKH and as P2WPKH, and 5,011 keys in both compressed and uncompressed form.

The same key gives a legacy address and a SegWit address, and spending from one reveals the key for both. The census finds 6,927 hashes with unspent coins under both formats, 93,972.89 BTC in all. For 3,301 of them, holding 49,154.61 BTC, no key is public yet, and the post notes that a spend from either form “would put the key on chain for both”.

Whether a given coin is in the count therefore depends on two things: the type of output it sits in, and whether the key or script of that address has already been revealed. Brave New Coin, reporting Glassnode’s 10 November update, draws the same line between exposure built into an address type and exposure created by how wallets reuse addresses. P2PK and Taproot outputs belong to the first kind, since the key is in the output itself.

Custody habits weigh heavily. In its May 2026 study, Glassnode found that roughly half of the bitcoin it could label as held by exchanges was exposed, against under 30% of the rest of the supply.

The limits the author sets on the count

The tool was written for draft BIP 362, but it does not count what that draft calls EXPOSED_AT_REST. The differences run in both directions and several are unmeasured, so duncan0k writes that it is not possible to say whether the draft’s definition would give a higher or a lower number.

Coins outside the count are not certified as hidden. A key may have been disclosed off-chain, or on-chain in a position the rules do not read. As a check, the author drew 250 uncounted outputs of each hash type at random and found none whose script had already been spent from, which gives a one-sided 95% upper bound of 1.19% per type. The post calls it “a sample, not a full pass”.

Nor does “public” mean “takeable”. For script-hash outputs, what is visible is a script, and the count says nothing on what an attacker could do with each one. The rules also count P2PK, Taproot and bare multisig outputs whether or not their keys are valid points on the curve: 19,915 outputs holding 0.17 BTC fail that check, the method page says.

The work can be rerun. The tables are published as a single 33.65 GB archive, and a full rescan reproduced them byte for byte in about 3.9 hours, the post says. The thread carried no reply on 19 November at 17:40 UTC, and the author’s forum profile gives no name or employer.

How the figure compares with Glassnode and earlier estimates

Glassnode’s first measure, published on 20 May 2026, gave 6.04 million BTC, 30.2% of issued supply, of which 1.92 million BTC exposed by format and 4.12 million BTC by reuse and custody practices. Its 10 November update raised the total to 7.12 million BTC, with 25.8% of supply exposed through address reuse, including 11.6% held on exchanges, Brave New Coin reported on 12 November. The outlet wrote that the comparison shows a change in measured exposure, not a demonstrated increase in what quantum computers can do.

The census lands within 1% of that total with a different test and a different base (unspent outputs, where Glassnode uses issued supply), and its format-only block of 1.94 million BTC is close to Glassnode’s May figure. Opening the discussion in September, duncan0k wrote that published estimates ran from about 25% to over 34% and that the spread was “definitional rather than measurement error”: tools disagree on whether Taproot counts, or what to report for a script never revealed.

What the sources say on timing, and what they leave open

None of these counts says when, or whether, an attack becomes possible. Glassnode states that its analysis “does not take a position” on that. A Google Quantum AI paper, as summarised by CoinDesk in an article its page dates 3 April 2026, estimated that a future machine with fewer than 500,000 physical qubits could derive a Bitcoin private key in roughly nine minutes. The largest quantum processors at the time had around 1,000, the outlet wrote.

The draft BIPs quote outside timelines. BIP 361 cites a McKinsey report for academic road maps that place a cryptographically relevant quantum computer “as early as 2027-2030”. BIP 360 points to US government guidance (NIST IR 8547) under which elliptic curve cryptography is due to be disallowed in federal systems after 2035. Both are planning dates quoted from other bodies, not measurements of a machine.

Abstract atom and computer chip facing a padlock next to an hourglass
The census measures which keys are visible, not when a machine could use them.

Three draft BIPs, none with an activation date

Draft BIP 362, by duncan0k, is an informational text that sorts outputs into four exposure levels. It was opened as a pull request on 18 September 2026 and is still open; the number does not yet appear in the index of the BIPs repository. Its description says it “changes no consensus, policy or P2P rule and proposes no freezing or forced migration.” On the forum, the user murch wrote on 16 September: “the proposed classification seems trivial to me, so I’m not sure I understand the value-add of a BIP here.” The next day murch added that the BIPs repository was a fitting venue if others found the text valuable. The reuse counts in the new post answer a review comment left by conduition on 4 November.

BIP 360, by Hunter Beast, Ethan Heilman and Isabel Foxen Duke, has draft status. It proposes a new output type, Pay-to-Merkle-Root (P2MR), introduced by soft fork: Taproot’s script tree without the key path, so no key rests in the output. Its authors say it resists long-exposure attacks only and that protection in the mempool “may require” post-quantum signatures, left to a later proposal.

BIP 361, by Jameson Lopp and five co-authors, is also a draft. It would first stop payments to quantum-vulnerable addresses, then restrict spending with today’s signatures after a flag day five years after activation. Its header lists as a requirement a post-quantum signature BIP that is still marked “TBD”. How a proposal moves from draft to code is explained in our guide to Bitcoin Improvement Proposals and who decides.

None of the three texts carries an activation date. The same holds for the consensus cleanup, a separate set of rules that developers began rehearsing on Testnet5 this week. Regulators are watching the migration question: IOSCO’s report on quantum risk, covered earlier today, says blockchains face a particularly hard upgrade. Other networks have put numbers on the cost, as with the XRP Ledger’s quantum-proof accounts.

The BIP 362 pull request stays open for comment, and duncan0k ends the post with a request: “If anything in the rules or counts looks wrong, that’s useful to hear.”

Sources

  1. duncan0k, How much BTC sits behind keys or scripts already on chain (block 975,019), delvingbitcoin.org, 18 November 2026 delvingbitcoin.org
  2. Bitcoin public-key exposure census: method and reproducibility snapshots.ark0.si
  3. BIP362: Output Public Key Exposure Classification, pull request 2294, bitcoin/bips github.com
  4. BIP 360, Pay-to-Merkle-Root (P2MR), draft, bitcoin/bips github.com
  5. Glassnode, Measuring Bitcoin's quantum-exposed supply, 20 May 2026 research.glassnode.com
  6. Brave New Coin on Glassnode's 10 November update, 12 November 2026 bravenewcoin.com

Leave a Reply

Your email address will not be published. Required fields are marked *