Same Money, Different Soul.
Bitcoin and Bitcoin Blake2b share every coin, key and rule up to the fork. The comparison below shows that the money is identical. What differs is who mines it, who filters it, and who sits closest to the table when the protocol changes.
Overview
In August 2026 a minority of Bitcoin users forked the chain. They kept the full history, kept the 21 million cap and kept the keys, but switched proof of work from SHA-256 to BLAKE2b and activated a data-limiting soft fork the majority chain had rejected. This page compares the two chains honestly, including where the newcomer is weaker and where its early promises are already being tested.
Three kinds of claim #
Most write-ups about this fork mix three different things. We keep them apart, because the interesting story lives in the gap between the third and the second.
The seven real differences #
Cryptography and monetary policy are identical everywhere: same 21M, same keys, same scripts, same quantum exposure. So this is not a question of which is harder money. The genuine differences concentrate in seven places.
Mining and decentralization intent
A fresh ASIC lineage (Sia-class BLAKE2b miners) and self-templating tools from day one. But three pools already mine about 72% of blocks, so the reset is an opportunity, not an outcome.
Design coreDecentralization →Blockspace data policy
RDTS limits non-financial data and caps block weight at 800,000 (about 300 KB) until September 2027. Money, not JPEGs, for now.
Design coreWhat the chain is for →Institutional capture
ETFs, corporate treasuries and a holder-funded developer consortium on one side. None of it on the other: uncaptured, but also unfunded.
Sovereignty coreWho funds the developers →Network scale and Lindy
Liquidity, hashrate, 16 years of uptime and 100M+ users. This is Bitcoin's honest moat, and it is enormous.
Bitcoin's advantageSecurity and scale →Legitimacy and the Schelling point
Which is "the real Bitcoin"? A contested minority against the established point of coordination.
Bitcoin's advantageEthos and philosophy →Who builds the blocks
On Bitcoin about four pools template around 70% of blocks. On Bitcoin Blake2b, DATUM self-templating is available and used by some pools, but the largest pool builds templates centrally.
Sovereignty coreWho picks the transactions →Spend freedom and the quantum freeze
Quantum risk is identical. The governance is not: Bitcoin has a live proposal that could freeze about 6.7M BTC, including Satoshi's. Bitcoin Blake2b has no such roadmap.
Sovereignty coreThe quantum fault line →At a glance #
| Bitcoin (SHA-256) | Bitcoin Blake2b | Verdict | |
|---|---|---|---|
| Short label | The incumbent chain, ticker BTC | The sovereign fork. Exchanges list it as BTCB2 or XBT | By degree |
| What it is | The original chain, proof of work is double SHA-256 | A minority hard fork. Proof of work switched to BLAKE2b; ships as Bitcoin Knots 29.4.1 and later | Differs |
| Lineage | Genesis 2009, continuous | Shares Bitcoin's full history to block 961,631. RDTS split at 961,632, BLAKE2b proof of work from 961,640 | Identicalup to the fork |
| Explorer | mempool.space and many others | mempool.guide, plus several community explorers | By degree |
| Node software | Bitcoin Core (majority), Bitcoin Knots (minority) | Bitcoin Knots v29.4.2.knots20260508 (2026-09-21) | By degree |
| Honest summary | Mature, liquid, increasingly institutional | A nascent minority chain growing fast. Many advantages are design intent, not yet measured reality | By degree |
How to read the tables #
Every row carries a verdict. Use the filter bar at the top to collapse the page to only the rows that matter to you. Measured figures are as of 2026-09-26 unless a row says otherwise.
Money
The monetary policy is the same, and that sameness is itself the finding. The magnitudes diverge: Bitcoin wins every scale axis, while Bitcoin Blake2b's claim rests on what has not happened to it yet.
Monetary policy #
| Property | Bitcoin (SHA-256) | Bitcoin Blake2b | Verdict |
|---|---|---|---|
| Max supply | 21,000,000 | 21,000,000, inherited | Identical |
| Divisibility | 1 BTC = 100,000,000 sats | Same | Identical |
| Issuance and halving | Every 210,000 blocks | Inherited. BIP-110 changes no subsidy and no halving | Identical |
| Premine or founder allocation | None, fair launch in 2009 | None, and no new issuance. Holders at the fork height own identical balances on both chains | Identical |
| Inflation trajectory | Disinflationary, toward zero | Same | Identical |
| Unit and ticker | BTC, sats. XBT is Kraken's long-standing code for Bitcoin | No official ticker. Exchanges use BTCB2, and some now use XBT, which is contested for exactly that reason | Differs |
Because the monetary rules are identical, the comparison is not "which is harder money". It is the same money with a different custody of the protocol's soul: who mines it, who filters it, who captured it.
Traits of money #
The eleven classic traits, applied to the two chains. Nearly all are inherited. Only history, costliness and the intent of decentralization move.
| Trait | Bitcoin (SHA-256) | Bitcoin Blake2b | Verdict |
|---|---|---|---|
| Verifiable | High | High, same proof-of-work verified ledger | Identical |
| Fungible | High | High | Identical |
| Portable | High | High | Identical |
| Durable | High | High | Identical |
| Divisible | High, 100M sats per coin | High | Identical |
| Scarce | High, 21M | High, 21M inherited | Identical |
| Established history | High, 16+ years | Low, forked in 2026 | DiffersBitcoin's biggest edge |
| Censorship resistant | High | High at protocol level. Higher against capture, lower against a liquidity exit | By degree |
| Unforgeable costliness | Very high, hundreds of EH/s | Low, about 35 PH/s, though up roughly 12x in three weeks | By degree |
| Openly programmable | High | High, inherited scripting, but RDTS restricts data until 2027 | Differsnarrower by design |
| Decentralized mining | High, but vendor and pool concentrated | A reset by intent: commodity Sia-class ASICs and DATUM. Pool concentration is already re-emerging | By degreethe core wager, still open |
Why it has value #
| Value driver | Bitcoin (SHA-256) | Bitcoin Blake2b | Verdict |
|---|---|---|---|
| Digital scarcity | Yes, 21M hard cap | Yes, inherited | Identical |
| Proof-of-work security | Enormous | Small, growing | By degree |
| Decentralization | Concentrated mining | Reset intent, concentration re-emerging | By degree |
| Censorship resistance | High | High, and off the radar | By degree |
| Immutability | Strong | Strong by inheritance, weaker against a 51% attack today | By degree |
| Utility and settlement | Global rails, Lightning | Early. Lightning forks exist only as prereleases | By degree |
| Network effects | Huge, 100M+ users, ETFs | Small | DiffersBitcoin's second moat |
| Inflation hedge | Yes, by policy and narrative | Same policy, only a thin market to express it | By degree |
| Store of value | Deep and priced | Thinly priced, around 0.0037 BTC | By degree |
| Self-custody | Yes | Yes | Identical |
| Transparent public ledger | Yes | Yes | Identical |
| Schelling point | The point of coordination | A contested minority | Differs |
| Non-correlation | Increasingly correlated with institutional markets | Uncorrelated, because illiquid | By degree |
The value properties are shared; the value magnitudes diverge. Bitcoin wins every scale axis: network, liquidity, costliness, time. Bitcoin Blake2b's whole claim sits on the rows marked Differs: data policy, independence from the Schelling point, and not being captured.
Markets and exchanges #
| Property | Bitcoin (SHA-256) | Bitcoin Blake2b | Verdict |
|---|---|---|---|
| Price | Deep, global, 24/7, about $83,900 | About $311 to $326, or 0.0037 to 0.0039 BTC, on two no-KYC exchanges | Differs |
| Market data listings | Everywhere | Not on CoinGecko or CoinMarketCap. Smaller aggregators disagree on market cap by a factor of 1,000 | Differs |
| Exchange listings | Every major venue, plus ETFs | NonKYC and Neoxa, plus a custodial beta OTC desk and a near-dead SafeTrade market. No regulated exchange | Differs |
| Liquidity | Billions per day | Reported volume of a few hundred thousand dollars a day, but only about $40 to $65 of resting orders per side near the best price on NonKYC | Differsvolume figures look inflated |
| On and off ramps | Abundant, KYC'd | Crypto to crypto only, no KYC | Differsfreedom against friction |
| Volatility | High but maturing | Extreme. Reported peak about $3,348 on 2026-09-05, low about $187 two days later | By degree |
| Value framing | Store of value, an asset | Use value and principle first, price second | By degree |
| Sound-money culture | Mainstream "Bitcoin Standard" | Inherited and intensified | Identical |
| HODL, DCA, stacking sats | Yes | Yes | Identical |
| Merchant acceptance | Large | Early. No merchant processor supports it yet | By degree |
| Community size | Huge | Small, high conviction | Differs |
Where it trades #
Read from the exchanges' public APIs on 2026-09-26 around 06:20 UTC. Volume is what the venue reports; see the depth note below.
| Venue | Pairs | Last price | Spread | 24h volume (reported) |
|---|---|---|---|---|
| NonKYC.io | BTCB2/BTC, BTCB2/USDT, AMM pools | 0.00370 BTC, about $311. USDT pair 314.4 | 0.50% (BTC), 0.12% (USDT) | 1,118 BTCB2 on BTC, 657 on USDT |
| Neoxa Exchange | BTCB2/BTC, /USDC, /USDT, /NEOX, /XMR, shown as XBT | 0.00388 BTC, about $326. USDC pair 318.2 | 1.61% | 1,292 BTCB2 on BTC, 658 on USDC |
| AlphaSwap OTC | XBT/USDC | Mirrors the two venues above | 0.09% fee | Custodial public beta, run by a mining pool, withdrawals reviewed by hand |
| SafeTrade | XBT/USDT | About $227 (aggregator) | n/a | About $140, effectively inactive unverified |
Resting orders within 10% of the best price on NonKYC's BTC pair came to about 0.13 BTCB2 bid and 0.20 BTCB2 ask, roughly $40 to $65 per side. That is hard to square with more than a thousand coins of daily volume. Neoxa priced BTCB2 about 5% above NonKYC at the same moment. Both exchanges warn that BTCB2 and BTC share an address format: send the wrong one and it is lost. Neither credits pre-fork balances; you deposit coins you have already split.
Around 2026-09-08 the project persuaded exchanges to relabel BTCB2 as XBT, the code Kraken and ISO-style conventions have long used for Bitcoin itself. The move drew criticism, and it is one reason searches by ticker mislead. Unrelated wrapped tokens on other chains are also called xBTC; they have nothing to do with this fork.
Protocol
BLAKE2b is a fine, modern, slightly faster hash. But faster and newer does not mean more secure, and quantum risk is identical. The hash is not the reason for the fork; hardware politics is. The more substantive technical change is the new block header.
The two hash functions #
The primitives on their own, independent of Bitcoin.
| Property | SHA-256 | BLAKE2b | Verdict |
|---|---|---|---|
| Origin | 2001, designed by the NSA, standardized by NIST (SHA-2) | 2012, Aumasson, Neves, Wilcox-O'Hearn and Winnerlein | Differs |
| Construction | Merkle-Damgård. Length-extension prone, which is why Bitcoin hashes twice (SHA256d) | Derived from BLAKE, a SHA-3 finalist, ChaCha based. Not length-extendable, so no doubling needed | Differs |
| Word size | 32-bit | 64-bit, fast on modern CPUs | Differs |
| Software speed | Baseline | About 20 to 50% faster on modern 64-bit CPUs | Differsmoot for mining: BLAKE2b ASICs already exist |
| Security margin | Battle-tested for 20+ years, no practical break | Equivalent collision and preimage guarantees | Identical |
| Output size | 256-bit fixed | Configurable up to 512-bit | Differsthe chain uses BLAKE2b-256 |
| Extensibility | Needs HMAC for keyed hashing | Native keyed hashing, salt, personalization, tree hashing, variable output | Differsmostly unused by proof of work |
| Quantum risk (Grover) | 256-bit falls to about 128-bit effective | Same | Identical |
| Quantum risk (Shor, signatures) | ECDSA and Schnorr on secp256k1 | Same, inherited | Identicalthe real quantum exposure |
| Standardization | NIST standard, universal | RFC 7693, widely used (Argon2, libsodium, Sia), new as a Bitcoin proof of work | By degree |
Chain technicals #
| Property | Bitcoin (SHA-256) | Bitcoin Blake2b | Verdict |
|---|---|---|---|
| Proof-of-work function | SHA256d, double SHA-256 | BLAKE2b-256 | Differs |
| Block header | 80 bytes | 164-byte v2 header, incompatible with Bitcoin Core after the fork | Differs |
| Pre-fork history | Shared | Byte-identical to block 961,631: same coins, same keys | Identical |
| Signatures | ECDSA and Schnorr on secp256k1 | Same, plus an opt-in unified sighash for replay protection | Identical |
| Address and script format | Bitcoin standard | Inherited, plus RDTS limits | Identical |
| Coinbase maturity | 100 blocks | 6,480 blocks (about 45 days) for coinbases from block 973,440, a temporary soft fork. A longer lock is under discussion | Differs |
| ASIC reality | Mature industry, Bitmain and MicroBT dominate | Mature Sia-class miners (Antminer A3, Goldshell SC series, iBeLink BM-S). A reset, not resistance | Differsnot "ASIC-proof" |
| The 2b versus 2s trap | Not applicable | BLAKE2b is not BLAKE2s. Kadena miners (Goldshell KD line) use BLAKE2s and cannot mine this chain | Differs |
| Difficulty adjustment | Every 2,016 blocks | Every 2,016 blocks, retained. Four maximum 4x increases in a row after the fork | By degree |
| Client maturity | Production | First BLAKE2b release 2026-09-02, current 29.4.2 from 2026-09-21 | By degree |
| Legitimacy | Community consensus | Minority and contested. The BLAKE2b change has no BIP; its spec is a set of Knots pull requests | Differs |
The header redesign and the "dumb miner" #
Post-fork blocks use a mandatory 164-byte v2 header. The ASIC, however, only ever hashes a standard 80-byte Sia work header, so Goldshell SC-series, iBeLink BM-S and Antminer A3 machines mine it unmodified, with no reflash. The miner becomes a minimal hasher; block construction and its commitments live in the node's header, not in the hardware.
| Change | Bitcoin (SHA-256) | Bitcoin Blake2b | Verdict |
|---|---|---|---|
| Transaction-count commitment | Absent, the Merkle ambiguity behind CVE-2017-12842 remains | Header commits to the transaction count, a proper fix | Differs |
| ASICBoost | Possible | Removed | Differs |
| Nonce space | 32-bit, topped up with extraNonce games | 128-bit ASIC nonce plus 128-bit machine nonce | Differs |
| Block-withholding defence | None in the header | An XOR key stops pool workers from recognizing a complete valid block | Differs |
| Timestamp field | 32-bit, runs out in 2106 | 40-bit block time | Differs |
| Timewarp guard (BIP94) | Not enforced on mainnet | Enforced on testnet4, still off on mainnet in 29.4.2. A pull request to enforce it (Knots #391) is open | Differsthe strongest criticism |
The redesign fixes several inherited header bugs and makes the miner a pure hasher. It also ships its own trade-offs: the timewarp guard is off on mainnet, exactly where it matters most, and a bigger header with a merge-mining field is new surface of its own. "Reduces attack surface" is true for the specific bugs fixed, not as a blanket claim.
Mining and governance
The measured state mostly favours Bitcoin. The political fact of this fork is the gap between users and miners. The governance question is who builds the block template, and three weeks of data show that question is not settled by the fork itself.
Security and scale #
Bitcoin Blake2b figures from the mempool.guide API, 2026-09-26.
| Property | Bitcoin (SHA-256) | Bitcoin Blake2b | Verdict |
|---|---|---|---|
| Hashrate | Hundreds of EH/s | About 34.9 PH/s, up from 2.84 PH/s on 2026-09-03 | By degreestill about four orders of magnitude smaller |
| Cost of a 51% attack | Nation-state scale | Low, the price of being new | By degree |
| Difficulty | Around 1014 | About 4.6 billion, up from about 71 million on 2026-09-03 | By degree |
| Block time | 10-minute target | About 8 minutes averaged over the last week | By degree |
| Reachable full nodes | About 25,400 public nodes | No crawler counts this chain. Roughly 1,100 public nodes run the Knots releases that carry the BLAKE2b rules, an upper bound | By degree |
| Network transport | Clearnet, Tor, I2P | The same options. Tor or I2P only is an operator choice, not protocol | By degree |
| Age | Since 2009 | Since August 2026 | Differstime in market is Bitcoin's moat |
Earlier coverage called this chain near-dead at about one block a day. By 2026-09-03 it was at height 966,368 with 2.84 PH/s. By 2026-09-26 it was at 974,193 with about 35 PH/s, after four consecutive maximum difficulty increases. Any number on this page will age quickly.
Decentralization, broken out #
| Axis | Bitcoin (SHA-256) | Bitcoin Blake2b | Verdict |
|---|---|---|---|
| Pool concentration | Top four pools about 68% of blocks in one week | Top three about 72% in one week: AlphaPool 33.6%, Lazarus 25.3%, DxPool 13.5%, then 36 smaller tags | By degreecomparable, and not what the fork promised |
| Nodes | Large, but one implementation sets policy | Knots only, strict and opinionated policy by design. Fewer nodes, aligned operators | By degree |
| Developers | Core maintainers, corporate-funded contributors | A handful of volunteer teams, most visibly Knots, privkeyio, paulscode, Kilombino and Retropex. See Builders | By degree |
| Block construction | Pools build nearly all blocks | DATUM self-templating is offered by CONVOY and Lazarus; the largest pool, AlphaPool, is closed source and builds templates itself | By degree |
On 2026-09-03 the largest pool held about 18% and the story was "no dominant pool". Three weeks later one pool holds a third of blocks, despite advertising a 30% self-cap. Mining decentralization on this chain is an intent and a toolset, not a result, and it should be judged by the pool table, not the pitch.
Nodes said yes, miners did not #
The entrenched mining industry did not respond to the BIP-110 user-activated soft fork at all. Miner signalling peaked at about 2.53%, far below the 55% threshold. That non-response is exactly what a proof-of-work change is designed to route around.
| Metric | Whole network | Dissenting side | Verdict |
|---|---|---|---|
| Public nodes (coin.dance) | 26,085 on 2026-09-03, 25,419 on 2026-09-26 | Bitcoin Knots: 4,644 (17.8%) then, 4,289 (16.9%) now | Differs |
| Miner signalling | About 97.5% stayed silent | About 2.53% signalled | Differs |
| ASIC vendors | Bitmain, MicroBT, Canaan: Chinese firms hold about 97% of the market | Goldshell and iBeLink, plus older Antminer A3 units | Differs |
| Silicon compatibility | SHA256d-only chips | BLAKE2b-only chips. Zero cross-compatibility either way | Differs |
About 18% of nodes ran the dissenting client, but only about 2.53% of miners would signal. Users showed up; miners did not. So the users changed the lock. A hash swap strands the entire SHA-256 installed base, which is also why that industry had no reason to follow. Vendor overlap is not literally zero (Bitmain made the A3), but the incumbent fleet does not carry over.
Who picks the transactions #
Censorship power lives in template construction, not in hashing. Whoever builds the block template chooses the transactions. A "dumb" hasher pointed at a classic pool is actually more censorable, because under Stratum V1 it blindly processes whatever template the pool hands it. The fix is self-templating: returning transaction selection to the miner's own node, through Stratum V2 job negotiation or DATUM.
| Property | Bitcoin (SHA-256) | Bitcoin Blake2b | Verdict |
|---|---|---|---|
| Who selects transactions | Mostly four pools. Foundry 26.78%, AntPool 17.79%, F2Pool 15.02%, ViaBTC 8.20% (mempool.space, 2026-09-03) | Split. Miners on DATUM pools build their own templates; the largest pool builds them centrally | By degree |
| Coercion target | A few nameable, pressurable pool operators | Also a few pools, but DATUM miners can leave for their own node at any time | By degree |
| Tools that provide it | Stratum V2 and DATUM exist but are under-adopted | DATUM gateways built for BLAKE2b from launch, several independent implementations | By degreesame idea, better default |
| Miner-side manipulation | ASICBoost present | ASICBoost removed | Differs |
| Running a node to mine | Optional: join a big pool and run nothing | Also optional now. Classic pools exist, so the early "you must run a node" state is over | By degree |
CONVOY (convoy.xyz), the pool announced in early September, is running. It is DATUM-based: miners run their own DATUM Gateway against their own Knots node and build their own templates, with a classic Stratum V1 endpoint documented as a failover only. Fees are 1% through your own gateway and 2% on the failover, paid out non-custodially from the coinbase under TIDES. It mined 34 of 1,263 blocks in the week to 2026-09-26, about 2.7%.
Dumb hardware alone does not prevent censorship. Self-templating plus no dominant pool does. Bitcoin Blake2b ships the first half by default; the second half is already slipping. Its governance edge today is that leaving a pool for your own template costs almost nothing, not that pools do not matter.
Blockspace
The cleanest ideological split: is Bitcoin a settlement and data layer, or sound money only?
What the chain is for #
| Property | Bitcoin (SHA-256) | Bitcoin Blake2b | Verdict |
|---|---|---|---|
| Non-financial data | Permitted and widespread since 2023: Ordinals, inscriptions, Runes, BRC-20 | Restricted by RDTS, which targets exactly those | Differsmoney, not JPEGs |
| Block size | 4,000,000 weight units. Validating the 2023 to 2025 data-heavy blocks is visibly slow on a resync | 800,000 weight units, typically about 300 KB, until 2027-09-01 | Differs |
| Mempool policy | Core default is permissive | Knots: strict data-carrier and spam filters | Differs |
| Fee market | Congestion and speculation on data | Intended monetary priority | By degree |
| Philosophy | Settlement plus data layer | Sound money only | Differs |
The RDTS rules #
RDTS is the Reduced Data Temporary Softfork, BIP-110, authored by Dathon Ohm in December 2025 under a BSD-3-Clause licence. It activated on the minority chain at block 961,632 in August 2026 and sets consensus limits until 2027-09-01, when they expire automatically:
- Output scripts of at most 34 bytes, except OP_RETURN at up to 83 bytes
- Data pushes and witness script arguments of at most 256 bytes
- Only witness v0 and Taproot during the deployment, with annexes and control blocks restricted
- A block weight limit of 800,000, from block 961,640
- Coins that existed before activation are permanently exempt, so there is no deadline to move old funds
- Explicit targets: Ordinals inscriptions, BRC-20, Runes and oversized OP_RETURN payloads
All of it is temporary. The data limits and the smaller block weight share the same expiry, 2027-09-01, unless a later change extends them. "Clean blockspace" is a one-year measure today, not a constitutional property. An earlier version of this page wrongly called the block-size cut the durable lever.
It failed on the majority chain, with about 2.53% miner support, and exists only on the minority fork. So the accurate framing is majority Bitcoin without RDTS against the minority who forked to get RDTS and BLAKE2b.
Sovereignty
Philosophically both chains are Bitcoin. The tension is that Bitcoin Blake2b claims to be the uncaptured embodiment of the original ethos, while by that ethos's own logic the majority chain is being absorbed by the institutional world it was built to escape.
Capture and censorship #
| Property | Bitcoin (SHA-256) | Bitcoin Blake2b | Verdict |
|---|---|---|---|
| Self-custody | Yes | Yes, inherited | Identical |
| Institutional capture | High and rising: ETFs, corporate treasuries, custodial concentration, and the two largest holders now help fund developers | No ETF, no institutional surface, no consortium. Also no funding | Differs |
| Regulatory attack surface | Large: KYC exchanges and ETF custodians are chokepoints | Minimal today, illiquid and off the radar | Differsfreedom, but few exits |
| Transport-level censorship | Clearnet exposed by default | Tor and I2P first by culture | By degree |
| Whose chain is it | Increasingly Wall Street's | Cypherpunk and individual | Differs |
The duplicate coins: what happens to institutional BTC at the fork? #
Every bitcoin that existed at block 961,631 also exists as BTCB2 at the same address, under the same key. That includes the coins of the largest holders. Strategy reported 846,000 BTC on 2026-09-20; BlackRock's IBIT held 734,261 BTC on 2026-06-30. Together that is about 1.58M fork coins, roughly half a billion dollars on paper at $311 each. So who gets them?
| Holder | What its filings say about forks | Who ends up with the fork coins |
|---|---|---|
| BlackRock IBIT | "The Sponsor will cause the Trust to irrevocably abandon the Incidental Rights and any IR Digital Asset". The custodian has "no authority ... to exercise, obtain or hold" them for the trust | Nobody, on paper. Shareholders "will not receive the benefits" |
| Fidelity FBTC | Abandons fork assets. If the trust receives one anyway, the sponsor must sell it and distribute the proceeds. No obligation to claim if that exposes its bitcoin to risk | Shareholders, only in the involuntary case |
| Grayscale GBTC | Abandons before every creation and redemption, and indemnifies its custodian for abandoned assets | Nobody, on paper |
| 21Shares ARKB | Abandons, though other passages let the sponsor claim and pass on proceeds | Unclear by its own text |
| Bitwise BITB | Disclaims fork assets. Anything received involuntarily goes to the sponsor and becomes its property | The sponsor, who may sell and pass cash on |
| Strategy (MSTR) | No fork policy at all. Forks appear only as a risk factor. Custodians: Anchorage, Coinbase Custody, Fidelity Digital Assets | Strategy could claim in principle, but its custodians have no duty to support the fork |
| Coinbase custody terms | Supports forks only at its "sole discretion" and is "not configured to detect, process, or secure" unsupported forked assets | No public statement on BTCB2 |
| BitGo policy | Passes fork value on only if the asset has over $100M market cap and $25M daily liquidity for 14 days | BTCB2 fails the liquidity test by orders of magnitude |
Coinbase holds the keys
Behind most of these ETFs sits one company. Coinbase Custody is the custodian, or one of the custodians, for IBIT, GBTC, the Grayscale mini trust, BITB, ARKB and several smaller funds; only Fidelity's FBTC keeps its bitcoin entirely elsewhere. In 2024 Coinbase said it held about 90% of all spot ETF bitcoin. It is also one of Strategy's three custodians, next to Anchorage and Fidelity Digital Assets, and it holds the coins of millions of retail customers. Whoever holds the keys is the only party that can technically reach the Blake2b coins. For all of these coins, that is Coinbase.
| Question | What we know, as of 2026-09-26 |
|---|---|
| Has Coinbase said anything about BTCB2? | No. No blog post, help page, listing or executive statement. At the split its status page simply reported Bitcoin systems as normal |
| Does it have to support the fork? | No. Its custody terms leave forks to its "sole discretion", and IBIT's filing says the custodian has "no authority" to hold abandoned fork coins for the trust |
| What does its own fork policy say? | Its 2018 policy promises that "customers should benefit to the greatest extent possible", but supports a fork only if it passes Coinbase's asset review and avoids risks including "sustained low liquidity" and security issues. BTCB2 fails the liquidity test clearly |
| What did it do before? | Bitcoin Cash (2017): credited to customers, trading from December 2017. Bitcoin SV (2018): withdrawal only, later liquidated. Bitcoin Gold (2017): never supported, and a California court ruled in 2020 that customers had no right to it |
| Where are the coins now? | Unknown. No public evidence that Coinbase has split, claimed, moved or sold any BTCB2. One analyst on X estimated that about 89% of the coins in Coinbase's published cbBTC reserve addresses had not moved since the split, so they could still be split unverified |
| Who owns them legally? | A grey zone. The ETFs have abandoned them, the custodian says it holds nothing for them, retail terms give customers no right to unsupported forks. The coins sit under keys that belong to one company, owned by nobody who claims them |
So the most likely outcome is that the Blake2b coins behind the ETFs simply stay where they are, untouched, unless Coinbase decides otherwise. There is one exception, and it is replay again. Every time Coinbase moves pre-fork coins on Bitcoin with an ordinary signature, the same transaction can be broadcast on Bitcoin Blake2b, and the fork coins follow the bitcoin to the receiver. Bitcoin University discussed how many Blake2b coins Coinbase holds and what it could do with them in Will Coinbase Steal Customers' Bitcoin?
Only people who hold their own keys received the new coins in a way they control. Everyone who holds bitcoin at an exchange, in an ETF or with a custodian depends on a decision someone else has not made. "Not your keys, not your coins" was never demonstrated more clearly than by a fork.
A signature that does not opt in to the unified sighash is valid on both chains. Knots' own help text says that signing the old way "gives up replay protection". So whenever a custodian moves pre-fork coins on Bitcoin with an ordinary signature, and the same coins have not already moved on Bitcoin Blake2b, anyone can rebroadcast that exact transaction on Bitcoin Blake2b. The BTCB2 then lands at the same destination as the BTC: the authorized participant, the exchange, the buyer, or the custodian's own change address.
Replay of unprotected transactions is not theoretical: it is documented by the project and demonstrated step by step in public guides, such as this video walkthrough. Whether a given custodian transfer leaks depends on how it was signed, whether the coins already moved on Bitcoin Blake2b, and whether it fits the fork's temporary data and block-weight limits. But it means "abandoned" coins need not stay put. IBIT alone redeemed or sold 106,148 BTC in the second quarter of 2026, and every such transfer after the fork is a potential leak of the matching BTCB2 to whoever received the BTC.
The same trap applies to you. If you sell or move pre-fork bitcoin without splitting first, you may hand your BTCB2 to the recipient without knowing it. If you move BTCB2 with an old wallet or stock hardware firmware, you may move your BTC too. Split first, with a wallet that signs with the unified sighash, and check that the wallet shows the transaction as protected. How to split safely →
Precedent
- Bitcoin Cash, 2017. GBTC's then custodian claimed the BCH without first telling the trust. Grayscale distributed it to shareholders of record, $0.61 a share, and remitted about $211.5M in sale proceeds.
- SegWit2X, December 2017, the closest analogue. There was no principal market, so the trust valued the rights at zero, and its agent abandoned them because selling would cost more than it raised. The ETFs' abandonment clauses grew out of this.
- Archer v. Coinbase (2020). A California appeals court held that Coinbase's user agreement created no right to Bitcoin Gold and no duty to support the fork.
- US tax, Rev. Rul. 2019-24. Fork coins are income only once the holder has "dominion and control". Coins sitting at a custodian that does not support the fork are not income until access is gained.
So who benefits?
- Not ETF shareholders. Four of the five largest spot ETFs have contractually abandoned fork coins, and Bitwise routes them to its sponsor.
- Possibly Coinbase, in practice rather than by right, because it holds the keys for most ETF coins and many customer coins, and no contract forces it to pass fork coins on. It has not said what it will do.
- Maybe Strategy, the one large holder without an abandonment clause, if it chooses to push its custodians. Nothing in its filings suggests it will, and its chairman has criticised BIP-110 publicly.
- Recipients of replayable transactions, possibly without knowing it, whenever pre-fork coins move without protection.
- Nobody at scale, through the market. With tens of dollars of resting orders near the best price, even 1% of these coins could not be sold at any quoted price. Paper value is not realizable value.
Institutions have pre-committed, in writing, to treat SHA-256 Bitcoin as bitcoin and to walk away from any fork. That is the Schelling point in legal form. For Bitcoin Blake2b it cuts both ways: the fork gets no institutional buyers, and it also carries no institutional overhang of coins waiting to be dumped, because those coins cannot reach a market. The first formal sponsor decisions on this fork, if any, should appear in the third-quarter filings due around November 2026.
Ethos and philosophy #
| Axis | Bitcoin (SHA-256) | Bitcoin Blake2b | Verdict |
|---|---|---|---|
| Founding myth | Satoshi, the 2008 white paper, a fair launch | A re-forging of that ethos after the original was captured | Differs |
| Exit or voice (Hirschman) | Voice: reform from within, and being out-lobbied | Exit: fork when reform fails | Differsthe cleanest split |
| Freedom and anti-authoritarianism | Foundational, now compromised by KYC and ETF rails | Reasserted | By degree |
| Self-custody as a right | Yes | Yes | Identical |
| Sound-money ethics | Yes | Yes | Identical |
| Low time preference | Yes | Yes | Identical |
| Antifragility | Yes | Literally: it survives as a minority chain | By degree |
| Energy | Industrial, cheap-power-seeking mining | Hobbyist and small-farm mining on second-hand Sia-class hardware | Differs |
| Maximalism | "There is no second best" | Claims to be the true continuation | Differscontested legitimacy |
| Emotional pitch | Digital gold, number go up | Reclaim the protocol's soul | Differs |
| Risk of the ideology | Complacency and capture | Tribalism and irrelevance | Differs |
| A way of life, not just money | Yes | Yes | Identical |
Who funds the developers #
The Bitcoin Security Consortium launched on 23 July 2026 with nine founding members: Anchorage Digital, ARK Invest, BlackRock, Block, Blockstream, Coinbase, Fidelity Digital Assets, Galaxy and Strategy. They pledged $15M over three years, with no pooled fund: each member funds developers and researchers it chooses. Its stated first priority is post-quantum cryptography, and it explicitly renounces any role in protocol governance. Nothing had changed by 2026-09-26.
The holdings context explains the concern. Strategy reported 846,000 BTC on 2026-09-20 and BlackRock's IBIT held 734,261 BTC on 2026-06-30, together about 1.58M BTC or 7.5% of the cap.
| Property | Bitcoin (SHA-256) | Bitcoin Blake2b | Verdict |
|---|---|---|---|
| Largest holders near developer funding | Yes, the two largest holders fund developers through the Consortium | None: no ETF, no treasuries, no consortium | Differs |
| Developer funding model | Institutional: credible, with capture optics | Unfunded and volunteer: uncaptured, and fragile | Differs |
| Product-incentive drift | High, ETF and BTC-backed credit products profit from wrapped holdings | None, there are no products | Differs |
| Pressure on quantum migration | Plausible through a funded roadmap | Same exposure, no institution steering the roadmap | By degree |
The legitimate concern: influence follows money, and the product incentives of ETF and credit issuers pull away from self-custody. It is fair to watch that the 21M cap and the freedom to spend stay sacrosanct.
Not a fact: claims that they plan to raise the cap. There is no evidence of any such proposal, and the Consortium disclaims governance. Funding is decentralized and the quantum threat genuinely needs paid work. "Uncaptured but unfunded" is a far more persuasive tension than a conspiracy.
The quantum fault line #
BIP-360 (P2MR, originally P2QRH, by Hunter Beast, Ethan Heilman and Isabel Foxen Duke) was merged into the BIP repository as a draft on 2026-02-11 and is still a draft. It adds an opt-in output type spent with post-quantum ML-DSA signatures, under the new bc1z prefix. Coins stay vulnerable until moved, and full migration is estimated at about seven years.
BIP-361 ("Post Quantum Migration and Legacy Signature Sunset", Jameson Lopp and others, April 2026, still a draft) is the flashpoint. Three years after activation it would disable legacy signature verification at consensus level. Coins not migrated would be frozen forever. About 6.7M BTC sit in addresses with exposed public keys, including roughly 1.1M BTC believed to be Satoshi's. Critics call it confiscation; its defenders would rather freeze dormant coins than let them fall to quantum attackers.
The threat is credible. Early in 2026 Google Quantum AI estimated that a machine with fewer than 500,000 qubits might run the attack in about nine minutes, faster than a block. Exposed public keys can be harvested today and cracked later.
| Property | Bitcoin (SHA-256) | Bitcoin Blake2b | Verdict |
|---|---|---|---|
| Quantum vulnerability | Yes, secp256k1 | Yes, identical | IdenticalBLAKE2b does not fix quantum |
| Concrete post-quantum roadmap | BIP-360 and BIP-361, both drafts | None yet | DiffersBitcoin is further along |
| Who steers the fix | An institutionally funded roadmap | No institution, community only | Differs |
| Freezing old coins on the table | Yes, BIP-361, including Satoshi's | Not proposed | Differs |
| Freedom to spend any coin | At risk through the freeze debate | Preserved, for now | Differsthe sovereignty crux |
Bitcoin Blake2b will face the same dilemma eventually, since its cryptography is the same. The difference is who sits at the table when it is decided: a funded institutional roadmap, or nobody yet. A sovereign-minded holder may prefer a chain that treats quantum exposure as the holder's own responsibility over one that might vote to freeze dormant coins for their own good.
Custody and privacy
Nearly every cryptographic primitive is inherited: same keys, seed phrases, address types and scripts. The differences are tooling maturity and one real cross-chain trap, which now has a documented fix.
Keys and custody #
| Property | Bitcoin (SHA-256) | Bitcoin Blake2b | Verdict |
|---|---|---|---|
| Private and public keys | secp256k1 | Same | Identical |
| Seed phrases and HD derivation | BIP39, BIP32 | Same | Identical |
| Address formats | SegWit, Bech32, Taproot | Inherited | Identical |
| Multisig, descriptors, PSBT | Full | Inherited. Shrike has a real 2-of-3 across SeedSigner, Coldcard and Krux on record | Identical |
| Hardware-wallet support | Official, from every vendor | Community firmware forks with replay protection for Trezor, Coldcard, SeedSigner and Krux, all unaudited and showing unofficial-firmware warnings. Stock firmware still signs, but those signatures replay. Nothing for Ledger, Jade or BitBox02 | By degree |
| Cold storage and metal backups | Mature | The same techniques apply | Identical |
| Cross-chain replay | Not applicable | A live issue for plain signatures. The opt-in unified sighash prevents it | Differs |
| Miner firmware trust | Known vendors | Often second-hand Sia-era ASICs, so factory resets and firmware hygiene matter | By degree |
| Inheritance tooling | Mature ecosystem | Same primitives, less tooling | By degree |
Replay and coin splitting #
Every coin that existed before the fork exists on both chains under the same private key. A plain spend of such a coin can be replayed on the other chain, moving your coins on the chain you did not intend to touch. This is demonstrated, not hypothetical; see the step-by-step replay protection guide.
The fix is Knots' unified sighash (SIGHASH_UNIFIED, flag 0x20, so 0x21 for the usual SIGHASH_ALL). A signature that opts in is valid only on Bitcoin Blake2b. Protection is per signature and one-way: signatures that do not opt in stay valid on both chains. The Knots wallet opts in by default, as do Shrike, the paulscode Sparrow fork, the pyblock-watch Android wallet from 0.8.2, fortis, and the hardware firmware forks above.
The documented way to split:
- Send to yourself on Bitcoin Blake2b with a wallet that signs with the unified sighash, and wait for it to confirm. That transaction cannot be replayed on Bitcoin.
- Then spend on Bitcoin as you normally would. Your Bitcoin Blake2b coins have already moved to an address the other chain never saw them go to.
A plain signature from an older wallet or stock hardware firmware replays and moves the coins on both chains. ANYONECANPAY signatures are weaker still, because that one input can be lifted onto the other chain. Check that your wallet shows the transaction as protected, and test with a small amount first. Your keys now control two coins, and separating them safely is on you.
Privacy #
| Property | Bitcoin (SHA-256) | Bitcoin Blake2b | Verdict |
|---|---|---|---|
| Base-layer anonymity | Pseudonymous | Pseudonymous, same model | Identical |
| KYC exposure | KYC dominates the on-ramps | No KYC by absence: the only exchanges are no-KYC | Differs |
| Chain surveillance | Heavy, a whole analytics industry | None targeting it yet | By degreeobscurity is not privacy |
| CoinJoin and PayJoin | Available | None found. One Android wallet offers Ricochet and PayNym, which are privacy tools but not CoinJoin | Differs |
| Network-level privacy | Tor optional | Tor and I2P first by culture | By degree |
| Address-reuse hygiene | User dependent | The same discipline applies | Identical |
Ecosystem
Broader than its age suggests, thinner than it looks. A month after the fork there are wallets, servers, firmware and Lightning builds, but almost every piece is unofficial, unaudited, and from a small number of teams. Some thinness is deliberate (no ETFs, custodians or analytics is the point); some is simply early.
The stack around the chain #
| Layer | Bitcoin (SHA-256) | Bitcoin Blake2b | Verdict |
|---|---|---|---|
| Node software | Core, Knots, btcd and more | Bitcoin Knots 29.4.1 and later, one implementation | By degree |
| Mining gateway | Stratum V2, many pools, pool-built templates | DATUM Gateway forks for BLAKE2b (CONVOY, innerhat-dev, others) and a Rust rewrite, RATUM | Differs |
| Miner hardware | The SHA-256 ASIC industry | Sia-class BLAKE2b ASICs: Goldshell SC series, iBeLink BM-S, Antminer A3. Some models (HS-Lite, HS6) still pending confirmation | Differs |
| Desktop wallets | Every wallet | Two Sparrow-based wallets, Shrike (23 releases, latest 2026-09-23) and a paulscode Sparrow fork, plus the Knots wallet. No Electrum wallet, because Electrum checks 80-byte headers in the client | By degree |
| Mobile wallets | Many | Early Android wallets: pyblock-watch, PyBLØCK, btcblake2b wallet, plus a Rust wallet core with mobile bindings | By degree |
| Explorer and API | mempool.space and dozens more | mempool.guide with a live REST API, and several community explorers | By degree |
| Electrum servers | Fulcrum, Electrs, ElectrumX | Shulcrum and a privkeyio Fulcrum fork, plus two Electrs forks. Shulcrum is verified on testnet4; mainnet verification is in progress | By degree |
| Lightning and layer 2 | Mature, with LSPs | Core Lightning and LND forks with tagged prereleases, unaudited. Channels opened before the fork need care | By degree |
| Merchant payments | BTCPay, Lightning, processors | None found. BTCPay would need header support in its indexer | Differs |
| Self-hosted node boxes | Umbrel, StartOS and more, official packages | Community StartOS and Umbrel packages for Knots, DATUM, mempool, Electrum servers, Lightning and wallets. Start9 itself ships nothing for this chain | By degree |
| Developer libraries | BitcoinJS, BDK, rust-bitcoin and more | Forks of rust-bitcoin, embit and drongo with v2 headers and the unified sighash. Unforked block-level tools break on the 164-byte header | By degree |
| Third-party services | Custodians, ETFs, analytics: enormous | None, a feature and a bug | Differs |
| Audits | Decades of review | Almost every fork of wallet, firmware and Lightning code describes itself as not audited | Differs |
Durability #
| Property | Bitcoin (SHA-256) | Bitcoin Blake2b | Verdict |
|---|---|---|---|
| Track record | 16+ years of uptime | Weeks | Differs |
| Developer bus factor | Many | A handful of teams carry most of the stack | Differsa real fragility |
| Capture resistance over time | Declining as it institutionalizes | Untested, but designed for it | By degreethe whole bet |
| Survival mode | Too big to kill | Too small to notice, a kind of camouflage | Differs |
| Mining economics | Industrial, cheap-power-seeking | Small and mid-size, on repurposed Sia hardware | Differs |
Treat Bitcoin Blake2b as a small bet that pays large if its thesis is true: insurance against capture, not an investment. Bitcoin's scale is real; so is the question of what scale has cost it.
Who is building
Almost all Bitcoin Blake2b software is open source on GitHub and built by volunteers, most of it forks of well-known Bitcoin projects. These are the most active contributors as of 2026-09-26, checked against the repositories themselves, plus one voice building awareness rather than code. The community keeps a fuller directory at awesome-bitcoin-blake.
The main players #
Bitcoin Knots and Luke Dashjr
The reference node. BLAKE2b proof of work, the 164-byte header, flag-day RDTS and the unified sighash all ship in its releases. Luke Dashjr also leads the DATUM gateway line behind the CONVOY pool.
- bitcoinknots/bitcoin node, v29.4.2 (2026-09-21)
- CONVOYMining/datum_gateway BLAKE2b DATUM gateway
- luke-jr/datum_gateway upstream of that line
privkeyio
The broadest contributor. The Shrike wallet, signing firmware for three hardware families, the libraries underneath, a Fulcrum fork and a Core Lightning fork.
- privkeyio/shrike desktop wallet, Sparrow fork
- seedsigner, cc-firmware, trezor-firmware signers
- drongo, rust-bitcoin, embit libraries
- Fulcrum, lightning server, Core Lightning
paulscode
A second, independent wallet and server line, plus an LND fork and a full set of StartOS and Umbrel packages for running the chain at home.
- paulscode/sparrow desktop wallet
- electrs-pruned Electrum server
- lightning-fork LND
- knots-blake2b-startos, umbrel-store packaging
Retropex
The explorer codebase behind mempool.guide, with BLAKE2b headers, unified-sighash display and a replay badge, and the Knots packages for StartOS and Umbrel.
- Retropex/mempool explorer
- knots-startos, mempool-startos, datum-gateway-startos StartOS
- umbrel-bitcoin Umbrel
Kilombino
Started Shulcrum, the first Electrum server for 164-byte headers, and builds an Android watch wallet and a mempool fork with BIP-110 violation detection.
- Kilombino/Shulcrum Electrum server, testnet4 verified
- pyblock-watch Android wallet
- mempool-bip110 explorer
Mining builders
Independent gateway, pool and miner software, so that no single implementation decides how blocks get built.
- innerhat-dev/datum_gateway BLAKE2b gateway
- iohzrd/ratum gateway and pool in Rust
- jasonsopko/electrs Electrs with v2 headers
- Antisys/cpuminer-bip110 CPU miner
Bitcoin University awareness and analysis
Matthew Kratter's educational channel, building in a different sense: explaining the Blake2b fork to a wide audience, walking viewers through replay protection and self-hosting, and arguing a sharp critique of the SHA-256 ecosystem, from Core policy to Blockstream. It also asks hard questions of the fork itself, such as whether it is too centralized.
TwentyOne.Life that is us
We maintain Shulcrum going forward, with protocol 1.8, and package it and the Shrike wallet for StartOS, so a home server can index the chain and serve wallets over Tor. Stated plainly: Shulcrum is verified on testnet4, mainnet verification is in progress, and the packages have not yet served a wallet holding money end to end.
- TwentyOneLife profile and overview
- Shulcrum Electrum server
- shulcrum-startos StartOS package, v2.1.2-5
- shrike-startos Shrike for StartOS, v2.5.5-19
By layer #
| Layer | Projects | State |
|---|---|---|
| Consensus and spec | Bitcoin Knots; BIP-110 for RDTS. The BLAKE2b change itself has no BIP | Released, one implementation |
| Mining | CONVOY, innerhat-dev and other DATUM gateways, RATUM, CPU and CUDA miners | Active, several implementations |
| Electrum servers | Shulcrum (Kilombino, TwentyOne.Life), privkeyio Fulcrum, jasonsopko and paulscode Electrs | Released, mainnet verification ongoing |
| Wallets | Shrike, paulscode Sparrow, Knots wallet, Android wallets, coin-splitting tools | Released, unaudited |
| Hardware signers | privkeyio forks for SeedSigner, Coldcard, Trezor; connorslab/krux-blake2b | Released, unaudited, unofficial firmware |
| Lightning | privkeyio Core Lightning, paulscode LND | Prerelease |
| Explorers | mempool.guide (Retropex), Kilombino mempool-bip110, community explorers | Live |
| Self-hosting | StartOS and Umbrel packages from Retropex, paulscode, privkeyio, TwentyOne.Life | Sideload, community maintained |
About this page
Open questions #
What this research still could not confirm, as of 2026-09-26.
- A real node count. No crawler counts Bitcoin Blake2b nodes. The 1,100 figure is an upper bound from user-agent strings.
- Trading volume. Reported exchange volume does not match the order books. Treat it as possibly inflated.
- Pool concentration. Whether AlphaPool's share keeps rising, and whether DATUM pools gain share.
- Per-model miner support for Goldshell HS-Lite, HS6, HS-Box II and iBeLink BM-S1, listed by vendors as pending.
- Audits. None of the wallet, firmware or Lightning forks has been audited.
- Merchant payments. No BTCPay or processor support found.
- Neither chain is "ASIC-proof" or "quantum-resistant". Treat any claim that either is as wrong.
Contact and tips #
Corrections, sources and questions are welcome on Nostr. If this page was useful, a Lightning tip keeps the research going.
Nostr address and Lightning address. Lightning tips are paid in Bitcoin on the SHA-256 chain.
Support this work in BitcoinB2B #
Donations in BitcoinB2B, the coin of the Bitcoin Blake2b chain:
1BH665bXvEqSuoWQUihiQiPpt2BqpzrgGDBitcoin Blake2b shares Bitcoin's address format and its genesis block, so this is a perfectly valid Bitcoin address as well, and nothing about it says which chain it belongs to. Send from a Bitcoin Blake2b wallet. Bitcoin sent to it is a different asset and is not a donation to this project, whatever your wallet shows you. There is no way to make that visible in the address itself; it is a property of the fork, not an oversight.
Sources #
- mempool.guide API: chain tip, hashrate, difficulty, pool shares
- mempool.space mining pools: SHA-256 pool share
- coin.dance: public node counts by user agent
- bitcoin-blake2b.org: project site, miners, FAQ, developers (formerly btc-blake2b.org)
- Knots v29.4.2 release, source and doc/unified-sighash.md; PRs #357, #358, #359, #385, #391
- BIP-110, BIP-360, BIP-361
- convoy.xyz: pool status, DATUM setup, fees
- NonKYC.io and Neoxa Exchange public APIs, 2026-09-26
- CoinCodex, CoinCarp: aggregator prices (secondary)
- news.bitcoin.com, 2026-09-08: the XBT ticker controversy (secondary)
- Miner guides from millionminer.com, bt-miners.com, asicmarketplace.com (secondary)
- awesome-bitcoin-blake: community directory
- Strategy press release, 2026-07-23: the Bitcoin Security Consortium
- SEC filings: IBIT, FBTC, GBTC, ARKB and BITB 10-Ks for FY2025; IBIT 10-Q Q2 2026; Strategy 10-K, 10-Q and 8-Ks of September 2026
- Coinbase Prime Broker Agreement (SEC exhibit, 2024) and the BitGo fork policy
- Grayscale GBTC 10-K FY2019 (Bitcoin Cash, Bitcoin Gold, SegWit2X); Archer v. Coinbase (2020); IRS Rev. Rul. 2019-24
- Coinbase, "How Coinbase approaches forked and airdropped assets" (2018, archived copy); ETF custodian filings and reporting (Bloomberg, CoinDesk, The Block); Crypto Finance on custody at a chain split (2026-08-03)
- Google Quantum AI resource estimate, early 2026
- RFC 7693: the BLAKE2 hash function