Bitcoin divided into two separate chains on 8 August 2026, triggered by the soft fork proposal BIP-110. The minority chain ground to a halt after producing just two blocks, while the main chain continued operating without interruption. The episode echoes earlier divisions over what Bitcoin's block space should carry, but stands apart in producing a chain that stalled almost immediately.

BIP-110 proposes to cap the volume of non-financial data in Bitcoin transactions for a period of one year, targeting content such as Ordinals inscriptions — a protocol introduced in 2023 that enables arbitrary data, including images and text, to be inscribed onto individual satoshis. The soft fork would limit data fields to 256 bytes and cap OP_RETURN outputs at 83 bytes, a carrier historically used to attach small amounts of metadata to transactions. To activate without a chain split, the proposal requires a quorum of 55% of blocks — specifically 1,109 out of 2,016 — within a two-week signalling window running from block 961,632 to block 963,647.

The origins of the dispute trace back to 2025, when Bitcoin Core removed the existing relay limit for OP_RETURN in version 30. Relay policies govern which transactions nodes forward to peers, as distinct from consensus rules that define which blocks are valid — a distinction at the centre of this debate. That decision led to the creation of BIP-444 and subsequently BIP-110. Strategy chairman Michael Saylor publicly opposed the proposal in a 110-point essay published in July 2026, arguing that consensus rules should address proven security risks rather than the perceived purpose of a transaction.

In the final signalling period before the split, only 51 blocks — 2.53% of the total — voted in favour of BIP-110.

AntPool Block Without Signal Triggers the Split

The two-week signalling window opened at block 961,632. From that height onward, nodes enforcing BIP-110 rules rejected any block that did not set the designated signalling bit — a marker in the block header that miners use to indicate support for a proposal.

The first block lacking the marker was produced by the mining pool AntPool. The main chain accepted it. BIP-110 nodes, however, discarded it and instead followed an alternative block from the mining group Roughnecks. From that point, two Bitcoin chains ran in parallel, with their shared history ending at block 961,631. Unlike the 2017 Bitcoin Cash hard fork over block size, which produced a competing chain with sustained economic activity, this division stemmed from a minority of nodes enforcing a soft fork rule without sufficient miner backing.

Roughnecks, which mines through the Ocean pool, emerged at the split under the name "110th Mining Division." A second block followed on the minority chain, bringing it to block 961,633, before production ceased entirely. The main chain, by contrast, continued producing blocks without interruption. Users on the main chain experienced no changes. Anyone operating a node with BIP-110 rules, however, now sees a different chain from the rest of the network.

A standard activation would have required substantial miner backing. BIP-110 sets the threshold at 1,109 of 2,016 blocks, or 55%, within the window from block 961,632 to 963,647. Only 51 blocks signalled during the period preceding the split, amounting to 2.53%. Since a soft fork becomes binding on Bitcoin only when sufficient miners enforce it, support fell far short of the requirement well before the split occurred. Proponents proceeded with the separation regardless.

Miner Boycott Freezes the BIP-110 Minority Chain

The minority chain's block deficit widened steadily. On the evening of 8 August, it stood at seven blocks. By the morning of 9 August, it had grown through 18 and 26 to a cumulative total of 57 blocks. Later that same day it reached 88, and at 15:27 UTC it hit 111 blocks. The main chain had advanced to block 961,744 by then, while the BIP-110 chain remained at 961,633. Measured against a ten-minute block interval, the lag amounted to roughly 18 hours. No newer data point had emerged by press time.

The standstill was not the result of an external attack. Miners withdrew voluntarily. In the first signalling period after the split, not a single block out of 113 signalled for BIP-110 — a drop from the previous level of at least 2.53%. Simultaneously, the hashrate attributed to Ocean on the BIP-110 chain collapsed from approximately 36 EH/s to roughly 1.25 EH/s, a 96.5% decline within a single day.

Roughnecks halted its mining operation on 9 August at 03:40 UTC and called on other BIP-110 miners to pause as well.

"We don't see this as a defeat for the BIP-110 movement, but as an escalation to the next step."
— Roughnecks, operator of the BIP-110 mining division

Established mining operations, by contrast, remained unified on the main chain. Foundry, F2Pool, AntPool, ViaBTC, and MARA all continued mining there throughout the episode. No formal boycott was announced; it consisted entirely of the large pools remaining on the original chain. An attack on the minority chain was therefore unnecessary, as it simply lacked sufficient computing power.

Exchanges and Custodians Ignore the Minority Chain

Infrastructure ultimately determines the economic relevance of a forked chain. A viable fork requires functioning wallets, sustained miner support, exchange listings, custody integration, and buyers. The BIP-110 chain currently lacks all of these components. A chain without such connections remains technically alive but economically irrelevant.

Neither Coinbase, Kraken, Binance, BitGo, Fireblocks, nor Anchorage publicly announced support for the minority chain. Coinbase and Kraken instead reported normal operations on the main chain through their status feeds. No major provider listed a separate token for the minority chain, meaning a tradable counterpart to the main chain's BTC never materialized. Anyone holding Bitcoin through an exchange or custodian remained on the main chain throughout, and institutional holders faced no need to take action. Michael Saylor estimated the share of hashpower remaining on the main chain at approximately 99.85%.

The market response was muted. Bitcoin traded at around USD 65,000 on the day of the split, and the fork itself produced no visible price movement. In the days that followed, the price held roughly at that level. By comparison, the last major soft fork — Taproot — activated in November 2021 at block 709,632 with broad miner support and no lasting minority chain.

Data Limits at the Heart of the Dispute

At its core, the conflict centres on what data belongs in a Bitcoin block. BIP-110 proponents seek to curtail non-financial content such as Ordinals inscriptions for one year through a consensus rule. The catalyst, however, was a change to Bitcoin Core's relay rules rather than a consensus rule. Proponents argue that the loosening of relay limits invited additional spam onto the blockchain. Opponents countered that it merely reflected the actual behaviour of miners.

The dispute reflects a recurring tension in Bitcoin's history between those who view the blockchain primarily as a settlement layer for financial transactions and those who see its censorship resistance as valuable for any data users are willing to pay to include. Previous flashpoints include the block-size debates of 2015–2017 and earlier disagreements over OP_RETURN capacity.

The regular activation path remains out of reach for the minority chain. It would need to reach block 963,648 — roughly 2,000 blocks ahead of its current position — having produced only two blocks so far. Under the proposal, the new rules would take effect from block 965,664 and remain in force for 52,416 blocks, approximately one year. At the current pace, that threshold is unattainable.

Proponents are therefore exploring a more radical alternative. Luke Dashjr, who maintains the Bitcoin implementation Knots, together with the pseudonymous developer Dathon Ohm, has proposed switching the minority chain's proof-of-work algorithm. The named candidates include RandomX, KT256, BLAKE3, and Scrypt. Such a switch would decouple the chain from existing SHA-256d ASICs, rendering the silent boycott by large mining pools irrelevant. The chain would, however, need to build its own hardware ecosystem from scratch. For now, BIP-110 remains a proposal without a chain capable of enforcing it.