How Bitcoin Resolved the AntPool–SpiderPool Reorg at Block 966,500

On September 11, 2026, Bitcoin briefly had two valid candidates for block height 966,500. SpiderPool and AntPool had independently mined blocks on the same parent, creating a temporary one-block fork. The race ended when the network built on AntPool’s version, making it the active-chain block and leaving SpiderPool’s otherwise valid block stale. 1820

What happened at height 966,500

Reports citing Galaxy Research say a monitoring node first received SpiderPool’s block, so that node temporarily treated SpiderPool’s branch as the chain tip. AntPool had independently produced a different valid block at the same height. Because blocks do not reach every node at the same instant, different parts of the network can briefly see different tips. 1835

The accepted AntPool block was reported with a timestamp of 11:30:30 UTC and 4,359 transactions. Fork-monitor data lists the competing SpiderPool block with a 11:35:19 UTC header timestamp. It also records SpiderPool’s block as first seen by its bcoin node at 11:30:42 UTC, while AntPool’s was first seen there later. That distinction matters: a block header timestamp is miner-supplied metadata, not a reliable measure of when every node received a block. 1740

The supplied reporting does not provide a comparably confirmed transaction count for SpiderPool’s stale block, so it should not be treated as established from these records.

Why AntPool’s branch won

Bitcoin nodes do not permanently choose a branch simply because it arrived first at one node. They follow the valid chain with the greatest accumulated proof of work.

At first, each pool had a one-block branch at height 966,500. A subsequent block at height 966,501 extended AntPool’s version. That made AntPool’s branch longer in accumulated work than SpiderPool’s one-block branch, and nodes converged on it as the canonical chain. SpiderPool’s block was therefore disconnected from the active chain and became stale. 2235

This is the normal resolution mechanism for an accidental fork race: miners ultimately direct their work toward the branch that has gained the proof-of-work advantage.

What happened to the blocks and rewards

Blockchain monitoring showed two versions of height 966,500: AntPool’s remained on the main chain, while SpiderPool’s was marked stale. They were not two permanent versions of Bitcoin’s ledger. 2240

AntPool’s accepted block was reported to have a total reward of 3.14112795 BTC. The prior research draft cited an approximately 3.1389 BTC reward for SpiderPool’s stale block, but the supplied primary-style excerpts do not independently substantiate that exact figure. What is clear is the outcome: a stale block is not part of the active chain, so its coinbase reward is not spendable as a confirmed main-chain reward. 22

Why the timestamps can look counterintuitive

The fork monitor’s data illustrates a common source of confusion. It lists AntPool’s block header time as earlier than SpiderPool’s, yet its observing node saw SpiderPool’s block first. 40

That is not a contradiction in Bitcoin consensus. Miners set block timestamps within protocol constraints, and block propagation takes time. A node’s local arrival order can differ from header-time order—and can differ from the order observed elsewhere on the network. What settles the competition is the accumulated work behind the branches, not either timestamp or one observer’s first-seen record.

The third reported one-block reorg in four weeks

Galaxy Research characterized the September 11 event as Bitcoin’s third reported one-block reorganization in four weeks, following events at heights 962,722 on August 16 and 963,853 on August 24. The common pattern was a short-lived competition between valid branches, followed by automatic convergence once one branch gained more accumulated work. 2635

A cluster of small reorgs does not by itself demonstrate a consensus flaw. One-block reorganizations are an expected consequence of proof-of-work mining: occasionally, separate miners find valid blocks close enough together that propagation delay briefly creates competing tips. 2325

What it meant for users

For ordinary users, a one-block reorganization is mainly a reminder that a first confirmation is not the same as finality. If a transaction appeared only in the stale block, its confirmation could disappear and the transaction could be rebroadcast or included in a later main-chain block. 2331

The September 11 incident was reported as a normal propagation-and-consensus outcome rather than a demonstrated double-spend, protocol bug, or attack. Merchants, exchanges, and other recipients handling meaningful value should continue to wait for confirmations appropriate to their risk, rather than relying on a transaction’s appearance in the latest single block. 2325