Bitcoin Mining Pool Centralization: What It Means for You

Bitcoin’s blockchain has been reorganized three times in the past month, and the pools behind two of those events now control nearly 45% of the network’s hashing power between them. None of the reorgs were deep, and none involved a double-spend. But the pattern has put a spotlight on something Bitcoin holders rarely think about: just how concentrated mining has become, and what that concentration could mean if the trend keeps going.

Here’s the short version. On September 11, 2026, at block height 966,500, two mining pools — AntPool and SpiderPool — each found a valid block at almost the same moment. Only one could survive. AntPool’s version won, and SpiderPool’s competing block was discarded, along with roughly 3.1389 BTC (about $240,000) in rewards that pool had briefly believed it earned. It was the third one-block reorg in four weeks, following similar events on August 16 and August 24. Galaxy Research, which tracks these incidents closely, called it routine. Structurally, though, it’s happening against a backdrop where Foundry USA and AntPool alone now command close to 45% of Bitcoin’s hashrate, and the top four pools combined control something close to 70%.

That’s not a 51% attack. Nobody is rewriting Bitcoin’s history or double-spending coins. But it’s worth understanding why this is happening now, what it would take to become a real problem, and what — if anything — it should change about how you think about the network you’re holding.

What Actually Happened: Three Reorgs in Four Weeks

A blockchain reorganization, or “reorg,” happens when two miners solve a valid block at nearly the same time. For a few minutes, the network is split — some nodes see one version of the latest block, some see the other. Bitcoin resolves this automatically: whichever chain accumulates more proof-of-work first becomes the accepted history, and the losing block gets orphaned. Any transactions that were only in the orphaned block go back into the mempool to be mined again.

This isn’t new or inherently alarming. Small, one-block reorgs have happened periodically throughout Bitcoin’s history — they’re a natural consequence of a global network where blocks propagate at slightly different speeds to different miners. What made this stretch notable is the frequency. Three in a month is unusual enough that multiple outlets, including Bitcoin.com News and Phemex, covered the September 11 event within hours.

The Three Events

• August 16, 2026 — one-block reorg at height 962,722.
• August 24, 2026 — one-block reorg at height 963,853.
• September 11, 2026 — one-block reorg at height 966,500, AntPool over SpiderPool, roughly 3.1389 BTC in orphaned rewards.

None of these approached anything close to a security incident in the traditional sense. For comparison, Bitcoin SV suffered a 100-plus block reorg in 2021 and Monero has dealt with an 18-block reorg tied to a hashrate attack. A one-block reorg resolves in minutes and, for anyone waiting for the standard six confirmations before treating a transaction as final, changes nothing. Still, three in a month is enough of a cluster that it’s worth asking why it’s happening now.

Why Mining Concentration Is Climbing in 2026

Part of the answer has nothing to do with mining pools directly — it has to do with where the hashrate went. Bitcoin’s network hashrate peaked near 1,151 EH/s in late October 2025. As of late September 2026, it sits in the 990–995 EH/s range, roughly 14–20% below that high, and the network has now spent well over 300 consecutive days without reclaiming its all-time peak. That’s the longest stretch of its kind in close to a decade.

A meaningful chunk of that missing hashrate didn’t disappear — it moved. Publicly traded miner IREN Limited disclosed cutting its Bitcoin mining capacity from roughly 50 EH/s to about 23.2 EH/s across 380 megawatts of infrastructure, redirecting that power toward AI cloud computing instead. The company has reported a contracted AI revenue run-rate near $4 billion annualized for 2026 — numbers that dwarf what the same power draw could earn mining blocks at current difficulty. IREN isn’t the only operator making this calculation, and every megawatt that leaves mining for AI compute is a megawatt that isn’t contributing to network security.

Fewer total participants competing for blocks means the remaining pools, almost by definition, end up with a larger share of what’s left. As of September 18, 2026, the breakdown looked roughly like this:

• Foundry USA — approximately 25.5% of network hashrate
• AntPool — approximately 19.2%
• F2Pool — approximately 14.5%
• ViaBTC — approximately 10.6%

Foundry USA and AntPool together sit close to 44.6% — not a majority, but close enough that if the trend continued, crossing 50% combined wouldn’t require a dramatic shift.

The Part That Deserves More Scrutiny: Proxy Pools

Here’s where it gets more interesting than a simple pie chart. Pool share numbers assume each named pool is an independent operator competing for blocks on its own infrastructure. Research into AntPool’s structure has found that isn’t quite accurate. Several pools that present themselves as independent operations are, according to that research, actually relying on AntPool-constructed block templates behind the scenes — meaning AntPool effectively decides what those “independent” pools’ blocks contain, even though the hashrate is nominally attributed elsewhere.

When that hidden relationship is accounted for, some estimates put AntPool’s effective network influence as high as 37–50% at various points in 2026 — a very different picture than the 19.2% headline figure suggests. This is the kind of concentration that’s genuinely harder to see from the outside, because it doesn’t show up cleanly in a hashrate distribution chart. It shows up in who actually controls transaction ordering and censorship decisions, which is arguably the metric that matters more for a proof-of-work network’s neutrality.

To be clear, none of this means AntPool — or anyone — currently controls a hard majority of Bitcoin’s hashrate outright. It means the honest answer to “how decentralized is Bitcoin mining right now?” is more complicated than the top-line numbers imply, and that’s worth knowing whether you’re a long-term holder, a miner, or someone building on top of the network.

Should You Actually Be Worried?

Probably not urgently, and here’s the reasoning rather than just the reassurance. A genuine 51% attack requires an entity to control a majority of hashrate and sustain that control long enough to consistently out-mine the honest chain — not just win one lucky block race. Three one-block reorgs over four weeks are consistent with normal network variance and slightly elevated block propagation friction, not with any single actor deliberately rewriting history. Galaxy Research’s assessment — that this is the third such event in four weeks and broadly unremarkable — lines up with how these incidents have looked historically.

That said, dismissing pool concentration entirely misses the point. The risk isn’t that Foundry USA or AntPool wakes up tomorrow and attempts an attack; publicly traded, regulated mining operations have every commercial incentive not to torch trust in the asset generating their revenue. The more realistic concern is structural: fewer independent decision-makers means less resilience if any single pool suffers an outage, gets compromised, or faces regulatory pressure in whatever jurisdiction it operates in. It also means transaction censorship — a pool choosing not to include certain transactions — becomes more plausible as a smaller number of entities gain influence over what gets confirmed.

What This Means If You’re Holding or Mining BTC

For most holders, the practical takeaway is almost anticlimactic: keep waiting for standard confirmations before treating a large transaction as final, and don’t panic-sell over headlines about reorgs that resolved within minutes. If you’re moving a meaningful amount of BTC, six confirmations remains a reasonable, conservative standard — a one-block reorg simply doesn’t threaten a transaction that’s already several blocks deep.

For miners, the calculus is different and more immediate. Difficulty sat at 132.76 trillion following a 4.16% upward adjustment on September 19, 2026, with hashrate hovering around 995 EH/s. That’s a market where smaller, independent operators face a real decision: compete on razor-thin margins against pools with better economies of scale, join a larger pool and add to concentration, or — like IREN — redirect power toward AI compute contracts that pay more predictably. None of those are wrong choices individually. Collectively, they’re what’s driving the concentration trend in the first place.

For anyone building on Bitcoin or relying on its settlement guarantees, the AntPool proxy-pool research is probably the most useful piece of this story to actually remember. Headline hashrate distribution numbers are a starting point, not the full picture, and it’s worth checking in on this periodically rather than assuming decentralization is a fixed, permanent property of the network.

How This Compares to Past Concentration Scares

Bitcoin has been here before, more than once. Back in 2014, GHash.io briefly exceeded 51% of network hashrate and voluntarily capped its own share after community pressure — a moment often cited as evidence that social and economic incentives, not just code, help keep mining decentralized. Concentration concerns resurfaced periodically through the following decade as Chinese mining dominance rose and then fractured after China’s 2021 mining ban scattered hashrate across North America, Central Asia, and elsewhere.

What’s different in 2026 is the driver. Previous concentration cycles were mostly about geography and energy costs. This one is being shaped by a genuine alternative use for the same infrastructure — AI compute — that’s pulling capacity away from mining entirely rather than just consolidating it under fewer roofs. That’s a new variable, and it’s one worth watching over the next few difficulty cycles rather than drawing hard conclusions from a single month of data.

None of this is financial advice. Mining economics, pool dynamics, and network security are complex, fast-moving topics, and this article reflects publicly available data as of September 22, 2026. Do your own research before making investment decisions, and never treat a single data point — including this one — as the whole picture.

Frequently Asked Questions

What is a Bitcoin blockchain reorg?

A reorg happens when two miners find a valid block at nearly the same time, temporarily splitting the network into two competing versions of the latest history. Bitcoin resolves this by keeping the chain with the most accumulated proof-of-work and discarding the other block, along with any rewards attached to it. Transactions from the discarded block simply go back into the mempool to be confirmed again.

Is a one-block reorg dangerous?

Not on its own. A single-block reorg resolves within minutes and doesn’t threaten any transaction that already has several confirmations behind it. It becomes a genuine concern only when reorgs go many blocks deep, which is what separates routine network noise from an actual attack.

What percentage of hashrate is needed for a 51% attack?

An attacker generally needs to control a majority of the network’s hashing power and sustain that control long enough to consistently out-mine the honest chain. Foundry USA and AntPool’s combined share near 44.6% as of September 2026 is significant but still short of that threshold, and mining pools controlling hashrate is different from a single entity controlling it outright.

Why is Bitcoin’s hashrate still below its all-time peak?

Hashrate peaked near 1,151 EH/s in October 2025 and has stayed below that level for over 300 consecutive days as of September 2026. A meaningful driver is miners like IREN Limited redirecting power capacity from Bitcoin mining toward AI cloud computing contracts, which currently pay more predictably than mining at current difficulty.

What are AntPool’s proxy pools?

Research into pool infrastructure has found that several pools presenting themselves as independent operators actually rely on block templates constructed by AntPool, giving AntPool influence over those pools’ block contents beyond what its headline hashrate share suggests. Some estimates put AntPool’s effective network influence as high as 37–50% when this hidden relationship is factored in.

Should I change how I handle Bitcoin transactions because of this?

For most users, no. Waiting for the standard six confirmations on significant transactions already accounts for the kind of short reorgs seen in September 2026. This is more relevant to miners weighing which pool to join and to anyone tracking Bitcoin’s long-term decentralization trends than it is to everyday holders moving coins.

Sources

• Bitcoin.com News — “Bitcoin Miners Found 2 Valid Blocks, Then One Was Left Behind” (September 13, 2026)
• Phemex News — “Bitcoin Network Records One-Block Reorg at Height 966500” (September 13, 2026)
• Shattered.io — “3rd Bitcoin Reorg in a Month: Top 2 Pools Hold 44%” (September 18, 2026)
• CryptoTimes — “Bitcoin Hashrate Marks 316 Days Below Peak as Miners Redirect Power to AI” (September 3, 2026)
• CoinWarz — Bitcoin Difficulty and Hashrate Charts (live data, accessed September 22, 2026)
• Cryptolexicon — “Difficulty up 1.31%, hashprice up 22%: what actually changed for miners” (September 7, 2026)