When people worry about Bitcoin centralisation, they usually mean one of two things. Ownership centralisation - the fact that a relatively small number of wallets hold a disproportionate share of the supply. Or mining centralisation - the concentration of hash power in a handful of countries and publicly listed companies.

Both are valid concerns. But there's a third layer of centralisation that gets far less attention, and it's arguably the most consequential of all: mining pools.

Right now, about 15 pools coordinate nearly all of Bitcoin's block production. The top two control almost half. And unlike mining companies - which own hardware but still submit work through pools - it's the pool operators who decide what actually goes into each block. They construct the templates. They choose the transactions. They are, in a very real sense, the gatekeepers of the Bitcoin blockchain.

Why pools exist in the first place

In Bitcoin's early years, anyone with a laptop could mine a block. Those days are long gone. At today's network difficulty of over 148 trillion, a miner running a top-tier ASIC machine would wait an estimated 15 years to solo-mine a single block. The block reward is 3.125 BTC - worth roughly $200,000 at current prices - but the odds of ever seeing it are vanishingly small for an individual operator.

Pools solve this by combining thousands of miners' computing power, finding blocks more frequently, and distributing rewards proportionally. Think of it like a lottery syndicate - you win less per draw, but you win far more often. Over 95% of all Bitcoin blocks are now mined by pools. For most miners, joining a pool isn't a choice. It's the only viable path to revenue.

The pools charge fees for this coordination - typically 1% to 4% of mining revenue. In exchange, they handle the heavy lifting: running full nodes, constructing block templates, validating work, and distributing payouts. It's a sensible arrangement. The problem isn't that pools exist. It's how few of them there are, and how much power that gives them.

What's actually at stake

To understand why this matters, you need to understand what pools actually control. Every Bitcoin block starts as a template - a list of transactions the pool operator selects from the mempool to include. The individual miners contributing hash power don't choose the transactions. They just solve the puzzle. The pool decides what goes in.

That means pool operators have the ability to censor specific transactions by simply not including them in their templates. They can prioritise certain transactions over others. They can build empty blocks that process no transactions at all. And because a handful of pools produce the overwhelming majority of blocks, their decisions have network-wide consequences.

This isn't a theoretical concern. It's a demonstrated one.

Each incident is worth examining. In 2021, Marathon didn't just theorise about filtering transactions - they built an explicitly "OFAC-compliant" pool that excluded transactions from addresses on the US Treasury's sanctions list. The result was blocks that were measurably smaller and less profitable than those mined by other pools. They were deliberately degrading the service to comply with government requirements that don't actually apply to miners.

The F2Pool incident in 2023 was arguably more troubling, because it was happening silently. Nobody knew until an independent researcher's monitoring tool caught the discrepancy. F2Pool was the third-largest pool at the time, responsible for roughly one in every seven blocks. If they'd continued filtering without detection, 14% of Bitcoin's block production would have been subject to US sanctions compliance - applied unilaterally by a Chinese-founded pool operator.

Follow the money

The financial stakes make this concentration even more significant. At current prices, the Bitcoin network generates roughly $33 million per day in miner revenue - about $12 billion annually. All of it flows through these pools on its way to the miners who earned it. The pools themselves take a 1-4% cut - still a healthy $120-480 million per year - but the real power isn't in what they earn. It's in what they control.

Foundry USA is backed by Digital Currency Group - Barry Silbert's empire that also controls Grayscale. AntPool is owned by Bitmain, the company that manufactures an estimated 85% of all Bitcoin mining chips. That's a structural conflict worth noting: the company that makes the hardware also runs the pool that coordinates how it's used, and bundles pool access into hardware sales contracts.

Both Foundry and AntPool require KYC from their miners. That means the identity of miners contributing to roughly half of Bitcoin's hashrate is known to two corporate entities. If a government wanted to apply regulatory pressure to Bitcoin's block production, these are the two doors they'd knock on first.

The escape routes

The picture isn't all grim. There are meaningful efforts to reduce pool-level centralisation, and they're worth understanding.

Stratum V2 is a new mining protocol that lets individual miners construct their own block templates instead of accepting whatever the pool operator builds. It effectively strips pools of their transaction selection power while still allowing them to coordinate work and distribute rewards. Both Bitmain and Foundry have signalled support for it.

OCEAN is a non-custodial mining pool built explicitly around decentralisation principles. It supports Stratum V2 and transparent block template construction. It's small - under 1% of hashrate - but it represents the philosophical counterweight to the Foundry/AntPool duopoly.

MARA Pool is an interesting case in the opposite direction. Marathon Digital runs roughly 3.6% of hashrate through its own vertically integrated pool - mining their own Bitcoin without relying on a third party. Ironically, it was Marathon's 2021 OFAC-compliant pool that first demonstrated the censorship risk. They've since abandoned that approach, but the vertical integration model itself shows that large miners can remove the pool dependency entirely.

And then there's the most powerful defence of all: miners can switch pools in minutes. Unlike mining hardware (which takes months to procure and deploy) or geographical location (which involves infrastructure, permits, and power contracts), changing pools is as simple as updating a configuration file. When GHash.io crossed 50% in 2014, miners left voluntarily. When F2Pool was caught filtering, the community's response was swift enough to force a reversal within hours. The exit barrier is low - and that matters.

Bitcoin's security model was designed to ensure no single entity controls the network. At the protocol level - proof of work, difficulty adjustments, the 21 million supply cap - it succeeds. But at the pool level, the system relies on something less elegant: the assumption that a small number of operators will continue to play fair, and that the community will notice and react if they don't.

So far, that assumption has held. Three times it was tested. Three times the community caught it, and three times social pressure forced a reversal. The question isn't whether pools can censor transactions. We know they can - they already have. The question is whether community vigilance will be enough when the pressure to comply comes not from a co-founder's personal views, but from a government subpoena.