The Role of Mining Pools in Bitcoin Decentralization

What Pools Actually Centralize — and What They Don't

Pool critics and Bitcoin maximalists often argue past each other because they measure different things. The useful frame is to split "who controls Bitcoin" into two layers: who owns the hashrate, and who coordinates the block template. Pools are light on the first and heavy on the second.

Hashrate ownership stays distributed. A pool like Foundry USA reports roughly 30% of the network's hashrate as of early 2026, but it does not own those machines. Hundreds of independent farms — public miners such as Marathon, Riot, CleanSpark, plus thousands of mid-size operators and home miners — simply point their ASICs at Foundry's stratum endpoint. If they unplug tomorrow, Foundry's share collapses the same day.

What a pool does centralize under the legacy Stratum V1 protocol is the block template: the specific set of transactions miners hash against. That single responsibility — deciding which transactions get included and whose coinbase reward address is written into the block — is where regulatory, commercial, and political pressure lands. Understanding the split is the first step to reading pool concentration charts honestly.

Pool Concentration in 2026 — A Snapshot

Looking at the numbers straight is the most honest way to talk about decentralization. As of Q1 2026, Hashrate Index pool data and CoinWarz hashrate charts tell a consistent story: four pools account for roughly 75% of blocks, and two of those four clear 25%+ each.

The uncomfortable number is the combined top two: ~52%. On paper, that is the theoretical threshold where two cooperating pools could attempt a 51% attack — double-spends, selective censorship, or chain reorganizations. In March 2026 a rare 2-block reorganization briefly spotlighted this risk, not because anyone attacked the chain but because concentration made the statistical coincidence noticeable.

Context matters, though. Pools are not monolithic actors — they are aggregators of independent miners. If Foundry announced tomorrow that it would censor certain transactions, large miners would have strong financial and reputational incentives to leave. That exit option is the core mechanism keeping the system honest, and it only works if miners actually exercise it. We'll come back to that.

Transaction Selection — Where Centralization Bites

Every ~10 minutes a pool that wins a block decides which transactions went into that block. Historically this was a dry technical detail because pools optimized for one thing — fee revenue. Pick the transactions paying the highest sat/vByte, stuff the block, collect the fees, move on.

Then regulatory pressure arrived. In May 2021, Marathon briefly operated an "OFAC-compliant" pool that filtered transactions linked to US Treasury SDN-list addresses using third-party wallet-scoring software. The experiment became public when one of those filtered blocks included just 178 transactions and earned roughly $2,903 in fees — while adjacent blocks on the same day cleared 1,100+ transactions and over $17,000. Community backlash was immediate and Marathon reversed course within weeks. CoinDesk's reporting on the Marathon OFAC blocks is still the cleanest primary source.

The episode illustrated three things at once. First, pool-level censorship is technically easy — it's just a filter in the template builder. Second, it's economically expensive: a censoring pool leaves fee revenue on the table, and over time miners reroute to pools that don't. Third, the fact that it became a headline at all is the feature, not a bug. Transparent block templates make censorship visible, and visibility is what gives miners the information to re-point.

Stratum V2 — Returning Template Power to Miners

The long-term structural answer to "what if pools misbehave?" is not to break up pools. It's to remove the pool's monopoly on block template construction. That is exactly what the Stratum V2 protocol specification does through its Job Declaration Protocol and Template Provider role.

Under legacy Stratum V1, the pool ships a template to your miner and your only choice is to hash it or disconnect. Under Stratum V2 with Job Negotiation enabled, the flow splits:

Adoption in early 2026 is still thin but no longer theoretical. Braiins Pool shipped full Stratum V2 support including Job Negotiation in February 2026, and the DEMAND (DMND) pool launched as the first pool designed entirely around V2. That is ~5% of network hashrate in the early-adopter camp — small, but the trend direction matters more than the current slice. Every percentage point of hashrate that migrates to miner-chosen templates is a percentage point where pool-level censorship becomes impossible by design.

Re-pointing — The Exit Valve That Keeps Pools Honest

Every discussion of pool concentration eventually asks the same question: if Foundry crossed a line, how long would it take miners to leave? The answer is technically seconds per machine, hours per fleet. A pool switch is a single config change — the stratum URL and worker credentials — applied across the farm. The ASICs hash identically; only the destination pipe changes.

This has played out in public before. When Marathon's OFAC experiment drew criticism in 2021, miners pointed at the pool were free to leave and competitors openly recruited them. More recently, when an individual pool has been suspected of template manipulation, hashrate visibly migrated within 24–48 hours — you can see it in any 7-day pool-share chart with timestamps.

Where Cloud Miners Fit In This Picture

Cloud-mining platforms sit on the miner side of the pool relationship, not the pool side. At NoxHash we operate our own fleet of Antminer S21 series and WhatsMiner M60/M66 units at 12–18.5 J/TH out of our Kazakhstan data center, and we point that hashrate at major FPPS pools on behalf of our customers. The hashrate you rent is real TH/s sitting in a real rack — not a paper contract — and it contributes to whichever pool the operations team chooses.

That means two things for our users. First, you are part of the miner-side leverage described above: if a pool misbehaves, our operations rotate our fleet like anyone else's, and your rented hashrate follows. Second, we deliberately diversify across more than one pool and track Stratum V2 adoption closely, precisely because concentration risk is real and passive monitoring is not enough. The honest version of cloud mining is not "we outsource pool choice to you" — it's "we run operations professionally so pool-level politics never touch your payouts."

Frequently Asked Questions