Mining Pools Explained: Who Controls Bitcoin’s Hashrate

Roughly every ten minutes, one machine somewhere on the Bitcoin network wins a race worth more than 3 BTC. For a single home rig, the odds of being that machine on any given day are longer than most state lotteries, which is why almost nobody mines alone anymore. Mining pools are the fix: cooperatives that bundle thousands of machines, share every reward, and turn a violent lottery into something closer to a paycheck. They are also the part of Bitcoin that most worries the people who care about decentralization, because a handful of them now decide what goes into most blocks. Here is how they work, who runs them in the middle of 2026, and why a quiet protocol upgrade is trying to hand power back to individual miners.

The problem pools solve: brutal variance

Bitcoin mining is a guessing game. Machines called ASICs race to find a number that produces a valid block hash, and the network adjusts its difficulty so that one winner emerges about every ten minutes no matter how much hardware is competing. In early July 2026 the network was churning through roughly 1.08 zettahashes per second at a difficulty near 134 trillion, according to CoinWarz. Against that wall of computing power, a single modern ASIC doing 200 to 300 terahashes per second would expect to find a block roughly once every several decades.

Lightning does occasionally strike. In April 2026 a solo miner with about 230 terahashes per second beat odds of around one in 28,000 on the day to claim a full block worth about $210,000, as CoinDesk reported; another win earlier in the year came from just $75 of rented cloud hashrate. These are lottery tickets that happen to hit, not a business plan. If you have bills to pay every month, you need predictable income, and that is exactly what a pool manufactures.

Shares, and how the money gets split

A pool hands each member an easier target than the real network difficulty. Every time your machine hits that easier target, it submits a share: cryptographic proof that you were genuinely working. Shares almost never turn into real blocks, but they let the pool measure exactly how much hashrate you contributed, and that is the basis for splitting the reward when someone in the pool does find a block.

How that split is calculated is the single most important thing to understand about any pool. Four models dominate, and they differ mainly in who absorbs the bad-luck risk when blocks come slowly.

Full Pay Per Share has become the default for large farms because it smooths both the block subsidy and the transaction fees into a steady daily figure, leaving the operator to eat the variance. Pay Per Last N Shares does the opposite: it pays nothing until the pool actually finds a block, then rewards the miners who stuck around, which discourages so-called pool hopping. Operators charge a fee for the service, usually somewhere between zero and 4 percent, with PPS-style schemes costing more because the operator is taking on more risk. The mechanics are laid out in Foundry’s own pool documentation.

The 2026 landscape: a few pools, most of the blocks

The uncomfortable truth is that a very short list of operators now coordinates the vast majority of Bitcoin’s hashrate. Foundry USA, backed by Digital Currency Group and anchored in the United States, has been the clear leader all year; AntPool, owned by ASIC maker Bitmain, sits second. Behind them, F2Pool, SpiderPool, ViaBTC and MARA Pool fill out a top six that together mine more than 80 percent of all blocks.

Those shares are rolling averages that shift every day; the ranking is far more stable than the exact percentages, and the figures here draw on Hashrate Index pool data and CoinDesk reporting. Consolidation is still happening: SBI Crypto said it will shut its pool on 31 July 2026, forcing roughly 2.2 percent of the network’s hashrate to find a new home, per Tech Times.

When concentration becomes a threat

At several points in 2026, Foundry and AntPool have together brushed or crossed the 50 percent line, reviving a fear as old as Bitcoin: the 51 percent attack. A group controlling a sustained majority of hashrate could, in theory, reverse its own recent transactions to double-spend, or refuse to confirm others. Coverage from Bitcoinist and others has flagged the milestone more than once this year.

There are real caveats. A pool does not own its hashrate; the machines belong to thousands of independent miners who can repoint them at a rival pool within minutes if an operator misbehaves, and the cost of mounting a sustained attack on Bitcoin is astronomical. The subtler problem is not attack but selection. Under the long-standing Stratum V1 protocol, the pool operator, not the miner, decides which transactions go into each block. That means a few operators effectively choose the contents of most of the blocks Bitcoin produces, and that is the concentration that turned out to matter most.

Stratum V2: handing block-building back to miners

Stratum V2 is a rewrite of the protocol that connects miners to pools, and its headline feature, Job Declaration, lets an individual miner build its own block template while the pool keeps doing the boring but valuable job of smoothing payouts. Transaction selection moves from the operator to the miner; the pool merely checks that the resulting block is valid.

In May 2026 the idea went mainstream. Seven of the largest pools, representing close to 75 percent of global hashrate, joined a Stratum V2 working group: Foundry, AntPool, F2Pool, SpiderPool, MARA Pool, Block Inc. and DMND, according to CoinDesk. It stopped being theory a few weeks later: on 25 June 2026, DMND and GoMining mined block 955,318 using Job Declaration, the first known production block whose contents were chosen by the miner rather than the pool, as documented by TFTC. The protocol also encrypts the miner-to-pool connection, closing a privacy gap that let internet providers snoop on mining traffic.

Ocean, DATUM and the decentralized wing

Stratum V2 is not the only route. Ocean, a pool that has made decentralized block construction its signature, pushes its own approach called DATUM, short for Decentralized Alternative Templates for Universal Mining. Rather than replace the old protocol wholesale, DATUM layers miner-built templates on top of it: the miner runs a full node and a DATUM gateway, assembles a block from its own view of the mempool, and sends that to Ocean only for reward coordination. The goal matches Stratum V2 even if the plumbing differs, as Spark’s research lays out. Braiins Pool, which helped pioneer the V2 spec, and DEMAND (DMND), the first pool built V2-native from launch, round out a small but growing camp that treats decentralized block construction as the point rather than a feature.

The censorship flashpoint

Why does it matter who picks the transactions? Censorship. In early 2026, developer research flagged that F2Pool, worth around 11 percent of hashrate, was again filtering out transactions tied to wallets on United States sanctions lists, reviving a debate The Block has tracked since Marathon ran a short-lived sanctions-compliant pool back in 2021. When one operator controls a tenth of all blocks, its filtering policy quietly becomes a delay applied to everyone else. Move template-building down to the individual miner, through Stratum V2 or DATUM, and no single operator can impose that filter across a large slice of the network. This is the strongest practical argument for the whole decentralization push.

Mining pools, staking pools and the SEC

Mining pools are a proof-of-work phenomenon, which today means mostly Bitcoin. Ethereum retired mining entirely at its 2022 Merge and now runs on proof-of-stake, where the equivalent cooperatives are staking pools that bundle validators instead of ASICs. Both socialize variance so that small participants can earn steady rewards, but their trust and reward mechanics differ, and United States regulators have treated them differently too.

For miners, 2026 brought welcome clarity. The SEC’s Division of Corporation Finance, under Chair Paul Atkins and his Project Crypto agenda, issued guidance stating that participating in a proof-of-work network, whether solo or through a mining pool, is not a securities transaction and does not meet the Howey test for an investment contract, because rewards flow from protocol rules rather than anyone’s managerial effort (SEC statement). That leaves pooled mining outside the securities regime, in contrast to some staking-as-a-service products that drew scrutiny in earlier years. The sharper pressure on pools comes not from the SEC but from the Treasury’s sanctions office, which is what drives the censorship question above.

Choosing a pool, and what comes next

If you run machines, the practical checklist is short: the fee, the payout scheme (FPPS for the steadiest income, PPLNS if you plan to stay put and want the fees), the minimum payout threshold, server locations near your farm, and, increasingly, whether the pool supports Stratum V2 or DATUM. That last item used to be a purist’s concern; after 2026 it is a mainstream one.

The economics remain tight. Hashprice, the miner-revenue metric coined by Luxor’s Hashrate Index, fell to a post-halving record low near $28 per petahash per day in the first quarter of 2026 before recovering into the mid-$30s after difficulty dropped about 10 percent, one of the largest declines of the year, and roughly 250 exahashes of older machines switched off. You can watch the number move on the Hashrate Index dashboard. With Bitcoin trading near $62,000 and the block reward fixed at 3.125 BTC until the 2028 halving, margins are thin enough that pool choice and electricity price decide who survives. The bigger story, though, is control: if miners keep reclaiming the right to build their own blocks, the fact that six logos still coordinate most of the hashrate will matter a great deal less than it does today.

By the HOGE Wire mining desk.