What Are Bitcoin Mining Pools? How They Work

A Bitcoin mining pool is a coordinated service where many miners combine hashrate via Stratum so the group finds blocks more often than any participant alone, then splits the subsidy and fees by a defined reward model after a 1 to 4% operator fee. Pools exist because solo variance is brutal at network scale near 994 EH/s (source: Hashrate Index).

Mining pools in one screen, what actually matters in 2026:

1. Network hashrate is running near 994 EH/s in May 2026, so a 100 TH/s miner solo-mines an expected block once every ~187 years and even a 335 TH/s S21 Hydro waits ~56 years; pooling collapses that variance to daily or per-block accounting Hashrate Index

2. Foundry USA holds about 30.95% of 3-day hashrate and AntPool about 20.55%, meaning two pools produce roughly 51% of current blocks, the recurring centralization concern (source: Hashrate Index)

3. Reward models split the variance risk between operator and miner: FPPS (Foundry, F2Pool 4%) gives fixed per-share income, PPLNS (AntPool 0% fee, F2Pool 2%) pays only when the pool finds a block, SOLO returns the full 3.125 BTC plus fees minus a small pool cut

4. Stratum V2 adoption accelerated in 2026 with Braiins Pool fully V2-capable and DEMAND launched as a V2-native pool, letting miners construct their own block templates rather than accepting pool-assigned ones (source: Stratumprotocol)

5. OCEAN's DATUM template system, Public Pool's zero-fee self-host option, and the Braidpool DAG proposal are the three active decentralization alternatives for miners who treat pool-level censorship as a real tail risk

This article walks through mining pools the way operating miners and serious researchers want them explained: why variance reduction is the whole reason pools exist, how Stratum shares and block templates actually flow between pool and miner, what the PPS/PPLNS/FPPS/SOLO/Prop payout math does to your income distribution, what the 2026 hashrate concentration looks like, and how Stratum V2 plus DATUM plus Braidpool are trying to move block-template control back toward the hashers. For vocabulary gaps, the Bitcoin glossary covers Stratum, nonce, share difficulty, and block template before going deeper.

What a mining pool actually is

A Bitcoin mining pool is a pool server plus accounting system that accepts work from many miners and distributes block rewards by a stated rule. It is not a separate network or token; it receives partial proof of work submissions over TCP and pays out to miner-supplied Bitcoin addresses after blocks confirm.

Understanding how Bitcoin mining works at the hardware level is the prerequisite for thinking about pools: the pool does not change the mining work itself. Every ASIC still hashes against the same network target; the pool only changes how wins get accounted for.

Why pools exist: variance reduction at industrial scale

Solo Bitcoin mining is a Poisson process. Each nonce attempt is independent, time between valid blocks is exponentially distributed, and the mean gap equals block time divided by that miner's share of network hashrate.

At May 2026 numbers, one Antminer S21 Hydro at 335 TH/s against 993 EH/s contributes ~3.4 × 10^-7 of global hashrate, for ~56 years expected solo wait. A 1 PH/s farm waits ~19 years; a 10 PH/s hobby operation ~1.9 years, still with huge variance.

Variance around these means is brutal at a small scale. A 335 TH/s solo miner with a 56-year expected time has a ~37% chance of waiting longer than that for their first block, far past machine economic life. The same statistical mechanic controls how Bitcoin confirmations arrive: pooled hashrate produces steadier wins that keep the network's 10-minute pace.

Joining a pool does not make hardware stronger or change global difficulty, as covered in how Bitcoin difficulty adjusts with competition. It replaces "one 3.125 BTC payout once a decade if average" with "a small payout every day that tracks contributed work." For most operations, that swap is the difference between a viable business and a lottery ticket.

How a mining pool works end to end

The operational flow from miner bootup to paid reward:

1. The ASIC opens a TCP connection to the pool's Stratum endpoint, authenticating with a pool-registered worker name.

2. The pool issues a job: block template plus share difficulty target well below the network target.

3. The ASIC iterates nonces and extra-nonces to build candidate block headers, double-SHA-256ing each.

4. When it finds a hash below the share target, it submits the share. Most do not also clear network targets, so most are accounting tokens only.

5. The pool validates each share, records it against worker count, and adjusts share difficulty as worker hashrate changes (vardiff).

6. If a share clears the full network target, the pool treats it as a valid block solution and broadcasts the block.

7. The coinbase transaction pays 3.125 BTC subsidy plus collected fees to a pool-controlled address.

8. After 100 confirmations (coinbase maturity), the pool credits earnings per reward model and pays once workers clear their minimum threshold.

The coinbase transaction is where pool versus miner control becomes visible on-chain. Tools like mempool.space read the coinbase to attribute blocks to pools by matching output address and coinbase tag bytes. For how transactions become part of mined blocks, the pool's block template selection is the step where transaction inclusion decisions happen, which is why block-template control is the core of the Stratum V2 debate.

Stratum, shares, and share difficulty in plain English

Stratum is the protocol connecting miners to pools. V1 dates to 2012, encodes messages as line-delimited JSON, and makes the pool authoritative over block templates.

A share is any hash below the pool's internal share difficulty, which is much easier than network difficulty. If the network is 132T and share difficulty is 65,536, miners produce shares hundreds of thousands of times more often than valid blocks. Each share is probabilistic proof of real hashing work at reported hashrate.

Shares have no on-chain existence. They are pool-internal accounting proofs; if the pool server fails, shares since the last snapshot are lost. For how proof of work secures Bitcoin, the consensus rule operates only on the network target. The share system is a statistical sampler of each miner's hashrate, nothing more.

The block template is the candidate block the pool sends: previous block hash, Merkle root of selected transactions, timestamp, nBits, and an extranonce range for the miner to iterate. In Stratum V1 the pool picks transactions. In Stratum V2 the miner can pick the core centralization fix the protocol was designed for. The Bitcoin developer reference documents the template interface (source: Bitcoin Dev Docs).

PPS vs PPLNS vs FPPS vs SOLO vs Prop: how reward models actually pay

Every major reward model answers the same question differently: who eats the variance when blocks arrive faster or slower than expected?

• PPS pays a fixed rate per valid share from statistical expected value. The operator absorbs all variance; fee is higher to compensate.

• PPLNS pays only when the pool wins a block, for shares in the most recent N-share window. Bad luck means miners earn less until blocks arrive; good luck means more. Fee is lower because miners carry variance. AntPool 0% and F2Pool 2% are the 2026 PPLNS reference quotes (source: D Central).

• FPPS is PPS plus an average transaction-fee component. Transaction fees have grown as a share of the block reward, a dynamic traced in why Bitcoin fees get high, and pure PPS miners missed that upside. Foundry USA defaults to FPPS, AntPool and F2Pool charge 4% FPPS.

• SOLO at a pool means: use the Stratum endpoint, keep the whole block reward when your share wins. Public Pool runs it at 0% fee with self-host option (source: Blockdyor). Multiple 5 to 50 TH/s solo operators found blocks in 2025 and early 2026, keeping the full 3.125 BTC subsidy plus fees (source: Mineshop).

• Prop pays proportional to shares in the current round. Largely deprecated because round boundaries let miners pool-hop during unlucky stretches; PPLNS's rolling window solves that.

Model choice is a risk tolerance statement. Expected value is similar modulo fees and luck; what differs is day-to-day distribution.

Pool fees, minimum payouts, and what miners should compare

Advertised fee is one line in a full pool comparison. The rest:

A pool advertising 0% fee with opaque share accounting is not automatically preferable to a 2% pool with a real-time public dashboard. mempool.space publishes live pool hashrate shares and block-finding history (source: Mempool.space).

From Blofin's exchange-operator vantage point, we observe miner-labelled deposit patterns across payout cycles, and the pools with the cleanest on-time minimum-threshold settlement are the ones whose depositors show the tightest inter-payout intervals. That is a different signal from advertised fees: operator reliability as behavioural data rather than marketing copy.

2026 hashrate distribution: Who actually mines Bitcoin right now

Pool concentration in May 2026, 3-day window from Hashrate Index Hashrate Index:

Foundry USA plus AntPool hold roughly 51% of recent blocks, the widely-cited "two pools control majority" headline (source: Cryptoslate). Pools are not themselves miners; they are settlement intermediaries for thousands of independent machines. The 51% figure measures concentration in the template-selection layer, not hardware ownership.

In late March 2026, Foundry USA mined seven blocks in a row, producing a rare two-block reorg (source: Cryptotimes). Seven-in-a-row is statistically expected at ~31% share once per few thousand blocks; nodes accepted the longer chain cleanly.

Centralization concerns and what they actually mean

Pool concentration matters because pools choose which transactions appear in candidate blocks. A majority pool does not get to rewrite consensus rules, which nodes enforce independently, but it does get disproportionate influence over transaction inclusion, ordering, and the minor selfish-mining advantages documented in protocol research. Risks worth distinguishing:

• Transaction censorship. A pool could refuse to include specific transactions (for example, transactions touching sanctioned addresses). If two pools at 51% of hashrate adopt the same censorship list, affected transactions get included roughly half as fast.

• MEV-style extraction. Pools can reorder or selectively include transactions to capture arbitrage or liquidation value. Historically muted on Bitcoin but attention grew through the ordinal and BRC-20 boom periods.

• Regulatory capture. A pool based in one jurisdiction is subject to that jurisdiction's rules. Foundry USA complies with US sanctions; ViaBTC serves Russian miners. Mandates that pools follow national blocklists would splinter the pool market rather than Bitcoin itself.

• Coordination risk. Five pools control ~80% of hashrate. Coordinated action by all five would require leaking to multiple operators without news leaking, and operators have reputation and withdrawal-run reasons to defect.

What pools cannot do, even at 100% hashrate share: change the 21-million supply cap, print coins, rewrite old transaction history beyond shallow reorgs, or force invalid blocks onto honest nodes. For the full taxonomy see what a 51% attack can and cannot do. Pool concentration is a real problem in narrow, important ways, not a civilisational collapse scenario.

How Stratum V2 and DATUM move template control back to miners

The centralization fix in active rollout is moving block-template construction from pool server to miner, so the pool becomes a share-accounting and payout utility and no longer selects transactions.

• Stratum V2. Binary wire format, end-to-end encryption, and a Job Declaration flow where miners submit their own templates Stratumprotocol. Braiins Pool ran fully V2-capable for over a year; DEMAND launched as the first V2-native pool in early 2026 (source: Bitcoinmagazine). Bandwidth savings ~60% pool-side and ~70% miner-side.

• DATUM (OCEAN). Decentralized Alternative Templates for Universal Mining. Miners run a Bitcoin node and construct templates locally, then hand shares to the pool for variance-reduction accounting D Central. Adoption grew through 2025 and into 2026 among operators treating pool-level censorship as compliance and principled risk.

• Braidpool (proposal). A peer-to-peer pool using a directed acyclic graph for share accounting, with no central pool server to compromise (source: GitHub). Each miner runs a node and constructs templates; the DAG records contributions without a privileged coordinator. Still in specification and implementation stages in 2026, not yet running production hashrate at scale.

• Public Pool. Zero-fee solo mining with self-host option. Miner connects ASIC, finds a block, keeps the full reward Blockdyor. No variance reduction; pure solo at infrastructure scale most hobbyists cannot run themselves.

At BloFin, when we watch pool market-share trends alongside Stratum V2 deployment and DATUM adoption signals, the centralization picture is less monolithic than a two-pool-majority headline suggests: template-construction sovereignty is already partially decoupling from hashrate aggregation, and that trajectory is the relevant metric to watch rather than raw pool share alone.

How to join a mining pool (the actual setup)

Assuming you already own working ASICs and have resolved electricity and cooling, joining a pool takes five steps.

1. Pick a pool and reward model. Decide on fee, payout method, Stratum V2 support, and minimum threshold. Small farms (< 500 TH/s) often prefer FPPS for smoothness; larger operations trade lower PPLNS fees for absorbable variance.

2. Create a pool account. Most pools require email, worker name, and payout Bitcoin address. Institutional pools like Foundry USA require KYC and contractual onboarding.

3. Configure the ASIC. In the miner's web UI, enter the pool's Stratum endpoint (for example stratum+tcp://btc.viabtc.com:443), worker username (account.worker), and password (x). Set a backup endpoint for failover.

4. Verify shares and hashrate. After connecting, pool dashboard shows shares within minutes. Pool-dashboard hashrate should match miner-reported hashrate within a few percent; the gap reflects stale and rejected shares.

5. Set payout threshold and address whitelist. Confirm the payout address, optionally whitelist it to lock funds to pre-approved destinations. For the operator-side variance tradeoff between the two dominant reward models, see PPS vs PPLNS.

Stale shares are submissions that arrive after the pool moved to a new job (another miner found a block). Stale rate of 1-2% is normal; above 5% suggests Stratum latency or farm network congestion.

Pool risks that do not disappear after you join

Pooling solves variance but introduces risks solo mining does not have.

• Payout risk. Operators can delay payouts, change policies mid-stream, or shut down without returning earnings. Concentrated in smaller, newer, unregulated pools.

• Kickback schemes. Some pools rebate fees to large miners, distorting headline-fee comparison for smaller operators.

• Censorship exposure. A filtering pool forces your hashrate into that policy. Stratum V2 and DATUM exist to remove this exposure.

• Regulatory risk. A pool caught between jurisdictions can have servers seized, banking cut, or withdrawals frozen. Splitting across two pools in different jurisdictions is an operational hedge.

• Withholding attacks. A minority at a PPLNS pool can submit shares but withhold any that also solve a block, reducing pool luck without missing their own payout. Detection is statistical and slow.

• Operator insolvency. If the pool goes bankrupt between block win and coinbase maturity (100 confirmations), unpaid earnings may become unsecured claims.

None of these erases mining's economics: electricity, hardware efficiency, and BTC price still drive whether operating is profitable at all, covered in is Bitcoin mining profitable.

Pool versus solo decision framework

For a miner deciding between pool and solo, the answer is almost always pool, but the "almost" matters.

The lottery-tolerant path is real: the expected wait for a 10 TH/s miner at 993 EH/s is ~1,870 years, but Poisson processes produce early hits. A miner treating it as a lottery with 1 year of runway and accepting the likely zero outcome is being rational about utility, not naive.

Common confusions worth flagging

"Pool hashrate is pool-owned hashrate." 

No. Pool hashrate is the sum of independent miners pointing ASICs at that pool's Stratum endpoint; they can redirect in minutes. Foundry does not own 296 EH/s of machines; it settles rewards for operators who collectively run that much.

"PPS is always better than PPLNS because it pays the same amount." 

Expected value is similar before fees; PPS typically has a higher fee because the operator absorbs variance. Over long horizons at matching fees, net income converges. PPS wins only if cash-flow predictability has a specific value.

"The pool picks transactions, so the pool controls Bitcoin." 

Pools pick transactions for block templates in Stratum V1, but consensus rules are enforced by full nodes, not pools. Pools cannot include invalid transactions, double-spend old coinbase outputs, or print coins. Pool transaction selection is a censorship concern, not consensus capture.

"Solo mining lets you skip all the pool fees." 

Solo skips fees but inherits full variance. Expected hourly income is similar before fees; the distribution is collapsed into rare huge payouts. For a small miner, 100% variance is incompatible with fixed electricity bills.

"Stratum V2 is already the default and centralization is fixed." 

V2 adoption is accelerating but still a minority of hashrate. Projected default for new ASIC shipments by end of 2026; existing fleets take years to roll. DATUM is a single-pool solution for now. Braidpool is not yet production-live.

Frequently asked questions

What is a Bitcoin mining pool in simple terms?

A Bitcoin mining pool is a service that coordinates many miners through the Stratum protocol, combines their hashrate, and distributes block rewards across participants by a stated rule once the pool wins a block. The pool runs an accounting system recording each miner's share contributions, constructs (or relays) the block template each miner works on, and pays out after blocks confirm at least 100 confirmations for coinbase maturity. Joining a pool does not improve the miner's hardware or reduce electricity costs; it only replaces rare large rewards with frequent smaller ones that track each miner's share of contributed work.

How do mining pools split the block reward?

Pools use reward models that shift variance between operator and miner. FPPS pays a fixed rate per valid share that includes a transaction fee component, with the operator absorbing variance; this is Foundry USA's default and costs 4% at AntPool and F2Pool. PPLNS pays only when the pool wins a block, weighted by recent shares, with AntPool at 0% and F2Pool at 2%. SOLO returns the full 3.125 BTC subsidy plus fees to whichever miner's share actually solves the block, minus a small pool cut. Expected net income is similar across models; what differs is variance.

What is the difference between PPS, PPLNS, and FPPS?

PPS pays a fixed amount per valid share regardless of whether the pool won blocks; the operator absorbs variance, reflected in a fee typically 2 to 4%. PPLNS pays only when the pool wins a block, weighted by shares in a rolling N-share window; miners absorb variance, so fees are typically 0 to 2%. FPPS mirrors PPS structurally but adds an average transaction-fee component to the per-share rate, so miners capture fee upside. None of the three guarantees profit; all three equalize to similar expected value after fees over long windows.

What mining pools are the largest in 2026?

Foundry USA leads with about 30.95% of 3-day hashrate in May 2026 at roughly 296.94 EH/s, followed by AntPool at 20.55%, F2Pool at 11.78%, SpiderPool at 7.62%, ViaBTC at 7.39%, SecPool at 5.08%, MARA Pool at 4.85%, Luxor at 3.00%, Binance Pool at 1.85%, and Braiins Pool at 1.62% (source: Hashrate Index). Foundry USA and AntPool combined produce roughly 51% of current blocks, the centralization headline most often cited. Those percentages measure template-selection concentration at the pool layer, not hardware ownership; the underlying ASICs belong to thousands of independent operators who can redirect hashrate between pools at any time.

How much do mining pools charge in fees?

Fees run from 0% to 4% of distributed rewards depending on the reward model. FPPS pools charge higher fees (AntPool 4%, F2Pool 4%, Braiins 2.5%) because the operator absorbs variance. PPLNS pools charge lower fees (AntPool 0%, F2Pool 2%, Braiins 0%) because miners carry it. PPS+ variants sit in between at 2.5%. Foundry USA's institutional FPPS schedule is tiered by quarterly hashrate and not publicly fixed. Stated fee is one line in a full comparison; minimum payout thresholds, Stratum V2 support, and operator track record all matter too.

What is Stratum V2 and does it fix mining pool centralization?

Stratum V2 is a 2022-onward protocol upgrade that uses a binary wire format, encrypts miner-to-pool traffic, and introduces Job Declaration so each miner can construct its own block template rather than accepting the pool's. It addresses template-level transaction censorship without touching variance reduction. Braiins Pool is fully V2-capable, DEMAND launched as a V2-native pool in early 2026, and V2 is projected to be default on new ASIC shipments by end of 2026. It moves template sovereignty to hashers but does not itself redistribute pool hashrate; aggregation and template selection are separate problems.

Can I still mine Bitcoin solo in 2026?

Yes, but variance is extreme. At May 2026 network hashrate near 993 EH/s, one 335 TH/s Antminer S21 Hydro faces an expected ~56-year wait between solo blocks, with a 37% chance of waiting longer. A 10 TH/s hobby miner faces an expected wait near 1,870 years. Multiple 5 to 50 TH/s solo operators did find blocks in 2025 and early 2026 through luck, keeping the full 3.125 BTC subsidy plus fees at zero pool cost. Solo is rational only as a lottery with a full-cost runway. Public Pool, CK Pool, and Braiins Solo run zero-fee solo services.

Researched and written by the BloFin Academy editorial team with AI-assisted drafting. All facts independently verified.

Disclaimer: This content is for educational purposes only and does not constitute financial, investment, legal, or tax advice. Crypto assets are highly volatile and carry significant risk of loss. Always verify local regulations and consult a qualified professional before making financial decisions.