Bitcoin Mining Pools in 2026: Who Really Builds the Blocks

On July 9, 2026, someone running a $150 Bitaxe, an open-source miner that draws about as much power as a household light bulb, solved Bitcoin block 957,382 and walked away with the full 3.1382 BTC reward, worth close to $200,000 at the time, according to CoinDesk. Stories like that make solo mining sound romantic. They are also statistical freaks. A machine that small has a vanishingly small chance of finding any given block, and for every lottery winner there are millions of machines that will grind for years and never solve one on their own.

That gap, between one lucky block and a steady paycheck, is the entire reason mining pools exist. A pool bundles thousands of machines, finds blocks far more often as a group, and pays each member a smooth wage for the work it contributes. Today the large majority of Bitcoin’s hashrate flows through fewer than ten pools. But pooling does something subtler than smoothing income, and it is the part most newcomers miss: for most of Bitcoin’s history the pool, not the individual miner, has decided which transactions go into every block it finds. In 2026 that quiet arrangement is finally being pulled apart, by a protocol upgrade called Stratum V2, by a scrappier rival design named DATUM, and by a reshuffle of the rankings that saw one established operator switch off entirely.

Why almost nobody mines Bitcoin alone

Bitcoin mining is a guessing game. Every ten minutes or so, the network runs a lottery in which machines race to find a number (a nonce) that, when hashed with the rest of a block, produces a result below a target set by the difficulty. Difficulty stood at about 126.23T after a slight downward adjustment on July 25, 2026, per Hashrate Index, and it retargets every 2,016 blocks to keep new blocks arriving roughly every ten minutes no matter how much hardware joins the race.

The winner takes the whole reward, which since the April 2024 halving is 3.125 BTC plus the transaction fees in the block, and it stays there until the next halving in 2028. At roughly $63,000 per bitcoin in early August 2026, a level that held even as the Fed stayed hawkish under Chair Kevin Warsh, that is close to $200,000 for ten minutes of luck. The catch is the luck. With total network hashrate hovering near 930 exahashes per second (that is 930 billion billion guesses every second), a single modern ASIC contributes a rounding error. A miner running a handful of top-tier machines might expect to find a solo block once every several decades, or never.

That variance is fatal for a business. A mining company with payroll, power bills, and loan payments cannot wait an unknown number of years for one lump sum. It needs predictable weekly revenue. Pools convert the lottery into a wage.

Solo mining still happens, and it still occasionally pays off in headline fashion. The trend even grew in 2026: that $150 Bitaxe drawing 15 to 21 watts took home a full reward worth about $200,000, one of roughly two dozen solo blocks found in a trailing year and up sharply from the year before, per CoinDesk. Those wins are celebrated precisely because they are so improbable. For everyone whose electricity meter cannot absorb years of zero income, a pool is the only rational choice.

What a mining pool actually does

Strip away the branding and a mining pool performs two very different jobs that happen to be bundled together.

The first job is payout smoothing, and it is the one everybody understands. Members point their machines at the pool and submit shares, which are near-miss solutions too weak to win a block but strong enough to prove the machine is doing real work at a known rate. The pool tallies those shares, and when any member finds a valid block, the reward is split in proportion to the work each contributed. A miner with 1% of the pool’s hashrate earns roughly 1% of the pool’s blocks over time, delivered as many small payments rather than one rare jackpot.

The second job is block construction, and it is the one that matters for Bitcoin’s politics. Someone has to decide which of the thousands of pending transactions in the mempool go into the next block, in what order, and with what fees collected. Under the long-dominant Stratum V1 protocol, that someone is the pool operator. The operator builds the block template (the list of transactions plus the coinbase payout) and hands it to members, whose machines simply hash the header they were given. The miners supply electricity and silicon; the pool supplies the block’s contents.

For most of Bitcoin’s history this division went unquestioned because it was invisible. As long as pools built ordinary, fee-maximizing blocks, nobody cared who assembled them. The 2026 debate is about what happens when a pool builds an abnormal block: leaving out legally sensitive transactions, prioritizing its own business, or simply becoming a single point that a government could lean on. It is a fight over what belongs in Bitcoin’s blockspace, the same tension that has already flared over data inscriptions and Taproot.

How pools pay: PPS, FPPS, PPS+, and PPLNS

The alphabet soup of payout schemes comes down to one question: who eats the variance, the pool or the miner?

Under pure PPS, the pool pays a fixed amount for every share and absorbs all the randomness itself, charging the highest fee for that insurance; the miner is paid only from the block subsidy. FPPS, the default at nearly every large industrial pool, works the same way but also folds in an average of network transaction fees, so miners share in fee spikes without waiting for their own machine to catch a fee-heavy block. PPS+ splits the difference, paying the subsidy PPS-style while distributing actual fees by PPLNS.

PPLNS flips the risk back onto the miner: you are paid nothing until the pool finds a block, and then only in proportion to your shares in the most recent window. Income is lumpier, but the design structurally discourages pool-hopping (jumping between pools to skim the easy rewards) and usually carries the lowest fee. Pool fees themselves generally sit in low single-digit percentages, from around 0% to 4% at the major pools tracked by Hashrate Index. For a small home miner, the practical takeaway is that FPPS gives the smoothest, most predictable income, while PPLNS can pay slightly more over a long horizon to those who can stomach the swings.

The 2026 hashrate map: who holds what

Because pools are where hashrate pools, a snapshot of the top operators is effectively a map of who assembles Bitcoin’s blocks. Here is the leaderboard from Hashrate Index in early August 2026, against a network running near 929 EH/s.

The concentration is real but not static. Foundry USA has led for years and still commands about a quarter of the network. What is new in the 2026 data is that F2Pool has edged past AntPool into second place, and that the top four pools together (Foundry, F2Pool, AntPool, and SpiderPool) now clear roughly 70% of blocks, a threshold that prompted headlines about a group of four dominating the resource.

Analysts often summarize concentration with the Nakamoto coefficient, the smallest number of entities that would need to collude to control more than half the network. For Bitcoin pools that number is currently three: Foundry, F2Pool, and AntPool together exceed 50%. It has hovered at three for most of 2026. The two largest, Foundry and F2Pool, add up to about 43%, so no single pair crosses the line, a slight improvement on the tighter readings from earlier in the year. Numbers like these are what critics point to when they call Bitcoin mining centralized. But the share of hashrate a pool commands is not the same as the power it holds, and understanding the difference is the key to the whole 2026 story.

The real risk isn’t a 51% attack

The nightmare scenario people imagine when they see two pools near 50% is a 51% attack: a majority coalition rewriting recent history to double-spend coins. It is worth being precise about why that is not the most realistic threat.

A 51% attack is expensive, self-defeating, and easy to spot. A pool that tried it would be visibly reorganizing the chain; its members, who own the actual machines, would flee within minutes; and the value of the bitcoin the attacker holds and earns would crater. The hardware does not belong to the pool. Miners can and do redirect it at will, which is exactly what makes pool dominance less dangerous than a pie chart suggests.

The more plausible risk is quieter: transaction selection. Because the Stratum V1 operator builds the block template, a dominant pool does not need to attack anything to shape what Bitcoin confirms. It simply chooses what to include and what to leave out. If a pool decides, or is ordered by a regulator, to exclude certain transactions, then a proportional slice of Bitcoin’s blockspace becomes unavailable to those users. No history is rewritten; some transactions just wait longer, or in the extreme are never mined by that pool at all. That is not a hypothetical. It already happened.

The censorship flashpoint: F2Pool and OFAC

The clearest proof that block construction is a real lever came from F2Pool, now the second-largest pool at nearly 18% of hashrate. Starting in 2023, the pseudonymous developer 0xB10C (known as b10c) used a tool called miningpool-observer, which compares the block template a neutral Bitcoin Core node would build against the block a pool actually mined, and found that F2Pool was quietly omitting transactions that spent from addresses on the US Treasury’s OFAC sanctions list. CoinDesk reported the finding, and F2Pool co-founder Chun Wang initially defended it, writing in a since-deleted post: “Why do you feel surprised when I refuse to confirm transactions for those criminals, dictators and terrorists?”

The episode recurred. b10c documented another batch of fifteen missing sanctioned transactions in December 2024, and the pattern has resurfaced periodically since. The reassuring part, and the reason this has not yet broken Bitcoin’s censorship resistance, is that every one of those transactions was picked up by a different pool in a later block. A single filtering pool at 11%, or even 18%, only delays sanctioned transactions; it cannot block them, because the rest of the network’s hashrate still includes them.

But the math changes as concentration rises, and it changes fast if a filtering pool cooperated with others or if regulators pressured several at once. Wang himself made the point that matters most, arguing that a censorship-resistant system “must be designed to resist censorship at the protocol level, rather than relying on each participant to act conscientiously,” and he later disabled the filter pending broader community consensus. That is precisely the argument the 2026 protocol upgrades are built to answer.

Why building the block is worth more than it used to be

Censorship is the political case for caring about block construction. There is an economic one too, and it is growing. Every block’s revenue has two parts: the fixed subsidy (3.125 BTC today) and the variable transaction fees users attach to get their payments included. Each halving cuts the subsidy in half, so over time fees become a larger share of what a block is worth. When the mempool is busy, whoever assembles the block chooses which fee-paying transactions to include and in what order, and that choice is money.

2023 and 2024 offered a preview of how large that swing can be. Waves of data inscriptions and token experiments periodically flooded the mempool, pushing fees so high that some blocks earned more from fees than from the subsidy. Whoever builds the template captures the upside of deciding what fills those bytes. On Ethereum this dynamic has a name, maximal extractable value, and a whole industry of specialist block builders; Bitcoin’s version is simpler but points the same way. As the subsidy fades toward zero over coming decades, the fee market, and therefore control of block construction, moves to the center of mining economics.

That is why the 2026 fight is not academic. Handing template-building back to individual miners does two things at once: it removes the single chokepoint a censor could squeeze, and it spreads the fee-ordering decision across thousands of independent participants instead of a few operators. The pool still smooths payouts and takes its cut, but the valuable, contestable job of deciding what a block contains stops living in one place.

Stratum V2 and the return of block construction to miners

Enter Stratum V2, a ground-up rewrite of the protocol that connects miners to pools. It does several things at once, including encrypting the miner-to-pool link to thwart hashrate hijacking and moving data more efficiently, but the feature that matters for decentralization is Job Declaration. Under Job Declaration, the individual miner runs its own Bitcoin node, builds its own block template, and tells the pool what it intends to mine. The pool’s role shrinks to what many argue it always should have been: validating work and smoothing payouts. The choice of which transactions go into a block moves from the operator back to the thousands of machines actually doing the hashing.

2026 is the year the idea left the lab. On May 7, seven of the largest pools (AntPool, Block, DMND, F2Pool, Foundry, MARA Foundation, and SpiderPool), together representing roughly 75% of global hashrate, joined a Stratum V2 Working Group to coordinate adoption. Weeks later, on June 25, the milestone that theorists had described for years finally happened in production: DMND, the DEMAND pool and the first V2-native pool (public since November 2025), mined block 955,318 using a template built independently by the mining firm GoMining.

The people involved framed it as a turning point. DMND chief executive Alejandro De La Torre said, “A miner just mined the first Stratum V2 block to power their own product end to end.” GoMining chief executive Mark Zalan put the history bluntly: “For years, mining pools have determined which transactions are included in Bitcoin blocks.” The whole point of Job Declaration is to end that default.

Braiins Pool, the direct descendant of Slush Pool (the first Bitcoin mining pool ever, launched by Marek Palatinus in November 2010 and rebranded in 2022), had pioneered V2 years earlier and runs it natively too. The Stratum V2 Reference Implementation team projects that new ASIC firmware will ship with V2 as the default by the end of 2026, which could carry adoption toward a majority of hashrate. The caveat is that a working group is not a deployment. Most ASICs in the field still run V1 firmware, and Job Declaration in its full form needs firmware most machines do not yet have. Signaling support is cheap; rewiring a global fleet is not.

DATUM and Ocean: decentralizing without new firmware

While the big pools coordinate around Stratum V2, a smaller pool has spent 2026 proving the same principle with different plumbing. Ocean, launched in November 2023 and co-founded by longtime Bitcoin Core developer Luke Dashjr, uses a system it calls DATUM (Decentralized Alternative Templates for Universal Mining). DATUM asks the miner to run a full node and a lightweight gateway that builds the block template locally, then submits it to the pool for payout accounting.

The crucial engineering difference is deployability. Stratum V2’s Job Declaration generally needs new firmware on the ASIC. DATUM is layered on top of existing Stratum V1, so a miner can build its own blocks today without waiting for a firmware upgrade. It trades some of V2’s elegance and encryption for the ability to run on hardware already humming in the racks. Ocean has attracted backing that punches above its 2.4% hashrate share, including a $6.2 million seed round led by Block Inc. chief executive Jack Dorsey and, since 2025, hashrate directed to it by the stablecoin issuer Tether.

Dashjr has been blunt about why any of this matters. Describing Ocean’s effect, he wrote on X: “Instead of two blocks made by the same pool, we get blocks made by two independent miners, exactly the same as if they weren’t using any pool at all.” The point is not that Ocean is large. It is that when its miners each build their own blocks, the pool stops being a single decider even though the payouts still flow through one place.

SBI Crypto’s exit and the churn beneath the rankings

Concentration figures make pools look like fixed institutions. They are not. On July 31, 2026, SBI Crypto, the mining arm of Japan’s SBI Group, permanently shut its public mining pool, stopping share submissions at 22:00 UTC on July 30. The pool had accounted for roughly 20 to 21 EH/s, about 2.2% of the network and enough to rank around eleventh worldwide, per CoinDesk. Its shutdown notice pointed departing miners toward other pools, including Braiins and Luxor, and left more than 20 EH/s of hashrate hunting for a new home over a matter of days.

The SBI exit is a useful reminder of two things. First, pool hashrate is mobile: more than 2% of the network’s security had to re-home almost overnight, and it did, mostly flowing to existing large pools, which nudges concentration up rather than down. Second, the pool business is genuinely hard right now. With hashprice (a miner’s daily revenue per unit of hashrate) sitting around $32 per petahash per day in early August 2026 and described by Hashrate Index as at or below breakeven for many operators, running a pool as a thin-margin service on top of a squeezed industry is not obviously attractive. When a name as established as SBI walks away, it says something about where the economics point.

The economics behind which pool a miner picks

For an individual miner, choosing a pool is a spreadsheet decision layered over a values decision.

The spreadsheet part is fees and payout scheme against expected revenue. On revenue of roughly $32 per petahash per day, a single percentage point of pool fee is real money at scale, which is why large farms negotiate hard and generally land on FPPS with a low headline fee. Reliability, payout frequency, minimum-withdrawal thresholds, and the operator’s track record of paying out fees honestly all feed the same calculation. So does geography and latency: a pool with servers far from the miner loses a sliver of revenue to stale shares that arrive too late to count.

The values part is what 2026 has added to the checklist. A growing minority of miners now weigh whether a pool supports Stratum V2 Job Declaration or DATUM, because building your own block template is the difference between renting your hashrate to an operator’s editorial choices and keeping that choice yourself. It also has a self-interested angle: if regulators ever pressure large pools to filter transactions, miners on template-sovereign pools are insulated. For most industrial miners the fee still wins, but the fact that block-construction sovereignty now appears on the list at all is the clearest sign the debate has moved from forums into procurement.

Pools, staking, and Bitcoin’s security budget

Mining pools are the proof-of-work version of a broader pattern. On proof-of-stake networks like Ethereum, which stopped being mineable when it merged to staking in 2022, the analogous institution is the staking pool: an operator that bundles many depositors’ capital, runs the validators, and distributes rewards, with the same centralizing gravity and the same argument about who controls block content (there expressed as MEV and proposer-builder separation). The economic engine differs (electricity and silicon versus staked capital), but the political question is identical: how do you get the efficiency of pooling without handing a few operators the keys to the chain?

That question sits inside the larger one of Bitcoin’s security budget, the total value the network pays miners to keep it honest, which HOGE Wire has compared against Ethereum’s staking-based model. As the block subsidy halves every four years, more of that budget must come from transaction fees, and fees are set by what fills blocks, which loops back to who builds the templates. Decentralizing block construction is not only an anti-censorship project; it is part of keeping the fee market competitive and legible as the subsidy fades. And all of it rests on a physical industry whose power sourcing HOGE Wire has mapped in detail, because the cheapest, most reliable electricity is ultimately what decides which pools grow and which switch off.

Where US regulators landed on mining pools

For American miners, the regulatory picture got clearer in 2026, and it broke in mining’s favor. In March 2025 the SEC’s Division of Corporation Finance published a statement clarifying that proof-of-work mining, including participating in a mining pool, is not a securities transaction, on the reasoning that a miner’s reward comes from its own computational work (an administrative or ministerial act) rather than the entrepreneurial efforts of a third party, so it fails the Howey test, as The Block reported.

In March 2026 the agency went further. In an interpretive release tied to Chair Paul Atkins’s Project Crypto agenda, the SEC extended the no-securities view to a range of network activities including mining, staking, wrapping, and airdrops distributed for no consideration. Atkins framed it plainly in the accompanying statement: “This is what regulatory agencies are supposed to do: draw clear lines in clear terms.”

Two caveats matter. The securities question is not the sanctions question: OFAC compliance is a Treasury matter, which is exactly the pressure that makes F2Pool-style filtering a live issue regardless of how the SEC treats pool rewards. And crypto derivatives such as hashrate futures fall to the CFTC, not the SEC. The upshot for a US pool participant is that earning pooled mining rewards is settled ground, while how blocks are built, and who can be leaned on to shape them, is one of the deadlines and open questions still shaping crypto in 2026.

How to choose a mining pool in 2026

For a miner weighing where to point machines in 2026, a few factors matter more than the marketing:

• Payout scheme and fee: FPPS for the smoothest income, PPLNS if you can absorb variance for a potentially lower fee; compare the all-in cost, not just the headline rate.
• Size and decentralization: a mega-pool offers frequent, reliable payouts but adds to concentration; a smaller pool spreads the network’s block-building around but pays more lumpily.
• Block-template control: check whether the pool supports Stratum V2 Job Declaration or Ocean’s DATUM if you want to build your own blocks instead of mining the operator’s template.
• Payout reliability and thresholds: look at the minimum withdrawal, payout frequency, and the operator’s record of honoring transaction-fee payouts.
• Geography and uptime: server proximity reduces stale shares, and published uptime and support responsiveness protect revenue.
• Transparency: does the pool publish its hashrate, blocks found, and fee policy? Third-party dashboards like Hashrate Index make claims easy to verify.

None of these is one-size-fits-all. A publicly traded miner optimizing quarterly numbers will weight fees and reliability; a sovereignty-minded operator will pay a little for template control. The healthy sign for Bitcoin is that, in 2026, that second column finally exists.

Frequently Asked Questions

What is a Bitcoin mining pool?

A mining pool is a service that combines the computing power of many miners so they find blocks as a group and share the reward in proportion to the work each machine contributes. It converts solo mining’s rare, all-or-nothing jackpots into steady, predictable payouts, which is why most of Bitcoin’s hashrate mines through fewer than ten pools.

Which Bitcoin mining pool is the largest in 2026?

Foundry USA, operated by Digital Currency Group, remains the largest with roughly a quarter of network hashrate (about 25%) in early August 2026, according to Hashrate Index. F2Pool sits second near 18%, having edged past AntPool, and the top four pools together account for roughly 70% of blocks.

Is it better to mine Bitcoin solo or in a pool?

For almost everyone, a pool is better. Solo mining offers the full block reward (3.125 BTC plus fees, worth around $200,000 in mid-2026), but a small miner’s odds of ever finding a block can stretch into decades. Pools trade that jackpot for frequent, predictable income. Solo mining makes sense mainly as a lottery-style hobby or for very large operations.

What is the difference between FPPS and PPLNS payouts?

Under FPPS (Full Pay Per Share), the pool pays a fixed amount per share plus an averaged share of transaction fees and absorbs the variance itself, so income is smooth; large farms prefer it. Under PPLNS (Pay Per Last N Shares), miners are paid only when the pool finds a block, so income is lumpier, but fees are often lower and pool-hopping is discouraged.

Can a mining pool censor Bitcoin transactions?

It can delay them. Because the pool builds the block template under the older Stratum V1 protocol, it can leave specific transactions out, as F2Pool did with OFAC-sanctioned transactions starting in 2023. Other pools still include them in later blocks, so one pool cannot permanently block a transaction. Newer designs such as Stratum V2 Job Declaration and Ocean’s DATUM let individual miners build their own templates, reducing the risk.

Yuki Tanaka is a mining and infrastructure correspondent at HOGE Wire, covering Bitcoin’s hardware, energy, and the protocol politics that decide what goes into a block.