Mining Pools Review · All pools

Bitcoin Mining Pools 2026: Concentration, Decentralisation or Stagnation? Ranking. FPPS. Proxy. Hashrate Marketplace. Stratum V2 and DATUM

covering 1 Jan - 11 Nov 2026 · published 13 Nov 2026 · next release expected 15 Nov 2026 (15th and 30th of each month)

• Even with the AI pivot, the leading pool share grew. The top-five pools' share went from 75.46 % in January to 76.71 % in November, and their hashrate rose 7 EH/s while every other pool together fell 13 EH/s, from 240 to 227.
• Pools enter the capital markets. SpiderPool's SPAC agreement, announced on 30 October, would take the pool public at a US$250 M valuation; the pool found 3,853 blocks in 2026 at about 90 EH/s.
• Hashrate aggregation pools: declared is not mined. A group declaring on average 125 EH/s on miningpoolstats finds almost no blocks under its own name. What a pool declares and what the chain shows are different questions, and it raises the question of which parent pool is the recipient.
• Payout visibility. Foundry paid 28.00 % of the year's blocks directly to large miner wallets that can be tracked. 25.1 % of all global payouts went through Cobo custody, and for 44.7 % the payout route is invisible for now.
• Custody is the concentration. Four pools - SecPool, WhitePool, Braiins and CloverPool - route their payouts into Cobo, built by founders from F2Pool with a clear understanding of the mining sector.
• Marketplace, not a pool. NiceHash leads with 198 blocks under its own tag and 104 inside Ocean; Braiins rolled out Hashpower and BitFuFu ($FUFU) teamed up with Finloop to sell hashrate through a token.
• Four pools mined 1,568 blocks in 2026 and shut down. Binance Pool stopped on 10 November, SBI Crypto in July (last block 29 July), Ultimus on 25 September, f2pool.kz closes on 21 November: about 60 EH/s against a network that averaged 962.5 EH/s over the year, roughly 6 % of the network's hashrate.
• Technology moves block control. DMND mined the first known Stratum V2 job-declaration block, 955318, in June; MARA signalled SV2 on its first block, 975369, on 11 November. Ocean is ramping up after BIP-110 with DATUM, which gives miners their own block, tag and visibility.
• Solo miners are the sector's R&D sandbox: the biggest group by number of people, small in hashrate, and with direct access to templates they will keep testing, modernising and taking their luck at finding blocks.

1 · Why pool concentration is the wrong frame, and what replaces it

Bitcoin mining transformed faster than any other part of its stack except one - pools. Hashrate, hardware, energy contracts and capital availability have all been rebuilt since 2020. The pool layer - the software that assembles the block and pays the miner - has barely moved: Stratum V1 and FPPS are still the backbone, and the main trigger for progress is not a yield product from a large pool but technological evolution. That is what this review is about.

Stratum V1 was a template monopoly by design until 2026. The pool builds the block, chooses the transactions and hands the miner work to hash. The miner is paid, not consulted. FPPS is the other half of the same arrangement: it converts hashrate into daily, predictable cash, which is the lifeline of a mining operation and the reason a miner accepts the arrangement at all. Together they explain why the market consolidated around a handful of pools without anyone deciding to consolidate it. When the sector needed a boost for new fleet deployment, especially in the two prior halving cycles, that FPPS support line was what let a miner get to ROI when capital access was limited - which is why AntPool's concentration was never about controlling hashrate so much as it was vendor financing for Bitmain in stimulating the sector to buy new hardware and grow hashrate.

Stratum V2 and DATUM break this stagnation. Job declaration lets the miner build the template and declare it; the pool validates and coordinates, but no longer decides what is in the block - it can be the miner's option, but a miner can still delegate all of it to the pool. This is not a change in who finds blocks: it is a change in who controls what the block contains. DATUM, which Ocean runs, offers the same control from the pool side. For the first time the choice is explicit: mine through a pool's template, or mine your own template inside a pool.

Control over the template is the thing a miner can hold. BMM's reading is that at least 150-200 large miners (approximately 70 already tagged via Ocean) with a few EH/s already have the operational capacity to run their own templates if the economics and internal policy allowed it, and not all will take the option, because their treasury and risk management need the predictable FPPS cashflow more than they need control. An own template means moving to PPLNS and accepting block-variance risk: acceptable for a large player whose EH/s still produces blocks every day or week, an option that could be harder for a small miner who may wait one to four weeks for a block.

The transition will not be driven by hashrate moving between pools but by miners deciding, one at a time, that control is worth more than a fixed daily payment - or choosing an active mix, with FPPS covering fixed monthly OPEX and PPLNS carrying the HODL strategy. BMM runs the 1 PH/s benchmark tracker as the visualisation of those strategies (buy and hold BTC; mine and sell daily; mine and hold), and of what the choice depends on: whether your funding base is in dollars or in BTC, and your time horizon. Bitdeer, which mines and divests BTC, and BitFuFu, which mines and holds, are the two live examples.

We shall not expect an instant improvement in miner economics from template control. A miner who builds his own template does not get richer from fees; he gets control of a lever that is small now and may not stay small. The real obstacle is the payment model that can generate a difference in picking FPPS versus PPLNS.

2 · FPPS: the cashflow that blocks the transition

Every argument about decentralisation runs into the same wall: miners need daily cash. FPPS pays the expected value of the block reward plus fees every day, whether or not the pool has found a block. That is a balance-sheet service before it is a pricing formula, and only pools with capital can offer it. It is also the main barrier to entry: a new pool cannot easily promise daily FPPS to a miner who has never heard of it, and it cannot fund the promise competitively.

FPPS is therefore the centralisation mechanism that no share table shows. A miner on FPPS has chosen predictable revenue over control: he takes the pool's template and the pool's payment because together they remove block variance risk from his business. A miner who builds his own template takes variance back with full risk-return, and under FPPS that variance is exactly what he is paying to avoid.

Only the narrative of transaction selection is, in the current state, light commercial reasoning. Fees were 0.56 % to 0.74 % of the block reward each month in 2026, and 894.5 BTC over the year: a miner who controls his template controls about 0.6 % of his revenue today. We would not expect 0.6 % to be rational grounds for a miner to change his payment model, but control over it is the only part of the block a miner can own. The technology has arrived before the economic reason to use it; the reason will come from fees, not from ideology.

What would change the balance: FPPS offered on top of a self-built template, so a miner does not have to choose between cash and control; pools that finance miners against their own templates; and fee growth, which is what turns 0.6 % into a decision. Another route is the additional fee uplift from merge mining and other services: DMND, for example, integrated Rootstock (RSK) merge mining into template formation, and RSK pays the miner directly rather than the pool, unlike the way merge-mining payments are normally structured.

3 · Back to pool concentration risk

The five largest pools - Foundry USA, AntPool, F2Pool, ViaBTC and SpiderPool - found 77.16 % of the year's blocks, and their share moved less than 2 % all year. A static share cannot tell us whether hashrate is flowing towards those five or away from them, because it measures a level, not a direction.

BMM initially applied other indexes, the Herfindahl index among them, and the problem is that they were built for markets that are static by comparison. Bitcoin mining is more dynamic and has sector-specific conditions: hashrate moves between pools month by month, block luck dominates short windows, and difficulty resets change every pool's revenue at once. So BMMAA was built to analyse the periodic addition and attrition of hashrate flowing through the network, from the position of top-pool concentration risk, as a live indication of the network's direction. It compares the growth of the top five with the growth of every other pool over the same period and returns one number from −1 to +1: negative means hashrate is moving away from the largest pools, positive means it is moving towards them.

Read that way, the year is not flat at all. BMMAA stood at −0.48 on 14 January, a blue reading, with the rest of the market growing faster than the top five. It turned positive through the spring and reached +0.63 on 21 October, a red reading, the strongest concentration reading of the year: the top five were growing 10.3 % against a fall of 6.9 % for everyone else. By 11 November it had swung back to −0.34, blue again: over the last 30 days the top five added 43.6 EH/s (5.7 %) while every other pool added 32.2 EH/s (16.5 %).

Three readings, three directions, in a year when the top-five share barely moved. That is why the index is published next to the share: concentration in Bitcoin mining is not a level that drifts upwards, it is a flow that regularly reverses. The end of the year shows it again - as the AI pivot pulls hashrate away from the large pools, the score turns blue, and a reader who only sees 77 % would not know.

More on the method and the live value: BMMAA Pool Concentration Risk.

4 · Who mined what: what pools declare, and what the blocks show

Some ranking lists are built on what pools declare. This review is built on what the blocks show: the coinbase of every block from BMM's node, mapped to the pool that produced it. The two views do not agree, and the difference is the structure of the market.

Five pools are large enough to lead: Foundry USA 28.00 % of the year's blocks, AntPool 18.19 %, F2Pool 13.15 %, ViaBTC 9.27 % and SpiderPool 8.55 %. Behind them a second group follows with real but smaller shares: MARA 5.10 %, SecPool 4.32 %, Luxor 3.25 %, Ocean 2.37 %, Binance Pool 1.84 % (still in the 2026 statistic but shut in November), Braiins 1.55 % and SBICrypto 1.26 % (still in 2026 but shut in July). Then there is a group that declares hashrate on miningpoolstats and finds almost no block under its own name: proxy pools, which BMM classifies as hashrate aggregation pools, passing hashrate to a parent pool so their blocks are counted inside that parent and their operators appear in no ranking.

However there are two groups that are easy to miss and matter for the totals. NiceHash is a hashrate marketplace rather than a pool: it carries its own tag for part of the year and appears as an Ocean member for the rest.

The table below is the physical record: for each pool and each month, the number of blocks it found. Blocks are the only figure BMM observes directly; hashrate is derived from them and from the network difficulty of that month, so the table stays with what the chain shows.

The table excludes the proxy pools - hashrate aggregation pools - whose hashrate is counted inside the parent pool, and the solo miners and untagged blocks, which are reported separately. The unnamed hashrate row is the difference between the network and the pools above it: that is the hashrate this table cannot name. November counts to 11 complete days, so its column is an eleven-day estimate and not comparable with a full month.

Blocks found per month, BMM's own attribution, 1 January - 11 November 2026. A pool with no blocks in a month shows a dash, not zero, and the unnamed row is the blocks no pool name could be attached to. Proxy pools and solo miners are inside the network total, not in the pool rows.

5 · Year 2026 has large pool closures

Four pools left the market in 2026, and together they were not marginal. At their peak months they represented about 60 EH/s against a network that averaged 962.5 EH/s over the year, roughly 6 % of the network's hashrate. They found 1,568 blocks in the year, 3.48 % of all blocks, before they stopped.

The four are not the same case. Binance Pool announced on 13 October and stopped on 10 November; our own record has it quiet from 7 November, six days without a block after a month of decline, which we cannot distinguish from a suspension - the closure is the operator's statement, not our inference. f2pool.kz, the Kazakh project F2Pool opened in December 2023, closes on 21 November 2026. Ultimus Pool simply stopped finding blocks on 25 September, and BMM assesses it as a Binance proxy on the evidence of their 2022 partnership and the timing of the closure. SBI Crypto last found a block on 29 July, having been a top-ten pool at the start of the year.

Peak EH/s is the pool's highest monthly estimate on BMM's 7-day basis, not a company disclosure; network average 962.5 EH/s from the same method. 1,568 blocks are 3.48 % of the year's 45,153.

6 · Hashrate Aggregation Pools (Proxy Pools): declared is not mined

BMM prefers to call this group hashrate aggregation pools rather than proxy pools. Most are built around business development: a regional or miner-group focus, run by people who maintain client relations, while the technical stack and the FPPS payment burden are outsourced to the parent pool that receives the hashrate flow.

Together they declare about 125 EH/s (snapshot 1 November, miningpoolstats) and find almost no block under their own name; BMM's coinbase reading does not find them. Their blocks are counted inside the parent pool, and their operators appear in no ranking. Among the ones BMM tracks: EMCD, k8x, Neopool, Headframe, tpool, Kryptex Poool, Sigmapool, Placepool and miningpool.ru.

The group has a strong footprint in Russia and the CIS. Russia was and still is the second-largest market by hashrate after the United States, previously after China, and has just been through a regulatory restructuring, with licensing for local operators and the exit or repositioning of foreign ones. That is a story about where hashrate goes next, not a footnote: BMM will have more focused regional editions that carry it as its own section.

BMM expects some of these pools to phase into transition with a review of their own business model, since pool fees continue to decline and custody or yield products carry limits and risks. Also, Russia's position on regulating pools could trigger some pools with Russian hashrate to reformat their positioning.

7 · Hashrate and control: Stratum V2 and DATUM

The operational route to decentralisation is to spread hashrate across more pools. The technology route leaves the hashrate where it is and moves control to the miner, and as long as the template sits with the miner rather than the pool, it triggers a redistribution of hashrate by itself. Stratum V2 does that through job declaration: the miner builds the block template and declares it, and the pool validates and coordinates the work without dictating its contents. The pool still finds the block; it no longer decides what is in it.

DMND is the first pool to roll out a Stratum V2 pool with a job declarator, and it mined the first known block produced this way: block 955318 on 25 June 2026, for GoMining. BMM flagged a tag within one ViaBTC block in February 2026 that carried the GoMining name. The first SV2 block was a confirmation that the stack works in production. Interest has been growing, visibly in the Stratum V2 Working Group. On 11 November MARA signalled Stratum V2 for the first time, tagging block 975369 Sv2/MARA - one block out of the eight MARA found that day, which BMM reads as a successful trial until MARA announces its position on Stratum V2.

DATUM, run by Ocean, writes the miner into the coinbase, so every one of Ocean's 1,072 blocks in the year can be attributed to a named client - the only client-level record in this review. The 146 tags in the member file resolve to 69 member names, eleven of them carrying more than one tag: a table listed by tag would overstate the number of miners by more than double, so the table below is by name.

Ocean's history runs through it, with an operational and management reset during the BIP-110 period; the pool has been accelerating both its hashrate and the number of tagged clients since. The largest single member is Peak Mining, formerly Northern Data, with 262 blocks or 24.6 % of Ocean's annual flows; the top five are Peak Mining, SoV, NiceHash, Penguin and Sympatheia, together 57.3 %.

This is what pool transparency looks like when a pool chooses it, and it is also its limit: Ocean is 2.37 % of the network. A named-client table for Foundry, AntPool or F2Pool cannot be built from the coinbase, because those pools do not write the name - and more importantly because clients do not ask for it or are not aware it is an option.

This brings us back to MARA, which set up its own pool to control the template, the tag and transaction selection in a way compliant with the OFAC sanctions list back in March 2021. With the Stratum V2 test under way, BMM would not exclude MARA expanding the offer to other miners, with the template run for other US-listed miners in a branded, OFAC-compliant way.

Blocks per month, Ocean's own tags, 1 January - 10 November 2026: the table's 1,067 blocks are the days the member file covers, and Ocean found 1,072 to 11 November. Every Ocean block carries a member tag. All other named members expand on the web version.

8 · Solo mining: the sector's R&D sandbox

In BMM's own data, solo and mini-pool tags account for 44 blocks in 2026, 0.10 %, falling to almost nothing after July. The community's weight is in people and experimentation, not in hashrate.

The solo mining community is real, and it is the biggest group by number of people compared with the personnel deployed by industrial miners. A Bitaxe is 1.4 TH/s, but 20,000 Bitaxe owners is 20,000 people - nobody buys a fleet of them, they buy one for the home - and they will keep taking their chance at a block reward.

The sector also needs solo mining because it is self-driven R&D. The last cycle had R&D at industrial scale: immersion, hydro cooling, container solutions, overclocking, because capital was expensive and less accessible and efficiency was the only lever. This cycle capital was far more available to miners, until the AI pivot, and R&D fell out of focus. So some interesting developments and initiatives will come from solo mining.

The 20,000-strong Bitaxe army is still there, and with direct access to block templates it will be prompted to test, modernise and add features, as well as to try its luck at finding those blocks.

DMND has published, as part of its sandbox initiatives, practical steps for solo miners, including those running a Bitaxe, to construct their own block templates and start hunting for mining block rewards, with an option for merge mining RSK.

9 · Hashrate marketplace, not a pool, and a growing segment

NiceHash is the one participant in this review that is not a pool but a marketplace for buying and selling hashrate: miners point rigs at it and buyers rent that hashrate to point wherever they choose. Its blocks therefore appear under its own tag for part of the year and inside Ocean's tags for the rest.

In 2026 its own tag found 198 blocks, and Ocean's tags name it for another 104. The monthly pattern is the clearest evidence of routing rather than mining: 21 Ocean blocks in January, none in August and September during the BIP-110 turmoil when its own tag carried the hashrate, then 41 of its own in August and 60 in September, before returning to Ocean in October and November. A pool's share would move with those shifts; a marketplace's does not, because the hashrate never left.

The segment is getting more competitive. Braiins launched Hashpower in January 2026, selling hashrate directly; BitDeer ($BTDR) has long had a cloud-mining business; BitFuFu ($FUFU) launched a BTC-denominated hashrate token in October 2026 with Finloop. We should expect to see more developments in this segment going forward.

BMM's own tags and Ocean's member tags. A block count here measures the marketplace's throughput, not a pool's share.

10 · Pools and capital markets

The company figures are from the 30 October 2026 announcement; the pool figures are BMM's own.

BMM expected Foundry to push this transition first, but DCG's decision favoured Fortitude Mining's earlier spin-off development with a Zcash pool as the new venture (Cointelegraph, 2026). What the data does show is the pressure behind it: Foundry's share fell from 29.8 % in January to 24.6 % in November, between the AI pivot of its large accounts, regional limits on its operations and continued fee depletion. However, we might still see some developments there.

SpiderPool took the lead in the year's capital-markets story instead. On 30 October 2026 Spider Digital Technology and Starlink AI Acquisition Corporation (NYSE: OTAI), a SPAC, announced a definitive agreement that would take the mining-pool business public in the United States at a valuation of US$250 million for the merger consideration. Completion needs both shareholder votes, an effective Form F-4 and NYSE listing approval.

Our data sizes the pool independently of the filing: 3,853 blocks to 11 November, 12.2 a day, about 90 EH/s, and roughly 12,098 BTC of gross mining rewards in 2026. A 1-3 % pool fee would put revenue at $10-31 M a year. We will follow up when the deal is finalised and SpiderPool's final valuation is set.

11 · Reward destination: where the money goes after the pool

Once a pool has found the block, two questions remain: who is paid, and where the payment goes next. Foundry answers the first directly, which is why it is the only pool where a listed miner can be named from the chain (CleanSpark, Riot, American Bitcoin, Cipher Mining, Core Scientific, HIVE and Bitfarms).

HIVE Digital is the clearest case of a miner choosing its settlement rail. Paid by Foundry, it routes about $354 M into a single Coinsquare deposit address across six wallets. A Canadian miner liquidating through a Canadian exchange, with an OTC desk alongside it, is what the addresses show.

Every other pool we track routes rewards into a custodian, and one custodian is shared. Cobo receives from four pools, measured as the Cobo share of each wallet's outflow: SecPool 98.1 %, WhitePool 99.6 %, Braiins 94.3 % and CloverPool 88.7 %. No other pool in the dataset shows that pattern. The pools differ; the custodian does not. Cobo was built by people who came out of F2Pool and holds a strong position in the mining community. The outflow completes the picture: Cobo's hot wallet sent $3.90 B to AntPool and $0.83 B to F2Pool.

Another discovery, from the current 45,153 blocks: 94 carry no pool name at all, 0.21 %, spread over 71 days, the last on 7 November. Each is a single-block address with no repeat behaviour, the signature of a small or one-off operator - solo mining - rather than an unlabelled large pool. The largest single unexplained wallet, 3GdjWJdk…, moved 7.8 times its own block reward, which makes it a distributor rather than a miner; it accounts for 61 of the 94 blocks, which could also indicate part of some off-balance-sheet flows for a large pool, a miner or a capital allocator with exposure to mining.

Two further groups are named as actors but not as companies. An unidentified treasury received 1,323.80 BTC and executes through NYDIG, and a private operator with a silent coinbase ran at about 16 EH/s at its peak and swept its rewards to a hub that sent $306 M out. Neither can be called a pool, and neither can be allocated to one.

Third-party address labelling, retrieved 11 November 2026. The residual falls from 20 blocks a month in January to 2 in October while total blocks rose, which is consistent with one-off small and solo miners rather than a hidden pool. Coinbase appears mostly as a label on other parties' addresses, not as a counterparty of its own.

12 · What BMM will watch next: the pool business model

This brings us back to the business models of mining pools and the next phase of their development. The sector is intense and mathematically brutal at eliminating inefficiency, and pools will be pushed by miners' demands, by external factors and by technology. With AI's advance, managing security and client-fund vaults becomes a high-risk business; expect smaller players to abandon satellite custody and yield products, especially as TradFi enters with stablecoin-based yield that only large balance sheets can tolerate at scale. Hedging and structured products on BTC and USD can equally be offered by established OTC desks and banks.

The important development for miners is a clear and transparent on/off ramp to fiat in their region. Banking is the perennial constraint; a miner should have at least two or three rails at any time.

One part of product development has been almost entirely ignored: hedging the difficulty reset at each epoch. In October, difficulty rose 11.75 % while miner revenue fell 10.52 %, and BTC went from $83,565 on 1 October to $77,437 on 31 October. In November BTC moved from $77,437 on 1 November to $73,928 on 11 November, and the 13 November reset is estimated at about −8 %, which would add about 8.5 % back to miners' revenue. That is a vital part of a miner's liquidity management and it is overlooked. Do not expect large insurance underwriters to offer anything meaningful here soon: the sector is too marginal and too complex for them to price. The offering will come from inside the sector, and pools could explore it - but any such product needs a liquidity provider and market makers, as any commodity business does.

Template control allows for the miner to become the centre point for capital and financing structuring discussions - it allows financing providers to deal an FPPS-type offering directly with the miner rather than the pool. This will be a big trigger to decentralisation in the sector with the support of the new technology. Bitcoin mining is niche but a solid market with a backbone - hashrate is a commodity.