Three Bitcoin mining pools now control more than half the network's block production. As of August 2026, Bitcoin's Nakamoto coefficient — the minimum number of pools required to exceed 50% of mined blocks — has fallen to 3. The Herfindahl-Hirschman Index (HHI) for pool concentration sits at appro...
"Block-template decentralization — Stratum V2 job declaration and OCEAN's DATUM — is the structural fix that raises the [Nakamoto] coefficient without changing who owns the hashrate." — D-Central, State of Bitcoin Mining H1 2026
Three Bitcoin mining pools now control more than half the network's block production. As of August 2026, Bitcoin's Nakamoto coefficient — the minimum number of pools required to exceed 50% of mined blocks — has fallen to 3. The Herfindahl-Hirschman Index (HHI) for pool concentration sits at approximately 1,492, bordering the U.S. Department of Justice threshold between "unconcentrated" and "moderately concentrated" markets.
The concentration is structural, not incidental. Post-halving margin compression, a 14% decline in mining difficulty from its November 2025 all-time high of 155.97 trillion, and widespread capital reallocation toward AI infrastructure have thinned the field. Smaller pools lack the economics to compete. SBI Crypto's July 31 shutdown — removing 20.9 EH/s (2.2% of global hashrate) — merely accelerated a trend years in the making.
The industry's proposed remedy, Stratum V2, gained commitments from pools representing 75% of hashrate in May 2026. But commitments and deployment differ. Only two pools run Stratum V2 natively in production. The gap between announced intent and operational reality defines Bitcoin mining's centralization problem in 2026.
As of the 7-day trailing average ending August 7, 2026, Bitcoin's pool distribution reads:
| Pool | Hashrate Share | Operator Region | |------|---------------|----------------| | Foundry USA | 25.5% | United States (DCG) | | F2Pool | 18.0% | China/Global | | AntPool | 17.0% | China (Bitmain) | | SpiderPool | 9.0% | China | | ViaBTC | 7.5% | China | | Top 5 Total | 77.0% | |
The top four pools command over 70% of global hashrate. Foundry USA alone coordinates roughly one-quarter of the network's block production. The combined share of Foundry, AntPool, and F2Pool briefly touched 60% on July 31, the day SBI Crypto exited.
Network hashrate stands near 942 EH/s, according to CoinWarz data as of August 7, 2026. After briefly crossing 1 ZH/s (1,000 EH/s) in January 2026, the network pulled back as miners curtailed operations in response to falling revenue.
Mining difficulty sits at 126.23 trillion — 14% below the 2026 high and 19.1% below the November 2025 all-time high of 155.97 trillion. This is the second time in Bitcoin's history that difficulty has fallen below year-prior levels.
Pool concentration does not equate to entity concentration in mining hardware. Foundry USA coordinates hashrate from hundreds of independent mining operations. Miners point machines at pools; they do not surrender ownership. The distinction matters, but not as much as proponents argue.
Under the dominant Stratum V1 protocol, pools — not individual miners — construct block templates. This means pool operators decide which transactions to include, which to exclude, and in what order. The entity running a pool with 25% of hashrate effectively has 25% control over Bitcoin's transaction selection.
Foundry is a wholly owned subsidiary of Digital Currency Group (DCG), one of the industry's largest conglomerates. DCG's portfolio includes Grayscale, Genesis, CoinDesk, and investments in over 160 crypto companies. AntPool is operated by Bitmain, the dominant manufacturer of ASIC mining hardware. These are not neutral infrastructure providers; they are vertically integrated conglomerates with strategic interests in the broader crypto ecosystem.
The structural risk: if two of the top three pools colluded or a single pool were compromised, the coalition could theoretically execute selective transaction censorship or, in a worst case, a double-spend attack. No such incident has occurred. The concern is architectural, not historical.
The April 2024 halving cut Bitcoin's block subsidy from 6.25 BTC to 3.125 BTC. The revenue reduction hit with a lag. By Q1 2026, the consequences were measurable:
These economics favor scale. Large operators with access to cheap power, institutional financing, and next-generation hardware survive. Everyone else faces a choice: consolidate into larger pools for better revenue smoothing, pivot to AI hosting, or shut down.
The top public miners reflect this consolidation. As of August 2026, Bitdeer Technologies operates 73.0 EH/s, MARA Holdings runs 70.7 EH/s, and CleanSpark manages 42.6 EH/s. These three firms alone represent roughly 20% of global hashrate. Meanwhile, MARA posted Q2 2026 revenue of $174.9 million, down 27% year-over-year; CleanSpark reported $138.0 million, down 30.5%.
Over $70 billion in cumulative AI and high-performance computing contracts have been announced across the public mining sector, according to CoinShares. Miners are not abandoning Bitcoin mining — they are hedging against its economics.
On July 31, 2026, SBI Crypto permanently closed its public Bitcoin mining pool. The pool stopped accepting mining shares at 07:00 JST (22:00 UTC July 30). No public explanation was provided.
At its peak, SBI Crypto's pool operated approximately 20.9 EH/s, roughly 2.2% of global hashrate and the 11th-largest active pool. In its final month, the pool's seven-day average hashrate had already fallen 64%.
The timing was notable. One week before the shutdown announcement, SBI Holdings disclosed a ¥46.7 billion ($289 million) agreement to acquire Bitbank, one of Japan's largest cryptocurrency exchanges. SBI narrowed its crypto focus to regulated exchange services while exiting the commoditized mining pool business.
The 20.9 EH/s displaced by SBI's closure flowed predominantly to the largest existing pools. On the day of shutdown, Foundry, AntPool, and F2Pool collectively reached 60% of recently mined blocks — a concentration level that would concern antitrust regulators in any conventional market.
The mining pool market in 2026 has split into two tiers, as documented by CryptoSlate. Tier 1 pools — Foundry, AntPool, F2Pool — have reoriented around institutional clients. They offer custom firmware support, dedicated account management, and preferential fee structures for operations deploying thousands of machines.
Independent and mid-size miners report declining service quality from top-tier pools: slower support response times, generic tooling, and fee structures that penalize smaller deployments. These miners are migrating to mid-tier pools that offer responsiveness and flexibility, but at the cost of higher variance in block rewards due to lower aggregate hashrate.
The economics reinforce the split. At a hash price near $30 per PH/s per day, any machine less efficient than a Bitmain S19 XP running on power priced at $0.06/kWh or higher is losing money. Home mining on residential electricity is, according to multiple analyses, a losing proposition in most U.S. markets.
The structural remedy to pool centralization is not breaking up pools — it is removing their control over block template construction.
Stratum V2 is an open protocol that allows miners, rather than pool operators, to select which transactions go into blocks. On May 11, 2026, seven pools representing approximately 75% of global hashrate — Foundry, AntPool, F2Pool, SpiderPool, MARA Pool, Block Inc., and DMND — joined the Stratum V2 Working Group. The commitment was the most significant signal of adoption since the protocol's specification was finalized.
As of August 2026, however, only two pools run Stratum V2 natively in production: Braiins Pool and DEMAND Pool (launched November 2025). The remaining signatories are in testing and integration phases. By year-end 2026, Stratum V2 is projected to reach 40-60% of network hashrate for basic protocol adoption, though usage of the job negotiation feature — the component that actually decentralizes template construction — is expected to remain below 10%.
OCEAN's DATUM protocol takes a different approach. Founded by long-time Bitcoin Core developer Luke Dashjr and backed by a $6.2 million seed round led by Jack Dorsey, OCEAN runs standard Stratum V1 to ASIC hardware while requiring miners to build block templates from a local full node. OCEAN currently commands roughly 2% of global hashrate. In April 2025, Tether announced it would deploy existing and future hashrate on OCEAN, signaling institutional interest in decentralized template construction.
Both solutions address the same problem: under Stratum V1, a pool with 25% of hashrate has 25% control over transaction ordering. Under Stratum V2 with job negotiation or DATUM, that pool coordinates reward distribution but not transaction selection. The Nakamoto coefficient for transaction censorship rises even if the hashrate distribution remains identical.
Bitcoin's mining centralization in August 2026 is not at crisis levels by the network's own historical standards — four pools controlled 75% of hashrate as recently as 2014. The difference in 2026 is scale. The network processes $10+ billion in daily settlement value. The economic incentives to maintain — or exploit — concentrated control are proportionally larger.
The HHI of 1,492 sits just below the DOJ's "moderately concentrated" threshold of 1,500. A single mid-tier pool shutdown could push the index above the line. The Nakamoto coefficient of 3 means that the minimum coalition for a 51% attack is three entities — three companies that would need to coordinate. This is a narrow margin for a network securing hundreds of billions in value.
The Stratum V2 commitments are encouraging but insufficient. Until job negotiation is deployed in production at scale, the protocol fix remains theoretical. The gap between signing a working group membership and running decentralized block construction in production is measured in engineering quarters, not press releases.
Bitcoin's mining pool concentration in August 2026 is the predictable result of post-halving economics compressing margins beyond the tolerance of smaller operators. The network is not broken — blocks are produced every 10 minutes, transactions clear, and no censorship has been documented. But the margin of decentralization has thinned.
Three pools controlling 50%+ of block production is an architectural vulnerability, not a crisis. The distinction depends entirely on whether structural remedies — Stratum V2 job negotiation and DATUM — move from commitment to production before the next concentration event. The data shows commitments are ahead of deployment. Whether deployment catches up before the market structure makes it irrelevant is the open question.