Timechain Wiki

Mining centralization concerns

16 min read

Bitcoin mining centralization is a recurring empirically-grounded critique across four dimensions: geographic concentration (post-2021 US-Russia-Kazakhstan after the Chinese mining ban), ASIC manufacturer concentration (Bitmain dominant, with MicroBT and Canaan as alternatives), pool concentration (the top 4-5 pools have controlled 75-90% of hashrate across most of Bitcoin's history), and institutional/public-company mining (Marathon, Riot, CleanSpark, others). Each dimension enables a distinct attack vector — regulatory pressure on concentrated geography, supply-chain attacks via manufacturers, coordination attacks via pools, and regulatory capture via institutional mining. The principled response: hashrate concentration is more visible than effective concentration (pool-level miners can defect), 17 years show no successful censorship attack despite recurring centralization, and decentralization initiatives (Stratum V2 with template-negotiation, Braidpool, OCEAN) reduce pool-level coordination capability. The critique has empirical merit; mitigations are partial; the trajectory is mixed.

Why this note matters

Paired controversy note (cross-section): the event-level controversy about Foundry USA’s persistent dominance and the AI-infrastructure-pivot dynamics lives in Mining pool centralization and the AI infrastructure pivot. This note engages the analytical structural-concentration concern.

Mining centralization is the most empirically-grounded critique of Bitcoin’s decentralization claims. Unlike abstract economic critiques, it operates on observable measurements — hashrate distribution, manufacturer market share, pool concentration, geographic distribution — that change visibly over time. The note catalogues the specific dimensions, each with different mechanisms and mitigation trajectories; engages the empirical record using observable metrics rather than rhetorical claims; surfaces the interactions among dimensions (geographic and pool concentration compound; institutional ownership and ASIC concentration interact); and articulates the realistic threat models — censorship via regulatory pressure, consensus attacks via pool coordination, supply-chain attacks via manufacturer compromise.

The defensible position: track centralization metrics actively, recognize that censorship resistance depends on structural conditions that aren’t permanent, and support practical decentralization initiatives (Stratum V2 adoption; pool diversity; geographic diversification) that reduce risk without centralized intervention.

The critique landscape

Four distinct centralization concerns deserve treatment.

Geographic centralization

The empirical pattern:

  • Pre-2021: China dominated Bitcoin mining; estimates ranged from 50-80% of global hashrate concentrated in Chinese provinces (Sichuan, Xinjiang, Inner Mongolia, others). The May-September 2021 Chinese mining ban produced the largest hashrate redistribution in Bitcoin’s history.
  • 2021-2024 transition: Chinese miners migrated to the US (~37-40% of global hashrate post-migration), Kazakhstan, Russia, Canada, and various other jurisdictions. Texas specifically emerged as a major US mining hub due to ERCOT grid characteristics.
  • 2024-2026 picture: US share has grown to approximately 40-45% of global hashrate; Russia, Kazakhstan, and Canada collectively control another 25-35%; the remaining hashrate is distributed across smaller geographies (Bhutan, Paraguay, Iran, Argentina, Norway, others).

The critique: regulatory pressure on a single jurisdiction with substantial hashrate share could compel coordinated censorship behaviour. A US ban on certain transaction types — or coordinated regulatory action affecting US-based mining operations — could in principle force significant hashrate share toward compliance.

Mitigations: geographic concentration is more diversified than the pre-2021 China era; pool-level operations span jurisdictions; mining migration is operationally feasible (demonstrated by the 2021 transition); the regulatory pressure threshold for forced coordinated censorship is high.

ASIC manufacturer concentration

The empirical pattern:

  • Bitmain dominates ASIC manufacturing, with the S19, S21, and S23 product lines representing approximately 60-75% of Bitcoin mining hardware shipped in 2020-2026
  • MicroBT (WhatsMiner) is the principal alternative, with approximately 15-25% market share
  • Canaan and several smaller manufacturers compete for the remaining ~5-15%
  • Intel entered the ASIC market in 2022 and exited in 2023, illustrating the manufacturer-base instability
  • The supply chain: Bitmain’s manufacturing is concentrated in Taiwan (TSMC fabrication); MicroBT similarly Taiwan-fabricated; supply-chain capacity is concentrated

The critique: a small ASIC-manufacturer base creates supply-chain attack vectors. A compromised manufacturer could:

  • Ship hardware with backdoors that enable specific firmware-level attacks
  • Selectively delay shipments to disfavoured customers
  • Coordinate firmware updates that enable specific censorship behaviour
  • Be coerced or compelled by jurisdictional regulators

Mitigations: open-source mining-firmware initiatives (Braiins OS; the BraidsOS+ project; Stratum V2 with template-negotiation); FPGA-based mining alternatives; the operational complexity of supply-chain attacks at scale.

Pool concentration

The empirical pattern:

  • The top 4-5 pools have controlled approximately 75-90% of Bitcoin hashrate across most of Bitcoin’s history
  • 2026 distribution: Foundry USA (~25-30%), AntPool (~15-20%), F2Pool (~10-15%), Binance Pool, ViaBTC, MARA Pool, others
  • Foundry USA’s dominance post-2021 is particularly notable; it operates in the US regulatory environment and has been the subject of specific concerns about transaction filtering

The critique: pools sequence transactions in the blocks they mine. A pool that filters transactions (refuses to include OFAC-sanctioned addresses, or specific Ordinals-related transactions, or other categories) can implement censorship without protocol-level violation. Coordinated filtering across the top 3-4 pools would effectively censor transactions at the network level.

Mitigations:

  • Stratum V2 with template-negotiation allows individual miners to construct their own transaction templates rather than accepting pool-constructed blocks; defection cost is low
  • Decentralized mining pools (Braidpool, OCEAN) reduce pool-operator coordination capability
  • Empirical track record: no successful coordinated pool censorship of Bitcoin transactions despite recurring concentration
  • Individual miner defection is profitable; a censored transaction provides extra fees to whatever pool includes it

Institutional and public-company mining

The empirical pattern:

  • Public-equity-financed mining-corporate sector grew materially post-2020: Marathon Digital, Riot Platforms, CleanSpark, Core Scientific, Hut 8, Bitfarms, others
  • Together these companies operate approximately 40-45% of global hashrate as of 2026 (energized-hashrate basis) — the highest public-miner share on record, up from ~25% two years earlier
  • Many operate primarily in the US under regulatory oversight
  • Several have been subject to specific regulatory scrutiny (SEC, environmental regulators, energy regulators)

The critique: publicly-traded mining companies are subject to regulatory pressure at multiple layers — securities regulators, environmental regulators, banking regulators that finance the operations. This creates structural pressure for compliance with whatever regulatory framework emerges; mining companies cannot easily defect because they have institutional shareholders and stock-market reputations.

Mitigations: institutional mining is offset by significant private-mining operations (the Russian-Kazakh share is mostly private); the regulatory landscape is currently favourable for Bitcoin mining in the US (post-2024 administration positions; the Strategic Bitcoin Reserve); institutional mining diversifies into multiple political jurisdictions.

Key proponents

The critique is advanced by diverse voices:

  • Academic researchers publishing on hashrate distribution and pool concentration (various Cambridge Center for Alternative Finance papers; academic blockchain research; the work tracking specifically Foundry’s market share growth post-2021)
  • Some Bitcoin developers and analysts raising concerns: Hasu, Nic Carter, Pierre Rochard, others have engaged the question with nuance
  • Critics including Frances Coppola, David Gerard, Molly White — frequently cite mining centralization
  • Greenpeace and environmental activists — engage centralization alongside energy critiques
  • Various regulatory commentators — including Treasury Department, OFAC, banking regulators concerned with transaction censorship capabilities
  • Within-Bitcoin advocates — many Bitcoin developers and miners themselves raise concerns about specific concentration trends and support decentralization initiatives

The critique is one of the few where mainstream-critic positions and within-Bitcoin-thinker positions substantially converge.

What’s right about the critique

Several empirical points are well-established:

Centralization metrics show real concentration. Hashrate distribution, ASIC manufacturer market share, pool concentration, and geographic distribution all show concrete numbers consistent with the critique. This is not a rhetorical observation.

Specific incidents demonstrate censorship capability. In 2021-2024, several US-regulated mining pools demonstrated transaction-filtering behaviour:

  • OFAC-sanctioned-address filtering by some US pools (limited; not network-wide; not consistently enforced)
  • Ordinals/Inscriptions filtering debates in 2023-2024 (some pools experimented with filtering; community pushback resulted in mostly-unfiltered policies)
  • MicroStrategy / institutional-corporate-account filtering has been raised as a theoretical concern; no widespread practice

The empirical capability exists; the practice has been bounded by community pressure and economic incentives but not by protocol-level prevention.

The trajectory is mixed. Some centralization metrics are improving (Stratum V2 adoption growing; geographic distribution broader than pre-2021); some are worsening (Foundry USA’s continued dominance; ASIC manufacturer concentration; institutional-mining growth). The direction is not uniformly favourable.

Compounding effects. Geographic centralization, pool concentration, and institutional ownership compound. A coordinated regulatory action affecting US-based mining could in principle reach 40%+ of global hashrate; combined with ASIC supply-chain leverage and pool coordination, the practical concentration is higher than any single metric suggests.

Empirical track record is limited. Bitcoin has operated for 17 years; the post-2021 centralization pattern is only 5 years old. The future-trajectory is uncertain.

The Bitcoin-side response

The response operates on several layers.

Empirical resilience since 2009

Bitcoin has experienced multiple periods of substantial centralization (pre-2021 China dominance; periodic 60-70% pool-concentration peaks) without coordinated censorship attacks. The empirical evidence is favourable to “the system resists centralization-based attacks even when centralized.”

This evidence is informative but not conclusive — past patterns may not generalize to future conditions.

Pool concentration is more visible than effective

A key distinction: hashrate participation in a pool ≠ alignment with pool censorship policy. Individual miners hash for pools because of the variance-smoothing benefit (more predictable revenue than solo mining). They are economically motivated to defect from any pool that begins systematic censorship — defection to a non-censoring pool is operationally trivial.

Stratum V2 with template-negotiation strengthens this defence: individual miners can construct their own block templates (including censored transactions) while still hashing for a pool’s variance-smoothing service. Adoption is ongoing; deployment matters.

The implication: pool-level concentration of 80% does not produce 80% effective censorship capability. The effective censorship requires miners to cooperate with their pools’ filtering policies, which is economically irrational.

Mining-decentralization initiatives

Several practical decentralization initiatives are active in 2026:

  • Stratum V2 — the protocol upgrade enabling miner-level template negotiation; adoption growing but partial
  • Braidpool — a Drivechains-related decentralized mining proposal; experimental
  • OCEAN — Jack Dorsey-backed decentralized mining initiative; operational but small market share
  • DEMAND — decentralized mining pool proposal
  • Solo mining at scale — various initiatives encouraging solo mining for variance-tolerant operators
  • Open-source mining firmware (Braiins OS+, Vnish) — reduces firmware-level supply-chain attack surface

These initiatives don’t eliminate centralization but they reduce its practical implications.

Regulatory-pressure scenarios are bounded

A coordinated regulatory attack on US-based mining would face several challenges:

  • Mining migration is operationally feasible (demonstrated by 2021 China transition; would require time but not infeasible)
  • Private mining and non-US jurisdictions would resist coordinated censorship; the network would split rather than universally comply
  • The Bitcoin community has mechanisms for emergency response (UASF historical precedent; emergency soft-forks; community coordination through technical channels)
  • The political-coordination cost of a state-led censorship attack is significant; the US would face diplomatic and economic consequences

Manufacturer concentration is less concerning than hashrate concentration

ASIC manufacturer concentration creates supply-chain attack capability but not direct censorship capability. A compromised manufacturer:

  • Cannot censor blockchain transactions directly
  • Cannot easily produce backdoor firmware that goes undetected (independent firmware audits; open-source firmware alternatives)
  • Cannot easily compel miners to accept compromised firmware (miners control firmware updates)

The risk is real but bounded.

The institutional-mining counter-balancing dynamic

Public-equity-financed mining operations are subject to regulatory pressure, but they are also subject to fiduciary duty to shareholders. A regulatory mandate that significantly hurts mining economics would face shareholder pushback; a mandate that helps Bitcoin’s price and adoption (per current US policy direction) is consistent with mining-company interests.

The 2026 picture has US regulatory framework favourable to Bitcoin mining; the historical pattern (regulatory uncertainty 2017-2021; favourable 2024-2026) suggests political-cycle dynamics rather than secular pressure.

Counter-arguments and tensions

”Foundry USA’s dominance is a real and growing problem”

The tension: Foundry USA’s market share has been notably persistent across 2021-2026; it operates in a regulated US jurisdiction and is subject to specific regulatory pressure. The historical pool-rotation pattern (where dominant pools tend to lose share over time) may not apply to Foundry given its institutional structure.

Response: Real concern. Foundry’s persistent dominance is the most concrete current centralization concern. Mitigations: (1) Stratum V2 adoption gives Foundry’s miners the ability to defect at the template level without leaving the pool; (2) several Foundry-adjacent miners participate in non-Foundry pools simultaneously; (3) regulatory pressure on Foundry would need to overcome significant shareholder and political resistance to translate into network-wide censorship. But the concentration trend deserves active tracking.

”Stratum V2 adoption has been slow”

The tension: Stratum V2 was published in 2018-2020; adoption in 2026 remains partial. Pool operators have economic incentives to delay adoption (template-negotiation reduces their economic power). The mitigation might be theoretical rather than operational.

Response: Partially valid. Stratum V2 adoption has been gradual; some pools have integrated it, others have not. The trend is positive but slow. The 2025-2027 adoption window will be informative — if adoption stalls, the centralization concern strengthens.

”The 2021 China migration set a precedent but doesn’t generalize”

The tension: The 2021 transition succeeded because Chinese miners could move to friendly jurisdictions. A coordinated regulatory pressure across multiple major jurisdictions simultaneously could prevent migration. The empirical resilience may not repeat.

Response: Valid. The 2021 transition relied on jurisdictional diversity that may not always be available. A scenario where the US, EU, China, and other major jurisdictions coordinated regulatory pressure on Bitcoin mining simultaneously is harder to imagine in 2026 (the regulatory landscape is more favourable than 2017-2021) but cannot be ruled out at longer horizons.

”ASIC manufacturer concentration is structurally hard to fix”

The tension: The ASIC manufacturer landscape is unlikely to materially diversify; specialized-hardware design economics favour consolidation. Bitmain’s dominance is likely persistent. Open-source firmware doesn’t address the hardware-level supply chain.

Response: Real concern. ASIC manufacturer concentration is structurally persistent. Mitigations are partial: open-source firmware reduces firmware-level attack surface; manufacturer geographic diversification (MicroBT vs Bitmain) provides limited diversification; FPGA-based alternatives are not economically competitive. The risk is bounded but real.

”Institutional mining could capture protocol governance”

The tension: Public-company miners with shareholder duties have economic incentives to support specific protocol decisions that benefit their business models. Concentrated institutional mining could influence protocol-evolution debates (see Protocol-evolution constraints) in ways that prioritize mining-business interests over broader user interests.

Response: Possible but bounded. Protocol decisions are not made by miners alone — the broader Bitcoin community (developers, node operators, users) participates. Past contentious upgrades (Block Size Wars; Taproot deployment) demonstrated that miner-dominated coalitions don’t always prevail. The institutional-mining political capture is a real concern at multi-decade horizons but is not currently visible at significant scale.

”The empirical track record doesn’t prove the future is safe”

The tension: 17 years of no successful censorship is informative but not predictive. Conditions evolve; adversaries adapt; regulatory environments shift. The defence-by-historical-track-record argument has limits.

Response: Valid. Track-record evidence provides directional confidence; it does not prove future resilience. Active monitoring of centralization metrics, Stratum V2 adoption, and regulatory developments is appropriate.

Verdict: Real empirical concern; mitigations partial; trajectory mixed; active tracking is appropriate

Mining centralization is a serious and empirically-grounded critique. Each of the four dimensions (geographic, manufacturer, pool, institutional) has real concentration; the dimensions interact and compound; the realistic threat models are non-trivial.

A serious assessment:

  • Current state: Mining centralization is substantial but not catastrophic. Bitcoin’s censorship resistance has held empirically since 2009 despite recurring centralization. Specific incidents (Foundry filtering experiments; 2024 Ordinals debates) have been bounded by community and economic pressure.
  • Trajectory: Mixed. Some metrics improving (Stratum V2 adoption; geographic broader than pre-2021); some worsening (Foundry persistent dominance; institutional-mining growth).
  • Mitigations: Real but partial. Stratum V2; decentralized mining initiatives; open-source firmware; geographic-diversification. Each reduces risk; none eliminates it.
  • Threat models: State-actor scenarios; coordinated pool censorship; supply-chain compromise. Each is bounded by current conditions but not eliminated.

This is a critique worth tracking actively. The 2025-2027 Stratum V2 adoption window and the trajectory of Foundry USA’s market share are the most informative near-term variables.

Open questions for further development

  • What is the right composite metric for “effective mining centralization” that accounts for hashrate distribution, pool template-negotiation adoption, geographic diversification, and institutional ownership? Current metrics emphasize raw hashrate distribution but understate effective decentralization.
  • The Foundry USA dominance trajectory is the most concrete current concern. What conditions would lead to Foundry’s market-share decline — competitive pressure from Stratum V2-enabled alternatives, regulatory pressure, customer defection, or something else?
  • Stratum V2 adoption is the single most important practical mitigation. What’s the realistic adoption curve, and what are the gating constraints?
  • The interaction between mining centralization and the long-term security budget (see Long-term security budget) is unstudied. Both critiques compound as Bitcoin matures; their joint trajectory matters more than each individually.
  • The 2028-2032 halving cycle will produce significant miner-economics pressure. How does that pressure interact with centralization dynamics — does it accelerate consolidation (smaller miners exit) or decentralize (institutional miners face profitability stress)?

Canonical sources for this note

Empirical data sources:

  • Cambridge Center for Alternative Finance — Cambridge Bitcoin Electricity Consumption Index and adjacent mining-distribution reports
  • Hashrate Index — pool concentration tracking
  • Various mining-pool reporting (Foundry USA, AntPool, F2Pool, others publish hashrate-share data)
  • Public-company mining filings (Marathon, Riot, CleanSpark SEC filings)
  • ChainArmor and similar analytics services tracking censorship behavior

Academic and analytical:

  • Various academic papers on Bitcoin mining centralization (Cambridge CCAF; academic blockchain research; specifically work by Holub, Johnson, Romiti, others)
  • BitMEX Research analyses of mining decentralization
  • Hasu — various essays on mining centralization and pool dynamics

Bitcoin-side engagements:

  • Carter, Nic — various essays on the mining centralization landscape
  • Lopp, Jameson — practitioner perspective on centralization risks. See Jameson Lopp.
  • Antonopoulos, Andreas — Mastering Bitcoin coverage of mining and pools
  • Various within-Bitcoin developer mailing-list discussions on Stratum V2 and decentralization initiatives

Mining-decentralization initiatives:

  • Stratum V2 specification and adoption tracking (stratumv2.com)
  • Braidpool project documentation
  • OCEAN (Ocean.xyz) project
  • DEMAND pool project
  • Braiins OS+ open-source mining firmware

Specific incidents:

  • Foundry USA Ordinals-filtering experiments (2023-2024)
  • OFAC-sanctioned-address filtering by US-regulated pools
  • The 2024 MARA Pool transparency disclosures
  • Pre-2021 China-mining-ban migration data

As of 2026-05-15: Foundry USA continues to dominate US-based mining; Stratum V2 adoption is growing; geographic distribution is broader than the pre-2021 era; institutional-mining growth continues.

Within the Criticisms section:

Mining section:

These mining-section notes are the natural homes for the operational mining treatment; this criticism note treats the centralization-as-vulnerability framing.

Technical foundations section:

Adjacent thinker pages:

The sub-MOC home:

Graph View

Backlinks