Bitcoin Energy Use Explained: How Much Power Mining Consumes and Why It Matters

Bitcoin's annualized electricity consumption sits near 138 TWh in the most recent transparent estimate (Cambridge Digital Mining Industry Report, April 28, 2025); the mix is 52.4% sustainable (42.6% renewables plus 9.8% nuclear), 38.2% natural gas, and 8.9% coal (source: Cambridge CJBS). Bitcoin uses real energy by design; the honest questions are how much, from what sources, on what trajectory, and whether the security budget is worth the spend.

Bitcoin energy use at a glance, 2026 specifics:

1. The Cambridge CCAF Bitcoin Electricity Consumption Index runs ~100-180 TWh annualised across low-central-high scenarios; the April 28, 2025 Cambridge Digital Mining Industry Report puts the central at 138 TWh, ~0.5% of global electricity (source: Cambridge CCAF)

2. Sustainable share is 52.4% per the April 2025 Cambridge study: 42.6% renewables (hydro 23.4%, wind 15.4%, solar 3.2%, other 0.6%) plus 9.8% nuclear; natural gas at 38.2% has replaced coal (8.9%) as the largest single source

3. Geographic distribution per Hashrate Index Q2 2026: US 37.4%, Russia 16.9%, China 12%, Paraguay 4.3%, UAE 3%, Oman 3%, Canada 2.6%, Ethiopia 2.5%, Kazakhstan 1.8%, Indonesia 1.8%; top three concentrate ~65% of 1,004 EH/s 30-day moving average (source: Hashrate Index)

4. Country / sector comparators: Argentina ~115 TWh (2023), Norway ~140 TWh (Statnett 2022), global gold mining ~240 TWh, global data centres 415 TWh in 2024 projected to 945 TWh by 2030 driven by AI (source: IEA)

5. ASIC efficiency improved ~7x over eight years from 98 J/TH (2018) to 12-13.5 J/TH on 2026 leading hardware (Antminer S21 XP at 13.5 J/TH; S21 XP Hyd at 12 J/TH); each generation cuts J/TH 20-30%, slowing as fabrication approaches 3-5nm

This guide explains Bitcoin energy honestly: why proof of work spends electricity by design, the actual 2026 numbers, the mix by source, geographic shifts post-China-2021, stranded-energy and flared-gas cases (Crusoe + Exxon-Bakken pilot), ERCOT demand response, the per-transaction-energy framing problem, the Mora 2018 study and Houy reply, the Greenpeace "Change The Code" 2022-2024 campaign, and the honest answer on whether the spend is worth it. For protocol context, see how proof of work secures Bitcoin and how Bitcoin mining works at the system level.

Why proof of work spends energy by design

Bitcoin's energy use is structural, not accidental. To add a block, a miner repeatedly hashes a candidate header until the result falls below a difficulty target. Each attempt consumes electricity. With the network at ~1,004 EH/s 30-day moving average in Q2 2026, miners collectively try ~10^21 hashes every ten minutes Hashrate Index.

The energy cost is the security mechanism. To rewrite Bitcoin's confirmed history an attacker must outpace the honest network's cumulative work, burning at least as much electricity as honest miners over the same window. Cheap block production means cheap history rewriting. The Bitcoin whitepaper's Section 11 derives this directly via Poisson catch-up probability (source: Bitcoin.org); the broader rationale appears in the Bitcoin developer reference (source: Bitcoin Dev Docs).

For the attack-cost arithmetic this enables, see what a 51% attack means for Bitcoin security; the $1.31M-per-hour 2026 attack-cost figure exists precisely because real electricity backs every confirmed block.

How much energy Bitcoin uses, and the source-by-source breakdown

The most cited authoritative model is the Cambridge Bitcoin Electricity Consumption Index (CBECI) at the Cambridge Centre for Alternative Finance, which publishes a low-central-high range using a transparent methodology layered over reported hashrate, ASIC efficiency curves, and assumed utilization Ccaf.

Source: Cambridge Judge Business School Cambridge CJBS; CBECI methodology (source: Cambridge CCAF); EIA Today In Energy (source: EIA).

The 138 TWh equals ~0.5% of global electricity consumption. Network-wide emissions in the same Cambridge study sit at 39.8 MtCO2e per year. Both are model estimates from a 49-firm survey covering 48% of global mining activity, extrapolated to the full network: the most rigorous public figures available, not direct meter readings.

The energy-source mix per the April 2025 Cambridge study:

Source: Cambridge Digital Mining Industry Report April 28, 2025 Cambridge CJBS.

Two findings warrant emphasis. First, sustainable share crossed 50% on a transparent third-party survey for the first time. Second, the coal-to-gas substitution is the dominant fossil-side shift: coal halved-and-then-some, gas filled the gap. Both patterns track the post-China-2021 migration to North American grids where gas is abundant and coal-grid mining lost share.

For how Bitcoin mining interacts with power grids, source mix follows where ASICs land, and ASICs follow electricity prices.

Country and sector comparisons that hold up

Comparisons to nation-states or other industries can develop intuition about scale, but only if the scope on each side is honestly stated. The table below uses the latest available figures with date stamps, drawing from IEA, national grid operators, and recent peer-reviewed mining-energy literature.

Bitcoin at 138 TWh is roughly the same order as Norway or Argentina's national consumption, about half of global gold mining, about a third of global data centres (themselves projected to more than double by 2030 driven mainly by AI), and ~0.5% of global electricity. Comparisons buy scale intuition, not equivalence; Bitcoin is one global industry, Argentina is a country of 47 million residents, and gold mining spans hundreds of operations across continents.

For why Bitcoin's 21-million supply cap matters distinct from energy, the value secured by 138 TWh is roughly $1-2 trillion in network capitalization plus the option value of an open monetary network. Those are what the energy spent buys.

Per-transaction energy is the wrong framing

A common comparison divides total Bitcoin electricity use by transaction count to produce a startling per-transaction figure (often quoted in the hundreds or low thousands of kWh). This calculation is mathematically valid but conceptually misleading.

Bitcoin spends roughly the same energy at 250,000 daily transactions or 500,000. Hashrate is a function of price-times-block-reward economics, not transaction volume. The meaningful unit is per-block-time energy: ~138 TWh a year buys ~52,560 ten-minute blocks plus all confirmed history.

For how Lightning settles thousands of payments under one on-chain transaction, the per-tx denominator is an artifact of where you draw the boundary.

Geographic shifts: Post-China-2021 dispersal and the 2026 map

The single largest event in Bitcoin energy geography was China's mining ban in May-June 2021, which forced ~50% of global hashrate to relocate within months. Five years on, the resulting distribution is more dispersed than the pre-2021 China-dominant concentration.

Source: Hashrate Index Q2 2026 heatmap Hashrate Index. Top three control ~65% of the 1,004 EH/s 30-day moving average.

Geography drives the energy mix: Paraguay and Ethiopia weigh renewables up via hydro; Russia and the Middle East raise gas share. The dispersal means a single country shock no longer takes ~50% of hashrate offline; China's 2021 ban dropped ~30-50% within months, while an equivalent event in any 2026 country would shave at most ~15-20%.

Stranded energy, flared gas, and grid balancing in the real world

Three claims appear in every Bitcoin energy debate: stranded energy, methane flaring mitigation, and grid balancing via demand response. Each is real in specific contexts. None is a general characterisation of all mining.

• Stranded energy. Electricity that can be generated but cannot be economically transmitted to existing demand: remote hydro, wind in transmission-constrained regions, geothermal in undeveloped areas. Documented examples include hydro mining in Sichuan during pre-ban wet seasons, wind-curtailment mining in West Texas, geothermal in Iceland and El Salvador, and remote hydro in Paraguay. The genuine stranded-energy share of global mining is small; most mining is grid-connected and competes with other loads.

• Methane flaring mitigation: the Crusoe/ExxonMobil case. Crusoe Energy's Digital Flare Mitigation runs mobile data centres burning associated petroleum gas at oil well pads that would otherwise be flared.

Source: Crusoe materials and CNBC coverage of the Exxon Bakken pilot (source: CNBC). ExxonMobil ran the Bakken pilot from January 2021; Crusoe sold its mining operation to NYDIG in March 2025 and refocused on AI infrastructure (source: Crusoe Energy). Methane has ~80x the warming potential of CO2 over 20 years, so combustion is a real reduction. The Fortune March 2022 critique fairly noted flared-gas mining is small relative to global flaring.

• ERCOT demand response in Texas. ERCOT has the most developed Bitcoin-mining demand-response programs globally. Miners categorised as Large Flexible Loads (LFLs) curtail when wholesale prices spike or grid stress emerges.

Source: US EIA Today In Energy (source: EIA). Texas SB 6 (passed June 2025, effective September 1, 2025) requires curtailment protocols for new large-load interconnections after December 31, 2025, formalizing what had been an ad-hoc demand response. Miners can curtail in seconds rather than the minutes-to-hours typical of industrial load, and curtailment credits partially offset forgone mining revenue.

For how Bitcoin mining pools coordinate hashrate, the pool layer is largely orthogonal to grid-services participation, which happens at the data-centre operator level.

Carbon intensity is what matters, not just TWh

Knowing how many terawatt-hours Bitcoin consumes tells you less than most headlines imply. Two operations consuming identical electricity can produce emissions differing by a factor of five or more depending on the source mix per region.

Carbon intensities by source range roughly from 4-12 gCO2eq/kWh for hydro and nuclear, through 11-45 gCO2eq/kWh for wind and solar (lifecycle), to 400-490 gCO2eq/kWh for natural gas combined cycle and 820-1,000+ gCO2eq/kWh for coal. Cambridge 2025's 39.8 MtCO2e for ~138 TWh implies a fleet-average ~290 gCO2/kWh, consistent with the gas-dominant mix (38.2%) plus coal residual (8.9%) offset by the 52.4% sustainable share. Renewable-share estimates have ranged 25-70%+ across earlier studies; methodological convergence around Cambridge's 52.4% is meaningful but is still a single transparent estimate, not a final answer.

For how Bitcoin is meaningfully decentralised at the consensus layer, geographic dispersal is structurally relevant: no single grid's coal mix dominates the global average.

ASIC efficiency curve: more hashes per joule each generation

Bitcoin's energy use has grown more slowly than its hashrate because ASICs become roughly 20-30% more efficient per generation. Joules-per-terahash (J/TH) is the standard metric; lower is better.

Source: D-Central ASIC efficiency analysis (source: D Central) and MineShop 2026 ASIC ranking (source: Mineshop).

Two implications: total energy use grows slower than hashrate (a 2x hashrate increase typically pulls only ~1.4-1.6x more electricity once efficiency-improved fleets phase in), and generation gains have slowed from earlier 50-100% steps to 20-30% as chip fabrication approaches 3-5nm physical limits.

For how Bitcoin's difficulty adjustment redistributes work, efficiency improvements cut joule cost per block but do not change the security spend ratio: miners compete the energy spend up to marginal revenue from block subsidy plus fees.

The strongest critiques, steelmanned

Both sides of the Bitcoin energy debate have serious arguments. The honest move is to state each one fully, then note where it holds and where it overreaches.

• Mora et al. 2018 on Bitcoin emissions and 2°C warming. The Mora et al. paper in Nature Climate Change (October 2018) modelled mass-adoption scenarios and concluded Bitcoin emissions could push warming above 2°C by 2033 (source: Nature). Nicolas Houy's "Rational mining limits Bitcoin emissions" (Nature Climate Change 2019) critiqued the methodology: Mora assumed each block was mined on hardware sampled from a self-compiled list of 62 devices (many obsolete) and ignored transaction-throughput constraints. Mora et al. replied (source: Nature) maintaining the broader caution. Net: the specific 2°C-by-2033 figure depends on adoption assumptions that have not played out; the broader caution that proof-of-work emissions matter at scale stands.

• Greenpeace "Change The Code" 2022-2024. Greenpeace USA, the Environmental Working Group, and Ripple co-founder Chris Larsen launched Change the Code, Not the Climate in March 2022, calling for Bitcoin to switch from proof of work to proof of stake (source: Greenpeace) and citing a potential 99.9% emissions reduction. The Bitcoin community's response was that the change was technically possible but socially and economically prohibitive: it would require coordination across miners, exchanges, custodians, and node operators with no consensus mechanism for it, would zero tens of billions in SHA-256 ASIC capex, and would change Bitcoin's security model in ways many holders consider central to the asset. The campaign was suspended around December 2024 with no protocol change. Ethereum's separate September 2022 transition demonstrated the switch is technically possible for a different chain with materially different governance.

• The opportunity-cost argument. Every TWh Bitcoin uses could in principle have served other demand (heat pumps, EVs, AI workloads). The counter-argument is that markets allocate electricity to the highest-bid use, and Bitcoin mining functions in many cases as a price-of-last-resort buyer (curtailable, sited at stranded sources, willing to pay below grid retail), which improves the economics of building generation and transmission. Both conditions exist in different operations.

Mitigations that actually exist in 2026

The mitigation landscape is concrete enough to enumerate rather than speculate about.

Each mitigation addresses one slice. None individually solves the footprint; collectively they explain why sustainable share rose from 37.6% (2022) to 52.4% (2025) without any protocol change.

For how miners, nodes, and wallets divide responsibilities, only the miner layer has direct energy implications; nodes and wallets are essentially zero-energy.

Honest framing: what we know, what we don't, what the verdict is

The honest answer on Bitcoin energy distinguishes facts, methodological uncertainties, and value judgments.

• What we know: Bitcoin uses ~138 TWh per year on the most rigorous transparent estimate (Cambridge April 28, 2025), about 0.5% of global electricity. Sustainable share was 52.4% in early 2025 (42.6% renewables plus 9.8% nuclear), up from 37.6% in 2022. Geographic distribution post-China-2021 is more dispersed: top three countries (US, Russia, China) hold ~65%. ASIC efficiency has improved roughly 7x in eight years. Crusoe flare-mitigation operations achieve 99.89% methane combustion vs ~93% baseline. ERCOT Bitcoin miners earned $30.6M in Q3 2025 curtailment credits.

• What is methodologically uncertain: The precise renewable share (Cambridge's 52.4% is rigorous but is from a 48%-of-hashrate survey; earlier studies spanned 25-70%+). Total energy use itself (CBECI's low-central-high range is ~100-180 TWh). Per-region carbon intensity (depends on hour-of-day grid mix). The counterfactual: would displaced electricity have served valued or wasted purposes?

• What is a value judgment: Whether the security and monetary properties Bitcoin provides justify the spend. How to weigh 0.5% of global electricity against the option value of an open monetary network securing $1-2T in capitalization. Whether ESG mandates should treat Bitcoin holdings as additive to portfolio emissions, neutral, or net-positive given grid-balancing services.

From BloFin's exchange-operator vantage point, the energy question shows up most concretely in client conversations about institutional ESG mandates and treasury allocation. Our practical observation is that the source-mix data (Cambridge 52.4% sustainable, the natural-gas-not-coal substitution, the dispersal away from single-country grids) has shifted enough that legacy "Bitcoin equals coal" framings no longer match the 2025-2026 evidence; whether that resolves the value judgment for any particular investor is a question we leave to their own ESG framework. Where we do take a position: any honest ESG read of Bitcoin should use date-stamped methodology-disclosed estimates (the Cambridge series remains the most defensible) rather than legacy advocacy figures from 2018-2021 that pre-date the China migration and the coal-to-gas shift.

Common confusions worth flagging

"Bitcoin uses as much energy as [country X]." 

Sometimes true at total TWh, but the comparison omits scope. A country supplies homes, hospitals, factories, transport, and government; Bitcoin is one global industry. Equal TWh does not imply equivalent value or burden.

"Per-transaction energy is huge." 

Bitcoin secures the entire confirmed history and current UTXO set independent of throughput; energy scales with hashrate, not transactions.

"Bitcoin mining mostly uses coal." 

Closer to true in the 2017-2020 China-dominant era. Cambridge 2025 puts coal at 8.9% versus 36.6% in 2022; natural gas (38.2%) and renewables (42.6%) dominate the 2026 mix.

"Renewables claims are just greenwashing." 

Some advocacy figures deserve scrutiny. The Cambridge April 28, 2025 study is third-party, transparent, and survey-based; treating it on the same footing as advocacy claims understates its rigour.

"Switching to proof-of-stake would solve Bitcoin's energy problem." 

Technically possible; socially and economically prohibitive on Bitcoin specifically. The Greenpeace "Change The Code" 2022-2024 campaign was suspended around December 2024 with no protocol change.

"More efficient ASICs reduce total energy use." 

Per-hash efficiency improves 20-30% per generation, but as long as block subsidy plus fees remain economically valuable, miners spend the marginal joule up to revenue. Better ASICs deploy more hashrate at the same energy cost.

FAQ

How much electricity does Bitcoin use in 2026?

The most rigorous transparent estimate is ~138 TWh annualised, from the Cambridge Digital Mining Industry Report published April 28, 2025 by Cambridge Judge Business School, about 0.5% of global electricity consumption Cambridge CJBS. The Cambridge Bitcoin Electricity Consumption Index publishes a low-central-high range that has run roughly 100-180 TWh over recent quarters, reflecting genuine model uncertainty about hardware mix, utilisation, and geographic distribution. Any single number without a date and methodology note should be treated cautiously; the figure shifts as hashrate and ASIC fleets evolve.

What share of Bitcoin mining uses renewable or sustainable energy?

The April 2025 Cambridge study puts the sustainable share at 52.4%: 42.6% renewables (hydropower 23.4%, wind 15.4%, solar 3.2%, other 0.6%) plus 9.8% nuclear. This is up from 37.6% in 2022 and is the first transparent third-party estimate to cross 50%. Earlier studies ranged 25-70%+ reflecting methodological dispersion. The Cambridge figure draws from a 49-firm survey covering 48% of global hashrate, the most rigorous public methodology available; broad claims of "mostly fossil" or "mostly renewable" both exceed what current data reliably support.

How does Bitcoin's energy use compare to other industries and countries?

Bitcoin's ~138 TWh is comparable to Norway's national consumption (~140 TWh in 2022 per Statnett), larger than Argentina's (~115 TWh in 2023), about half of global gold mining (~240 TWh on mid-2020s estimates), and about a third of global data centres (415 TWh in 2024 per IEA, projected to 945 TWh by 2030 driven mainly by AI). Globally Bitcoin is ~0.5% of total electricity consumption. These comparisons develop scale intuition, not equivalence, since each comparator has different scope, services, and population.

Does Bitcoin mining help reduce methane emissions from oil flaring?

In specific operations, yes. Crusoe Energy's Digital Flare Mitigation achieves 99.89% methane combustion efficiency at oil well pads, versus ~93% for baseline flaring, cutting CO2-equivalent emissions by ~63% Crusoe. ExxonMobil ran a Crusoe pilot in the North Dakota Bakken from January 2021; Crusoe sold its mining business to NYDIG in March 2025 and refocused on AI infrastructure. Flare-gas operations are a small share of global hashrate but a meaningful real-world slice of the mitigation landscape.

Why does Bitcoin use energy at all if proof of stake exists?

Proof of work converts electricity into the cumulative-work property securing Nakamoto consensus: rewriting confirmed history requires outpacing the honest network's energy spend. Proof of stake substitutes staked capital. Both are valid security models. Ethereum transitioned PoW to PoS in September 2022; Bitcoin has not. The Greenpeace "Change The Code" 2022-2024 campaign asking for that switch achieved no protocol change and was suspended around December 2024. The Bitcoin community broadly views PoW as central to the asset; changing it would zero tens of billions in SHA-256 ASIC capex with no precedent on Bitcoin.

What is the per-transaction energy of Bitcoin and is it a useful number?

A per-transaction figure is calculable (annual TWh divided by tx count) but conceptually misleading. Bitcoin's energy use is set by hashrate and the security target, independent of transaction throughput; the network spends roughly the same energy at 250,000 daily transactions or 500,000. What is being secured is the entire confirmed history and current UTXO set, not individual transactions. Off-chain layers like Lightning further confound the per-tx denominator: one on-chain transaction can settle thousands of off-chain payments. Per-block-time energy (~2.6 GWh per ten-minute block at 138 TWh/year) is the more meaningful unit.

Is the energy mix actually improving over time?

Yes on the available data. Cambridge's transparent third-party estimate moved from 37.6% sustainable in 2022 to 52.4% sustainable in early 2025. The biggest shift is coal share dropping from 36.6% to 8.9%, with natural gas (25% to 38.2%) and renewables-plus-nuclear (37.6% to 52.4%) absorbing the gap. Geographic dispersal away from China's 2021 coal-heavy mix toward US gas, Paraguay hydro, Ethiopia hydro, and other lower-carbon grids is the structural driver. Whether the trajectory continues depends on grid decarbonisation in mining-host countries.

Researched and written by the BloFin Academy editorial team with AI-assisted drafting. Factual claims independently verified against the Cambridge Centre for Alternative Finance Bitcoin Electricity Consumption Index, the Cambridge Judge Business School Digital Mining Industry Report of 28 April 2025 (138 TWh and the 52.4% sustainable energy mix), the Hashrate Index Q2 2026 Global Hashrate Heatmap for country distribution and the 1,004 EH/s 30-day moving average, the IEA Energy and AI report (April 2025) for the 415 TWh global data-centre figure and the 945 TWh 2030 projection, US EIA Today In Energy for Texas ERCOT large-flexible-load data, Crusoe Energy's Digital Flare Mitigation methane-combustion materials with CNBC and CoinDesk on the ExxonMobil-Crusoe Bakken pilot, the Mora et al. 2018 Nature Climate Change paper and the Houy "Rational mining limits Bitcoin emissions" rebuttal, the Greenpeace "Change The Code" 2022 launch and 2024 suspension reporting, D-Central and MineShop ASIC efficiency rankings, and the Statnett and IEA country pages for the Norway and Argentina comparators.

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.