Quantum NISQ Miners and Proof-of-Work Energy: The Promise of a Swedish-Sized Savings and Testing Its Limits

01.11.2026

Bitcoin and other proof-of-work systems are criticized for their enormous energy consumption. But what if a quantum computer—a device typically portrayed as energy-intensive—could radically reduce the energy required to mine a single block? A 2023 paper estimates savings of approximately  126.7 TWh per year , or about the same as Sweden’s annual consumption. The question isn’t whether this figure is impressive, but rather  what assumptions make it true and how to verify it . The paper examines the work of Kearney and Perez-Delgado, which evaluates the energy efficiency of switching from Bitcoin ASIC miners to quantum miners: a NISQ device without error correction, as well as quantum miners with single-layer and double-layer Shor codes. The baseline is the Antminer S19 XP and CBECI network data. The authors apply the Landauer limit, an estimate of the number of erasable bits, and the inefficiency coefficient of a classical ASIC, and then transfer this coefficient to quantum architectures. Under this assumption, quantum miners are indeed many orders of magnitude more energy efficient. However, the claimed savings are  theoretical engineering estimates and predictions , not measured results. The article proposes a testable hypothesis, a critical analysis of the numerical assumptions, an experimental design, and a research agenda linking quantum mining with NISQ, error correction, Grover’s algorithm, mining economics, and blockchain security.

Keywords:  quantum computing, NISQ, Proof-of-Work, Bitcoin, Grover’s algorithm, Landauer limit, Shor code, energy efficiency, cryptography, blockchain.

Content

⚙️ Detailed research option

  1. Central question and testable hypotheses
  2. Why the topic is important and what remains beyond the borders
  3. Mechanism: Proof-of-Work, Grover and Landauer Limit
  4. Methodology and initial data of the article
  5. Results: Three quantum architectures versus ASICs
  6. Critical Analysis: Where Evaluation Is Strong and Where It Is Fragile
  7. Relationship with other works and scientific context
  8. Selected research questions and evidence
  9. What experiments will actually test the hypothesis?
  10. Economy, network and social consequences
  11. Conclusion: what can be stated now
  12. Sources

1. Central question and testable hypotheses

⚙️ Detailed research option

The central question of the article is:  can the transition from classic ASIC miners to quantum ones, including NISQ miners without full error correction, provide a measurable and sustainable reduction in energy per mining block with a comparable success rate and comparable block time?

This question is broken down into three testable hypotheses.

HypothesisFormulationWhat could disprove it?
H1. Energy advantageA quantum miner, with equal block success probability and equal block time, consumes less energy per block than a modern ASIC by a margin of at least 10 6 .Direct measurement of the total power consumption of a quantum miner, including control, cooling, metering, and classical processing, shows whether ASICs are equal or inferior.
H2. Robustness to error correctionEven with a two-layer Shor code, the advantage remains if the inefficiency of the quantum device does not exceed approximately 5×10 11  relative to the theoretical minimum.The actual inefficiency of a quantum miner with ECC turns out to be higher than this threshold; or the overhead of ECC, control, and cryogenics is not fully accounted for.
H3. NISQ suitabilityMining is a suitable task for NISQ because it does not require a high probability of success and allows for short runs followed by resets.The depth of Grover’s scheme for SHA-256 is so great that a NISQ device loses coherence before producing a useful result; or the cost of repeated runs makes mining unprofitable.

The key verification criterion  is not “does a quantum mining algorithm exist?” but “does a physical implementation exist whose  measured total energy consumption per block  is lower than that of an ASIC, with the same success probability and block time.” This criterion is what separates a scientific hypothesis from a beautiful theoretical prediction.

2. Why the topic is important and what remains outside the boundaries

⚙️ Detailed research option

Proof-of-work is one of the most common consensus mechanisms in cryptocurrencies. A miner must find a nonce such that the block header hash is below a target value. This is a deliberately computationally difficult task, resulting in enormous energy consumption. According to the Cambridge Bitcoin Electricity Consumption Index, the article cites a value of  126.7 TWh per year  for Bitcoin; this is greater than the annual consumption of Norway (124.3 TWh) and about the same as Sweden. The authors estimate the potential savings from switching to quantum miners to be approximately the same 126.7 TWh per year.

The interest here is twofold. First, environmental and economic: even a partial reduction in mining energy would be noticeable. Second, scientific: NISQ devices have long sought practical applications. Mining doesn’t require a universal quantum computer and doesn’t require a high probability of success—this makes it an unusually convenient testing ground for NISQ.

Boundaries of the article.  We do not consider:

  • A quantum computer attack on Bitcoin’s ECDSA signatures is a separate cryptographic threat;
  • quantum key distribution and post-quantum cryptography as such;
  • quantum sensors and quantum communications;
  • a complete economic model of the Bitcoin price and the cost of quantum hardware;
  • legal and political mechanisms for banning PoW.

This limitation is important: the article is about  mining , not about “hacking Bitcoin with a quantum computer.”

3. Mechanism: Proof-of-Work, Grover and Landauer Limit

⚙️ Detailed research option

3.1. Classic mining

⚙️ Detailed research option

Bitcoin uses a HashCash-like PoW: a miner forms a block, tries a nonce, and calculates SHA-256. Success is achieved when the hash is less than the target value. The article estimates the success rate of a single ASIC per block at approximately  0.000070% . This is crucial: a miner isn’t required to find a block with near certainty. They can make multiple attempts with a low probability of success.

3.2. Quantum Miner as a Search for Grover

⚙️ Detailed research option

The quantum miner in the paper implements nonce search using Grover’s algorithm. The oracle marks states for which the block hash does not exceed the target value. Iterations are then performed: oracle, Hadamard transform, phase shift, and Hadamard again. The number of iterations is estimated as  t = sqrt(N/M), where  N is the size of the search space,  M = maxTarget / Difficulty. The appendix to the paper specifies  1280 gates per Grover iteration  and a search space of  2^256.

The NISQ variant without error correction uses  512 qubits , which are measured at the end. While unitary evolution is running, data is not erased; the main erasure occurs when the output is measured. This yields a theoretical Landauer minimum  of 1.4336×10−18 J  per  block.

3.3. Landauer Limit

⚙️ Detailed research option

Minimum bit erasure energy:

E = k B  T ln 2 · B

where  k_B is the Boltzmann constant,  T is the heat sink temperature,  B and is the number of erased bits. For a classic miner, the article assumes a 256-bit hash, 8588 NAND gates per hash, and 0.625 erase bits per NAND. At 84,000 TH per block, this yields approximately  4.72× 1020  erased bits and a theoretical minimum of  1.324 J  per block.

3.4. Error correction

⚙️ Detailed research option

For ECC variants, the Shor code is used. The paper considers one and two layers. Each correction step requires measuring 12 qubits over two rounds. The total number of correction steps is estimated as  ECSteps = t × g × d, where  g = 1280, and  d is the number of qubits to be corrected. For the single-layer variant, the number of erased bits is approximately  1.339×10 11 , while for the two-layer variant, it is  1.607×10 12 . The theoretical minimums are  3.7497×10 −10  J  and  4.4996×10 −9  J  , respectively.

4. Methodology and initial data of the article

⚙️ Detailed research option

The baseline is a modern Antminer S19 XP ASIC miner with a performance  of 140 TH/s . The article also includes a power consumption  of 3.010 kW , which yields an inefficiency ratio  of 1:379  relative to the Landauer limit. The second ratio,  1:1706 , is derived from CBECI network data: the current Bitcoin hashrate is approximately  201 million TH/s , the energy per block is approximately  3.2267 GJ , and the share of a single device is approximately  0.000070% , which yields approximately  2.25869 kJ  per block for this ASIC.

The authors then multiply the theoretical minimums of quantum architectures by the same factors of 1:379 and 1:1706. This is the central assumption:  a quantum miner, in terms of its inefficiency relative to the theoretical optimum, is in the same order of magnitude as a classic ASIC .

Evidence status.  All numbers for quantum miners in the paper are estimates, not measurements. They are derived from the Landauer model, the number of gates, the number of ECC steps, and the transfer of the classical inefficiency coefficient. There is no experimental quantum miner in the paper to confirm these numbers.

5. Results: Three quantum architectures versus ASICs

⚙️ Detailed research option

The key figures in the article are summarized below. For a classic ASIC, the actual power consumption is approximately  2258.69 J  per block. For quantum architectures, the theoretical minimum and forecast are shown at a ratio of 1:1706. The last column is the ratio of the classic actual consumption to the forecast for the quantum architecture; this is a more conservative comparison than the ratio to the theoretical minimum.

ArchitectureTheoretical minimum, J/blockForecast at 1:379, J/blockForecast at 1:1706, J/blockWin rate against ASIC at 1:1706
Classic ASIC1,324—2258.69 (actual)1×
NISQ without ECC, 512 qubits1.43×10 −185.43×10 −162.445× 10−15≈9.2×10 17
Quantum miner, 1 layer of Shor3.75×10 −101.42×10 −76.40×10 −7≈3.5×10 9
Quantum miner, 2 layers of Shor4.50×10 −91.70×10 −67.68×10 −6≈2.94×10 8

Even the most expensive quantum variant—a dual-layer ECC miner—according to the article’s estimates, consumes approximately  2.94×108 times  less energy per block than a classic ASIC. For the NISQ variant without ECC, the claimed gain is even more dramatic. This is where the estimated annual savings of approximately  126.7 TWh come from .

The article also provides a robustness table: how much less efficient a quantum miner must be than its theoretical minimum to even match a classic ASIC. For the dual-layer ECC variant, this threshold is approximately  5.02×10 ; for the NISQ variant, it is approximately  1.58×10 . This means that even if a quantum miner is hundreds of billions of times worse than its ideal limit, the dual-layer ECC version still retains an advantage.

An important distinction.  The figure “1.58×10^ 21 ” in the article is the ratio  of actual classical  consumption to  the theoretical quantum minimum , not a prediction taking into account the actual inefficiency of the quantum device. If we apply the same factor of 1:1706 to the quantum minimum, the gain of the NISQ variant becomes approximately 9.2×10^ 17 . This is still enormous, but methodologically these are different statements.

6. Critical analysis: where the assessment is strong and where it is fragile

⚙️ Detailed research option

6.1. Strengths of the work

⚙️ Detailed research option

  • An explicit baseline is used – a specific ASIC and network data, rather than an abstract “Bitcoin power consumption”.
  • The Landauer limit is applied sequentially: first to the classical miner, then to quantum architectures.
  • Three levels are considered: NISQ without ECC, one Shor layer, and two Shor layers. This provides a range rather than a single optimistic figure.
  • The authors explicitly acknowledge that reaching the Landauer limit is unrealistic and test robustness through threshold inefficiency coefficients.
  • The work connects quantum mining with the NISQ agenda: mining does not require a universal fault-tolerant computer.

6.2 Where do restrictions begin?

⚙️ Detailed research option

LimitationWhy is this important?What to check
Transfer of the inefficiency coefficientA classic ASIC and a quantum miner have different sources of loss: cryogenics, lasers, microwave control, vacuum, measurement, and shielding. A single loss factor of 1:1706 may not be sufficient.Full energy balance of a quantum miner: cooling, control, reading, classical processing, downtime.
Lack of experimentAll quantum numbers are calculated. There is no demonstration of SHA-256 mining on a quantum device with energy measurement.A small-scale prototype with reduced nonce space and fair energy measurement on Oracle Query.
Circuit depth and NISQ noiseThe NISQ variant without ECC assumes that 512 qubits and Grover’s scheme have enough time to operate before decoherence occurs. This needs to be proven, not assumed.T1/T2 score, oracle fidelity, number of iterations, probability of success before reset.
Search spaceThe article uses [nonce]  2^256, while the classic nonce in Bitcoin is a 32-bit field; the space is expanded by header and extranonce variants. An accurate model is needed, otherwise the comparison with ASICs is invalid.An explicit model of entry, acceptable header changes, and success probability.
Network effect and complexityIf mining becomes cheaper, new miners may join the network; difficulty will increase; the total network energy will not necessarily fall by 126.7 TWh.Network simulation with difficulty adjustment, miner entry, and rebound effect.
Numerical inconsistenciesThe article’s tables contain points that require verification: for example, for single-layer ECC, one table lists a very high threshold, while a direct calculation yields approximately 6×1012 , not 6× 1021 . The text also mixes comparisons with the theoretical minimum and predictions with the inefficiency coefficient.Independent recalculation of all tables, publication of code and initial assumptions.
Source statusThis is a preprint; independent replication and peer review would strengthen the findings.Third party verification, calculation replication, experimental demonstration.

It’s impossible to state based on the article’s results:  “a quantum miner already saves 126.7 TWh”; “Bitcoin has become energy efficient”; “a NISQ miner has been experimentally proven.” The correct formulation is:  under the accepted assumptions, the model predicts enormous potential; this is a testable hypothesis, but one that has not yet been experimentally confirmed .

7. Relationship with other works and scientific context

⚙️ Detailed research option

The article doesn’t exist in a vacuum. It draws on several lines of research that are useful to explicitly link.

Work / lineRole in analysisWhat does this give for our article?
Bard, Kearney, Perez-Delgado, “Quantum advantage on proof of work” [14]Quantum acceleration of PoWShows that Grover’s algorithm is applicable to mining; our paper adds an energy analysis.
Preskill, “Quantum computing in the NISQ era and beyond” [15]NISQ AgendaExplains why NISQ applications are sought in narrow tasks; mining is a candidate.
Fruchtman, Choi, Technical Roadmap for Fault-Tolerant Quantum Computing [16]The boundary between NISQ and fault toleranceHelps to understand why a miner does not need a universal fault-tolerant computer.
Kearney, Perez-Delgado, “Vulnerability of blockchain technologies to quantum attacks” [10]Quantum threat to blockchainIt distinguishes between two scenarios: signature attacks and energy-efficient mining are different tasks.
CBECI [9]Energy consumption baselineGives 126.7 TWh and grid power consumption for a ratio of 1:1706.
Kim, Ryou, Jun, SHA-256 hardware [21]Gate evaluation8588 NAND gates per hash is the basis for the Landauer limit of a classic miner.
IEA, Sweden [22]Scale of comparison126.7 TWh is about the same as Sweden’s annual consumption; this is a rhetorically strong argument, but not a convincing one.

The connecting thread is this:  PoW provides a specific task ;  Grover provides a quantum algorithm ;  NISQ provides a feasible platform ;  Landauer provides a lower bound on energy ;  ECC provides a price for reliability ;  and network economics determines whether theoretical efficiency translates into real savings . It is at the intersection of these five lines that the main scientific interest lies.

8. Selected research questions and evidence

⚙️ Detailed research option

The question constructor contains questions that directly address the hypothesis being tested: problem statement (1–6), mechanism (7–12), apparatus (19–26), errors (31–35), fair comparison (37–42), evidence (43–46), cryptographic context (49–56), economics (63–66), and future (69–72). Below is a map of “question → what the article answers → status of evidence.”

QuestionAnswer within the articleStatus
1–6. What problem are we solving, what is the hypothesis and boundaries?Problem: PoW energy consumption; hypothesis: a quantum/NISQ miner can dramatically reduce the energy per block; boundaries: no attack on ECDSA and no full economic model.production
7–12. What is the mechanism and what is measured?PoW as nonce search; Grover; SHA-256 oracle; 512 qubit measurement; Landauer limit; bit erasure energy.theory + assessment
19–26. What platform, noise, scaling and bottlenecks?NISQ without ECC; ECC with Shor code; 512 qubits; cryogenics and control are not fully included in the model; the main bottleneck is circuit depth and decoherence.open question
31–35. What is the error model and the cost of correction?One and two Shor layers; 12 measured qubits in two rounds; growth of erased bits and energy; the two-layer variant still wins by model.grade
37–42. What are we comparing it to and what is stated?Comparison with Antminer S19 XP; equal block success rate; claimed theoretical/calculated advantage, not experimental.calculation
43–46. What constitutes evidence and is it reproducible?Proof is a model and calculation; no experiment; the code and data in the article are not presented in full; independent replication is needed.not confirmed by experiment
49–56. The Cryptographic Context of BitcoinWe are discussing PoW hashing, not ECDSA signatures; the public key and signature attack are a separate topic.demarcation
63–66. Economy and Social ValuePotential savings of 126.7 TWh; but the cost of quantum hardware, centralization, and rebound effects are not fully estimated.forecast
69–72. Open Questions and Next StepsA prototype, a full energy balance, a network model, and a test of the robustness of the conclusions under pessimistic assumptions are needed.experimental plan

9. What experiments will actually test the hypothesis?

⚙️ Detailed research option

  1. Small-scale quantum PoW prototype.  Implement Grover for SHA-256 with a reduced nonce space (e.g., 8–16 bits) on a real NISQ device. Measure: oracle fidelity, number of iterations, success probability, energy per run, control energy, and read energy.
  2. A fair comparison with ASICs.  Compare not the “ideal quantum minimum” with the “real ASIC,” but  the measured total power consumption  of the quantum prototype with the measured power consumption of the ASIC with the same block success rate and block time.
  3. ECC experiment.  Test how real errors and Shor’s correction affect the erase count and energy. Compare with the formula  ECSteps = t × g × d and table values.
  4. Full lifecycle.  Consider manufacturing, cryogenics, lasers, microwave electronics, vacuum, shielding, traditional post-processing, and downtime. Without this, the “energy per unit” may be underestimated by orders of magnitude.
  5. Network simulation.  Model the entry of quantum miners, difficulty adjustments, and potential growth of the overall hashrate. Determine whether the annual savings of 126.7 TWh is maintained or whether a rebound effect occurs.
  6. Independent recalculation of tables.  Reproduce all the numbers in the article, eliminate inconsistencies, and publish the code. This is a cheap but important step for trusting the results.

A minimal demonstration that will convince.  A real quantum miner running a reduced task, for which the success probability, time to reset, total power consumption, and cost per successful block have been measured—and which, under the same conditions, outperforms a classic ASIC. Until such a demonstration is available, this is a strong theoretical assessment.

10. Economy, network and social consequences

⚙️ Detailed research option

Even if a quantum miner is energy efficient at the device level, this doesn’t automatically translate into a 126.7 TWh reduction in network energy. This is due to mining economics. If the cost of energy per unit of hashrate falls, mining becomes more profitable, new participants join the network, and the difficulty rises. A rebound effect occurs: some of the gains can be offset by the increase in overall computing volume. This is a classic energy efficiency problem and cannot be ignored.

Furthermore, a quantum miner is a specialized and expensive device. If it’s only available to a few operators, there’s a risk of mining centralization. This doesn’t make the idea a bad one, but it adds a dimension missing from a purely energy-based calculation:  who controls effective mining ?

Finally, the comparison with Sweden’s consumption is strong rhetoric, not proof. 126.7 TWh is an order of magnitude that helps the reader get a sense of scale. For a scientific paper, what’s more important isn’t “Sweden,” but the metric being tested: J per block with equal success probability and equal block time.

11. Conclusion: what can be stated now

⚙️ Detailed research option

The short answer is:  under the article’s assumptions, yes, quantum miners, including the NISQ variant without full error correction, can provide a huge energy advantage. Even a two-layer ECC miner, according to the model, outperforms ASICs by a factor of approximately 2.94×108 ,  and the loss threshold is so high that the conclusion seems robust to very poor engineering.

But  this  isn’t  a measured savings or  a  proven fact. The central assumption—the transfer of the classical inefficiency coefficient to a quantum device—requires experimental verification. Cryogenics, control, decoherence, circuit depth, network rebound effects, and centralization are not fully accounted for. There are sections of the tables that require independent recalculation.

The most interesting question to test for the future is:  is it possible to build a NISQ miner that, with  equal success rates  and  fully accounting for its lifecycle energy,  actually consumes less than an ASIC? If so, this will not just be a pretty estimate, but a real technological advance. If not, we will discover the specific limitation (noise, depth, cooling, or economics) that ruins this pretty picture.

The main takeaway for the reader:  Quantum mining is a rare example of a problem where a NISQ device doesn’t have to be universal and fault-tolerant to be of interest. This makes the topic scientifically valuable. But the value here isn’t in the sensational “quantum computer will save Bitcoin,” but in an honest verification of  where exactly a theoretical advantage translates into measured energy .

12. Sources

⚙️ Detailed research option

  1. J. Kearney, C. A. Perez-Delgado.  Quantum Blockchain Miners Provide Massive Energy Savings . arXiv:2306.03321v1. Main text analyzed: NISQ/ECC miner scores, Landauer bound, Tables I–II, Grover’s algorithm.
  2. D. A. Bard, J. J. Kearney, and C. A. Perez-Delgado.  Quantum advantage on proof of work . Array, 15:100225, 2022. Quantum acceleration line PoW.
  3. J. Preskill.  Quantum computing in the NISQ era and beyond . Quantum, 2:79, 2018. The context of NISQ and its limitations.
  4. A. Fruchtman, I. Choi.  Technical Roadmap for Fault-Tolerant Quantum Computing . UK EPSRC, 2016. The boundary between NISQ and fault-tolerant systems.
  5. JJ Kearney, CA Perez-Delgado.  Vulnerability of blockchain technologies to quantum attacks . Array, 10:100065, 2021. Context of quantum threats to blockchain.
  6. Cambridge Center for Alternative Finance.  Cambridge Bitcoin Electricity Consumption Index (CBECI) . URL:  https://ccaf.io/cbnsi/cbeci .
  7. Bitmain.  Antminer S19 XP / S19j Pro . ASIC performance and power consumption data.
  8. M. Kim, J. Ryou, S. Jun.  Efficient hardware architecture of SHA-256 algorithm for trusted mobile computing . In Information Security and Cryptology, 2008. Evaluation of SHA-256 gates.
  9. International Energy Agency.  Sweden: Countries: IEA . URL:  https://www.iea.org/countries/sweden . Benchmark annual consumption for Sweden.
  10. National Quantum Initiative.  Directions in Quantum Science . URL:  https://www.quantum.gov/national-quantum-initiative/ . The context of computing, networks, and sensors.
  11. NIST.  Post-quantum cryptography: FIPS 203, 204, 205 approved , August 13, 2024. URL:  https://csrc.nist.gov/news/2024/postquantum-cryptography-fips-approved . The context is post-quantum migration, not mining.

Status note: References 1–5 and 8 are scientific/technical papers; 6–7 and 9 are data and specifications; 10–11 are context for trends and standards. Article numbers 2306.03321v1 in this article are cited as author estimates, not as independently verified measurements.