Quantum Computers: Energy-Efficient or Power-Hungry?

 

Only forty years ago, the number of mobile-phone users in the world was tiny. Consider the 1980s. Mobile phones were not yet part of people’s everyday lives; in fact, they were hardly within reach of ordinary people at all. At that time, total global electricity consumption was about 7,300 terawatt-hours (TWh) per year. [1 terawatt = 1 million megawatts = 1 trillion watts.] By 2024, following the rise of thousands of electricity-dependent technologies—including

mobile phones, smartphones, personal computing, cloud computing, data streaming, artificial intelligence, and global data centres—worldwide electricity consumption had risen to more than 27,000 terawatt-hours. In 1990, there were only about 12 million mobile-phone subscriptions worldwide. In other words, only about 24 out of every 10,000 people had a mobile phone. Yet by 2024, the total number of mobile-phone subscriptions had reached 9.1 billion—about one billion more than the world’s total population.

The increase in global electricity demand is not due simply to the use of mobile phones or computers. Individually, these devices do not consume very much electricity. But even such low-power devices have created an enormous supporting infrastructure around them—mobile towers, networks, cloud servers, data centres, cooling systems, software platforms, and so on—all of which require large amounts of electricity to operate continuously.

Now, as the world looks with enormous interest toward a transition from conventional computing to quantum computing, a natural question arises: will quantum computers be more energy-efficient than conventional computers, or will they become power-hungry machines whose demands place an intolerable burden on the world’s electricity sector?

There is no simple answer to this question. Mobile phones and computers may be small devices, yet they have created electricity-intensive ecosystems around them. In the same way, when assessing the energy demand of quantum computers, we must remember that a quantum chip itself may be small, but the infrastructure built around it—and the amount of electricity required to operate that infrastructure—will ultimately determine whether quantum computing is energy-efficient or power-hungry (1–4).

Will Quantum Computing Be Environmentally Friendly?

Quantum computers are already being regarded as a revolutionary technology capable of solving many problems far more efficiently and in much less time than conventional computing. But at what cost? This raises an important question: will quantum computing reduce overall electricity use, or will future quantum-computing data centres create another major burden on the electricity grid?

The answer is not straightforward. The quantum processor—the chip on which the actual quantum computation occurs—may itself use relatively little energy. However, the complete support system required to keep that processor operating—ultra-low-temperature cooling, lasers, microwave-control electronics, error-correction computation, classical servers, networking, and data-centre infrastructure—may consume substantial amounts of electricity. Quantum computing therefore should not automatically be regarded as ‘green computing’ or environmentally friendly computing. Whether it ultimately becomes a truly sustainable technology will depend on several factors: what type of quantum technology is used, what it is used for, how efficient the supporting systems are, and whether quantum computers are reserved for problems that offer a genuine quantum advantage. It may take many years, even after quantum computing becomes operational at scale, before these issues can be determined with confidence.

 

Lessons from History

The history of the mobile phone provides an important lesson—and perhaps a warning. A mobile phone in a person’s hand uses very little electricity compared with a household refrigerator, air conditioner, or electric vehicle. But that is not the right way to think about its energy use. A mobile phone does not operate independently; it is connected to an entire network. The mobile-phone network did not merely add billions of handheld devices—it created a complete global digital ecosystem. This ecosystem includes cellular base stations, fibre networks, routers, cloud services, streaming platforms, app stores, social-media services, artificial-intelligence systems, and enormous data centres.

The scale becomes clearer when we look at the electricity involved. According to the International Energy Agency, data-transmission networks consumed about 260–360 terawatt-hours of electricity in 2022, equivalent to roughly 1–1.5% of total global electricity consumption. Mobile networks accounted for about two-thirds of that energy use [4]. In addition, the world’s data centres alone consumed about 415 terawatt-hours of electricity in 2024—roughly twice Bangladesh’s annual electricity demand. Data centres now account for around 1.5% of global electricity demand. By 2030, this figure could more than double to about 945 terawatt-hours [5].

Why is the comparison with mobile-phone electricity use important for quantum computing? Because, as widespread users of mobile phones, we can easily understand how an entire system grows around an individual device. A smartphone is not really just a phone—it is merely the visible endpoint of a vast digital infrastructure. In the same way, a quantum computer is not merely a quantum chip; it is the visible part of a complex hybrid infrastructure. A future quantum data centre may integrate many systems simultaneously: cryogenic refrigerators (which allow the quantum-computing system to operate at ultra-low temperatures), lasers, vacuum systems, microwave electronics, quantum processors, classical control systems, real-time error-correction processors, storage, networking, and conventional high-performance computing systems. Their combined electricity demand could be enormous.

 

Bits versus Qubits

To understand how energy use in quantum computers should be assessed, we first need to understand the fundamental difference between conventional and quantum computers.

The basic unit of information in a conventional computer is the binary digit, or bit, whose value is always either 0 or 1. Energy use in such computers arises mainly from several sources: switching vast numbers of transistors, transferring information between processors and memory, storing information, and removing heat from the system. In modern high-performance computing and artificial intelligence, the greatest electricity consumption comes from GPUs, CPUs, memory systems, networking equipment, and cooling. This is why conventional data centres have already become major electricity consumers. According to the International Energy Agency, electricity use by data centres has grown by about 12% per year in recent years, far faster than the overall rate of growth in electricity demand [5].

Quantum computers work differently because they use qubits rather than ordinary bits. A simple way to understand a qubit is this: whereas a conventional bit can hold only one of two values—0 or 1—at any given moment, a qubit can exist in a combination of 0 and 1 at the same time, a condition known as superposition. In addition, multiple qubits can be linked so that the state of one is correlated with the state of another; this phenomenon is called quantum entanglement. Together, these two properties allow quantum computers to explore many possibilities in parallel, whereas conventional computers may have to test them one by one.

Because of these properties, quantum computers can, in principle, solve some problems in quantum simulation, quantum chemistry, materials design, cryptography-related algorithms, optimisation, and certain sampling tasks in far fewer computational steps than conventional computers. A useful quantum computer may therefore solve some specific problems using much less electricity. But that is not the whole picture.

 

Not a Stand-Alone Chip, but a Hybrid Machine

We should remember that today’s quantum computers are not general replacements for conventional computers. They are essentially specialised accelerators that work together with classical computers. IBM’s Quantum System Two, for example, is designed so that multiple quantum processing units can be connected to conventional data-centre systems. Its cryogenic quantum infrastructure, conventional runtime servers, and modular qubit-control electronics operate together [6]. In other words, a practical quantum computer is not simply a chip but a hybrid system in which quantum and conventional technologies work side by side.

 

Dependence on Hardware

The amount and pattern of energy use in a quantum computer depend largely on the hardware platform being used—and this is where much misunderstanding arises. Different qubit technologies have very different energy requirements.

Superconducting qubits: IBM and Google quantum-computing systems use superconducting qubits. To operate, these systems must be maintained at temperatures very close to absolute zero. IBM’s quantum processors are kept only about 0.01 degree above absolute zero, requiring an extremely complex cryogenic system [6].

Trapped-ion qubits: These systems require ultra-high-vacuum and laser systems to operate.

Neutral-atom qubits: These also depend on laser and vacuum systems.

Photonic qubits: Although the processors in these systems do not necessarily require ultra-low temperatures, they rely on highly sophisticated optical systems, and some designs may require cryogenic detectors.

Silicon spin qubits: Although these are an integrated form of conventional semiconductor technology, they still require a cryogenic environment for the qubit system to function effectively.

Because of this hardware diversity, broad claims about the electricity demand of quantum computers can be misleading unless a specific technology is identified, since each platform has a very different energy profile. However, in every quantum system, the electricity required by the supporting hardware is likely to be no less—and may well be greater—than that required by conventional computing systems.

 

Measure the Result, Not Just the Machine

Directly comparing conventional and quantum computers is difficult because their electricity consumption varies enormously from machine to machine. A laptop may run on only a few watts. A desktop computer may use a few hundred watts. A server may require several kilowatts, while a leading supercomputer may consume many megawatts. So asking how many watts a particular computer uses is not especially meaningful. The more useful question is: how much total energy is required to obtain each useful result? A small, noisy quantum computer may use much less electricity than a supercomputer, but if it cannot solve a practically useful problem accurately, its real energy efficiency is poor because it produces no usable result. Conversely, a large, reliable quantum computer may consume considerably more electricity while operating, but if it can solve in a few hours a problem that would take a supercomputer years, then the total energy used per solution may be far lower than for conventional computing.

This distinction is extremely important, and it makes the mobile-phone analogy relevant once again. In the mobile-phone era, the device in a person’s hand is not the main source of electricity demand; the network and cloud infrastructure are. Similarly, in the quantum-computing era, the qubit chip itself may not be the main energy consumer. Cooling, control systems, error correction, and classical supporting infrastructure may dominate. Research suggests that in large-scale quantum-computing systems, the total electricity consumption of a quantum data centre—including cooling—could become sufficiently large to be a significant concern. Studies also indicate that the electricity required to keep quantum-computing equipment cold may greatly exceed the energy required for the computation itself [7].

 

Power-Hungry Error Correction

One of the main challenges in quantum computing is quantum error correction. Qubits are extremely sensitive. Changes in temperature, fluctuations in electromagnetic fields, cosmic rays, small deviations in the quantum state, or even weak unwanted interactions with the environment can disturb them. Unless these disturbances are controlled, accurate results cannot be guaranteed. Preventing them requires continuous measurement, correction, and computation. Considerable energy may therefore be consumed even before a useful result is obtained, and a continuous supply of electricity is required. Google’s Willow chip has demonstrated an important advance in this area: as the system size increased, the error rate could be reduced significantly. Google itself has emphasised, however, that improving the quality of each qubit is more important than simply increasing the number of qubits [8]. The reason is that a fully fault-tolerant quantum computer may require many physical qubits to create a single reliable logical qubit. As the number of qubits rises, the requirements for measurement, control pulses, classical decoding operations, and cooling also increase—along with the demand for energy.

 

Can Energy Savings Be Achieved?

Can a quantum-computing system be energy-efficient? Many kinds of research are under way to answer this question. Some recent studies suggest that quantum computers may indeed save energy for certain problems, but this advantage does not arise automatically in every case. Researchers Meier and Yamasaki developed a theoretical framework for comparing the energy consumption of conventional and quantum computation. Using Simon’s problem, a well-known problem in quantum complexity theory, they showed that quantum computation could provide a dramatic energy-saving advantage [9]. Other research indicates that energy savings in quantum computing depend strongly on quantum-gate fidelity and the degree of quantum entanglement among qubits. For some current noisy quantum-computing systems, and for problems involving moderate levels of entanglement, conventional tensor-network simulation may be more energy-efficient than real quantum hardware [10]. Thus, energy efficiency in quantum computing is conditional rather than universal.

Will future quantum-computing data centres therefore place additional pressure on electricity demand? In the near future, probably not on a global scale. Conventional computing and AI data centres are already operating worldwide, while quantum data centres remain limited and relatively rare. In the longer term, however, the answer will depend on how quantum technology is used. If quantum computers are reserved for selected high-value problems—such as drug discovery, materials design, quantum chemistry, nuclear physics, logistics, cryptography, and climate modelling—the average energy used per solution may actually fall. But if quantum data centres expand inefficiently, with large cryogenic overheads, poor utilisation, and excessive error-correction burdens, they could create substantial new electricity demand.

We already have the example of the pre-mobile and post-mobile eras. In the age of mobile technology, mobile phones did not merely add the electricity needed to charge billions of devices; they created an entire world of always-on communication, cloud storage, video streaming, digital payments, location services, and AI-driven applications. Electricity demand in this ecosystem has grown at a remarkable rate. In the same way, quantum computing may add not only the electricity needed to run quantum processors but an entirely new layer of specialised data-centre infrastructure. Whether that infrastructure ultimately proves sustainable will depend on how the technology is used.

Five Priorities for Sustainable Quantum Computing

If quantum computing is not used with the right level of efficiency, the entire system could become extremely electricity-intensive. Energy efficiency therefore needs to be considered from the very beginning of large-scale quantum-computing development. Five key priorities can be identified.

1. Make energy efficiency a core metric: Energy efficiency should be treated as a major design criterion for quantum hardware. Quantum computing should be assessed not only by qubit count and gate fidelity, but also by joules per logical operation, joules per algorithm, cooling energy per qubit, and energy per useful result. The real question should be how much total energy the entire quantum system consumes to produce a correct and useful answer.

2. Improve cryogenic efficiency: In superconducting and many spin-qubit systems, cooling is one of the largest sources of energy consumption. Improvements in cryostats, thermal packaging, heat-leak reduction, cryogenic multiplexing, and control electronics should focus on reducing low-temperature thermal loads. The electricity required for cooling may exceed the energy used by the computation itself.

3. Make error correction more efficient: If thousands of physical qubits are needed for every logical qubit, the energy demand of large machines will rise rapidly. Higher qubit fidelity, better error-correcting codes, faster measurements, efficient decoding algorithms, and lower-power control electronics could reduce the electricity demand of quantum computing.

4. Use quantum computers selectively: Quantum processors should not be used for tasks that conventional computers can perform more efficiently. In this sense, quantum computers are better regarded as specialised accelerators than as universal replacements for classical data centres.

5. Connect quantum computing to an appropriate electricity system: The strategies now being discussed for AI and cloud data centres—renewable electricity procurement, energy storage, grid-aware scheduling, efficient cooling, transparent reporting, and improved power-usage effectiveness—should also be applied to quantum computing. Because data-centre electricity demand is expected to grow rapidly by 2030, future quantum systems should be planned in locations where clean electricity, cooling infrastructure, and adequate grid capacity are readily available [5].

Alongside these priorities, we also need a clear and transparent understanding of quantum computing. Public discussions often compare only the quantum chip with a supercomputer, while ignoring the energy costs of refrigeration, control electronics, calibration, classical pre-processing, and post-processing. A fair comparison must consider both the complete system and the complete task. Quantum computing will be genuinely energy-efficient only when its algorithmic advantage outweighs its real physical and infrastructure overheads.

 

Quantum computers have the potential to reduce electricity demand by solving selected problems in far fewer computational steps than conventional computers. For certain scientific, industrial, and cryptographic tasks, they may therefore become much more energy-efficient than classical supercomputers. But this advantage is not automatic. Quantum computing can become part of a more sustainable computing future only if energy efficiency is built into the technology from the outset. Better qubits, lower cooling overheads, more efficient error correction, selective use of quantum processors, and reliable electricity systems for future quantum data centres could make quantum computing energy-efficient. Uncontrolled expansion, however, could turn the entire system into a massive electricity consumer—one whose demand may be particularly difficult for developing countries to meet.

 

References

[1] Statista, "Global electricity consumption 1980–2024," 2026. [Online]. Available: https://www.statista.com/topics/6462/global-electricity/

[2] Worldmapper, "Mobile subscriptions 1990," 1990 data summary. [Online]. Available: https://worldmapper.org/maps/mobile-subscription-1990/

[3] International Telecommunication Union, "Facts and Figures 2024: Subscriptions," 2024. [Online]. Available: https://www.itu.int/itu-d/reports/statistics/2024/11/10/ff24-subscriptions/

[4] International Energy Agency, "Data centres and data transmission networks," 2023. [Online]. Available: https://www.iea.org/energy-system/buildings/data-centres-and-data-transmission-networks

[5] International Energy Agency, "Energy demand from AI," 2025. [Online]. Available: https://www.iea.org/reports/energy-and-ai/energy-demand-from-ai

[6] IBM, "IBM Quantum Computing: Hardware and roadmap," 2026. [Online]. Available: https://www.ibm.com/quantum/hardware

[7] M. J. Martin, C. Hughes, G. Moreno, E. B. Jones, D. Sickinger, S. Narumanchi, and R. Grout, "Energy use in quantum data centers: Scaling the impact of computer architecture, qubit performance, size, and thermal parameters," IEEE Transactions on Sustainable Computing, vol. 7, no. 4, pp. 864–874, 2022. [Online]. Available: https://arxiv.org/abs/2103.16726

[8] Google Quantum AI, "Meet Willow, our state-of-the-art quantum chip," 2024. [Online]. Available: https://blog.google/technology/research/google-willow-quantum-chip/

[9] F. Meier and H. Yamasaki, "Energy-consumption advantage of quantum computation," PRX Energy, 2025. [Online]. Available: https://arxiv.org/abs/2305.11212

[10] D. Jaschke and S. Montangero, "Is quantum computing green? An estimate for an energy-efficiency quantum advantage," arXiv preprint, arXiv:2205.12092, 2022. [Online]. Available: https://arxiv.org/abs/2205.12092