How can we scale quantum computing in the most energy-efficient way?

Artificial intelligence (AI) is reshaping the digital world at breathtaking speed, powered by vast data centres, ultra-fast networks and ever-denser compute clusters, as each new generation of AI models requires significantly more processing power. Yet the world's digital infrastructure is already approaching physical and energy limits.

Today, a single cutting-edge AI chip can draw as much electricity as an entire household and large training runs can consume as much energy as an entire city. The International Energy Agency, therefore, expects global electricity consumption by data centres to roughly double by 2030.

This raises the question: how can we expand computing without expanding energy use?

Quantum computing is often cited as a partial answer, promising far greater computational power at a dramatically lower energy cost. But we rarely explain why, in which cases or compare what quantum platforms deliver the most computation per kilowatt-hour, despite significant differences.

For now, governments and industry must decide which architectures are not only scalable but energy-scalable: those that offer the highest computing power with the lowest energy consumption.

The starting point is physics. Any computer is a physical system that must obey the laws of thermodynamics, so information processing is never energy-free. At the heart lies Landauer’s principle, which tells us that whenever information is erased, a minimum amount of energy must be lost as heat, a limit that no technology can bypass.

This fundamental limit helps distinguish classical from quantum computers. In everyday computing, this limit is overshadowed by engineering inefficiencies, so we rarely notice it. But as problems become more complex, classical algorithms must erase a rapidly growing number of intermediate bits, which means that even in principle, their minimum possible energy use can grow exponentially.

Quantum algorithms handle information fundamentally differently. Their logic operations are reversible, allowing intermediate states to be “uncomputed” instead of erased. This allows a quantum computer to examine many possible solutions simultaneously before extracting a small final answer.

For the same complex problems, theory tells us that the minimum energy needed by well-designed quantum algorithms grows more slowly. In these regimes, quantum computers are not just faster – they can be exponentially more energy-efficient.

These thermodynamic advantages matter only if the hardware keeps the energy cost of computation close to the theoretical limit. Quantum processors sit in elaborate physical setups, from cooling to control electronics and in practice, this supporting infrastructure often dominates power consumption.

Whether a platform depends on large dilution refrigerators or runs at or near room temperature will be a major determinant of its long-term energy profile. Understanding energy efficiency requires comparing not just algorithms but the hardware platforms that implement them.

Superconducting quantum computers rely on circuits that work only at temperatures close to absolute zero and therefore sit in multi-stage dilution refrigerators.

In current cloud systems, total power consumption is typically around 25 kilowatts (kW) – most of that electricity goes into refrigeration and supporting equipment rather than quantum bits (qubits).

Neutral-atom quantum computers work differently: they use individual atoms as qubits, held in arrays of laser beams in ultra-high vacuum and controlled by light pulses. They operate at or near room temperature and do not require millikelvin cryogenics or a cooling process that requires refrigeration.

Existing neutral-atom devices with a few hundred qubits report total system power below 10kW. Even at today’s modest scales, two quantum computers with comparable processor sizes can, therefore, differ by roughly a factor of three in power draw, depending on whether they rely on extreme cooling.

Looking ahead to larger, fault tolerance at scale (which can continue operating despite physical disturbances and operational errors), technology roadmaps suggest that electrical power demand for full-scale quantum computers could differ by up to two orders of magnitude between architectures.

Energy-efficient quantum hardware matters most where better solutions directly cut real-world energy use. Its first benefit is computational: some tasks can be solved with far less energy than on classical machines.

Equally important, quantum computing can help design and run a greener, more sustainable future, especially in energy systems and mobility, where optimization challenges abound, from wind-farm layout and renewable-generation forecasting to transport planning and traffic management.

Here, the electricity for quantum algorithms is tiny compared with potential system-level savings, so even modest gains can yield large reductions in resource use and emissions.

Materials science and chemistry offer another powerful pathway: better batteries, catalysts and membranes for energy storage, hydrogen and industrial processes, yet many crucial simulations remain beyond classical hardware.

Many battery and industrial materials are too complex for today’s computers to simulate accurately. Quantum computing can model them more faithfully, accelerating better batteries, more efficient catalysts and lower-loss membranes — reducing overall energy demand.

These advances also feed into AI. Training and serving large models depend on many intertwined hardware, software and architecture choices and their energy use is rising.

Quantum-inspired and quantum-based optimization is being explored so that future systems deliver equal or better performance with fewer operations and less energy per task.

Whether quantum computers will significantly speed up training frontier models remains an open question but using quantum tools to streamline AI workflows is already a promising direction.

As global computational demand accelerates, quantum computing should be seen not as a niche research pursuit but as a necessary pathway to sustain digital progress without increasing energy consumption.

Realizing this potential requires large, coordinated efforts across research, industry and the public sector, as well as sustained public and private investment to address key technical challenges such as large-scale error correction.

As well as efficiency, quantum processing offers a new way to master complexity, potentially redefining the architecture of the digital economy. Whether this future is environmentally viable depends on decisions today.

Governments and industry must ask: which technologies can scale without locking the digital world into high power demand?

If energy efficiency is to become a strategic priority, it must be embedded as a core design principle across the entire quantum ecosystem – from research funding to infrastructure planning. Only then can quantum computing rise alongside AI while keeping our digital future within the limits of the physical world.