Some quantum computers might need more power than supercomputers

Large quantum computers may be able to solve problems that are impossible even for the best traditional supercomputers, but some of them may require much more energy to do so than these supercomputers.

Existing quantum computers are relatively small: most have fewer than a thousand building blocks called qubits. They are also prone to errors during operation due to the fragility of these qubits. This makes these computers incapable of solving the economically and industrially important problems they were expected to excel at, such as aiding in drug discovery. Researchers largely agree that really useful quantum computers must have a radically higher number of qubits and the ability to correct errors, making them fault-tolerant quantum computers (FTQC). But achieving this goal remains a challenging engineering challenge, in part because there are several competing designs.

Olivier Ezratty V Quantum Energy Initiative (QEI)international organization, says one of the overlooked issues when building utility-scale FTQCs is their potential energy consumption. At the Q2B Silicon Valley Conference in Santa Clara, California, December 9 he introduced preliminary estimates of this. Amazingly, some FTQC projects have outperformed the world's largest supercomputers in power consumption.

fastest supercomputer in the worldEl Capitan, at Lawrence Livermore National Laboratory in California, uses about 20 megawatts of electricity, about three times the power consumption of the nearby city of Livermore, population 88,000. Ezratti estimates that the two FTQC projects, which scale to 4,000 logic or error-corrected qubits, will require even more. The most energy-hungry of them may require up to 200 megawatts of energy.

Based their estimates on publicly available data, proprietary information from quantum computing firms, and theoretical modelsEzratti identified a wide range of possible energy implications for future FTQCs that range from 100 kilowatts to 200 megawatts. Notably, Ezratti estimates that the three FTQC projects currently in development will ultimately require less than 1 megawatt of power, which is comparable to typical supercomputers used in research centers. In his opinion, this spectrum could influence the development of the industry, for example, increasing the market for quantum computing if less power-intensive designs begin to dominate.

The large difference in projected power consumption primarily reflects the variety of competing ways in which quantum computer companies create and use qubits. In some cases, power consumption is driven by the need to keep various parts of the device cool, such as some light qubits, where light sources and detectors perform worse when warm. Ezratti says this can be especially energy-intensive. In other cases, such as qubits made from superconducting circuits, entire chips must be placed. in giant refrigeratorswhile quantum computers, based on trapped ions or ultracold atoms, require energy for the lasers and microwaves that drive the qubits.

Oliver Dial at IBM what does superconducting quantum computerssays he believes the company would need just under 2 or 3 megawatts to operate a large-scale FTQC. That's only a fraction of what is projected to be needed for hyperscale AI data centers, and could be even less if FTQC were integrated with an existing supercomputer, Dial said. The team at ultracold atom quantum computing company QuEra estimates its FTQC will require about 100 kilowatts, which is at the lower end of Ezratti's spectrum.

Xanadu, which builds quantum computers based on light, and Google Quantum AI, whose quantum computers are based on superconducting qubitsdeclined to comment. PsiQuantum, which also makes qubits from lightdidn't answer New scientistrequest for comments.

Ezratti says there are also many costs associated with the traditional electronics that are used to control and monitor qubits, especially when it comes to FTQC, where qubits are given additional instructions to detect and correct their own errors. This further complicates matters because it means that the details of error correction algorithms also affect the devices' power consumption. Additionally, there is the issue of how long a quantum computer must run to complete an operation, since the energy savings that come from using fewer qubits may be offset if they have to run longer.

To untangle all these factors—the base cost of energy to create qubits, the cost of cooling and controlling them, and the cost and runtime of quantum software—the industry must develop standards and tests for determining and reporting the energy footprint of its machines, Ezratti says. This is part of QEI's mission. According to him, similar projects are being implemented in both the United States and the European Union.

Just as the entire quantum computing industry is still evolving, Ezratti says his work is in the early stages and should lead to efforts to better understand FTQC power consumption and use that understanding to reduce it. “There are many, many technical options that can help reduce the energy footprint.”