The Alps are cold, but they’re nothing compared to a superconducting quantum computer. At the very low temperatures (or isolated conditions) required for working Quantum Processing Units (QPUs), our standard understanding of energy and thermodynamics doesn’t apply anymore. Meanwhile, achieving those extreme conditions requires powerful classical technologies, like cryostats, lasers or vacuum chambers. This creates lots of open questions about the energetics of quantum technologies, right from the fundamental quantum scale to that of room-sized classical machines. I took a trip to the Alps to try and find some answers.

The Quantum Energy Initiative (QEI) organised their second ever workshop this January. The initiative aims to create a forum for stakeholders in the quantum industry, including both academia and industry, to explore issues of energy in quantum technologies. The workshop took place in Grenoble, France; a city 360-degree surrounded by mountains. While there wasn’t much skiing on the slopes, there was plenty of science on the slides, and in this blog I’ll outline some insights from the week.

Cool ideas on the quantum scale

What temperature is a qubit? There is no straightforward answer to this question, but there is lots of exciting research in the growing field of quantum thermodynamics to understand energy exchanges on the quantum scale, with interesting implications for quantum technologies.

A talk by Lindsey Oftelie from NEST, Pisa, was about “dynamic cooling”, in which logic gates are used to shift energy from some qubits to others. This cools the desired set of qubits, reducing their noise. While the original idea was proposed over two decades ago, it was discarded because the most popular platform at the time (NMR) worked at high temperatures. Now that quantum computers are working at low temperature, Oftelie has brought dynamic cooling back into fashion with some promising results and experiments.

Meanwhile Géraldine Haack, University of Geneva, gave an invited talk on a counterintuitive idea: the uncontrolled dissipation of a qubit’s energies in the environment can actually be a resource for figuring out the qubit’s state. Usually, characterising a quantum state needs lots of measurements, which quickly becomes unfeasible. Haack’s result could lead to new techniques for mitigating errors using knowledge of the dissipation mechanism instead of, or together with, those measurements.

While representing OQC at the QEI, I gave a talk about some of my PhD research at the University of Oxford on quantum thermodynamics. The amount of useful energy (work) you can extract from a superposition of quantum states is directly related to the fundamental information-based property about whether they are perfectly distinguishable. I spoke about how to reconcile two different approaches to thermodynamics, showing that it is impossible to build a “universal work extractor” that deterministically gets useful energy from qubit superpositions; it has to be a special-purpose machine.

A methodology for estimating energy

Another theme of the workshop was estimating the energy consumption of quantum computers and how it will scale. Two of the founders of the Quantum Energy Initiative, Alexia Affuèves from MajuLab, Singapore and Robert Whitney from CNRS, Grenoble (who led this year’s workshop organisation), and collaborators, recently developed a methodology for full-stack quantum computing resource estimations. It is called MNR: Metric, Noise, Resource.

In the paper proposing MNR, they analyse a specific (albeit idealised) case-study to minimise the power consumption of an error-corrected quantum computer. By drawing a comparison with classical computing, they outline and determine conditions for a “quantum energy advantage” that is different from standard quantum computational advantage.

At the workshop, academic and industry researchers presented their results from applying the methodology to make the first steps of energy resource estimations, with the long-term aim of optimising energy resources of future quantum computers. This included the different qubit modalities of superconducting, photonic, neutral atom and silicon hardware, as well as software and algorithms.

The macro-scale challenge ahead

Workshops like this one create a platform to share ideas and make progress. Yet, anticipating and engineering the resources required for technologies before they exist is not an easy task. While I set out to the Alps looking for answers, I came back with more questions.

For instance, how should we compare the energy consumption of quantum computers with classical ones? Classical computing is matured while quantum is in early stages; and the way we run tasks on future quantum computers will also likely be different to how tasks are run on today’s CPUs and GPUs. More questions arise when considering the broad range of possibilities for the architecture of fault-tolerant, large scale quantum computers. Even those built with the same qubit modality could end up with very different structures and resource requirements. At OQC, we’ve been considering the role that quantum computers will play in the energy landscape when integrated into data centres, which you can read more about in our recent report.

As the QEI moves forward, they aim to bring more people from the macro-scale side of quantum into the community — the engineers working on the enabling technologies, such as cryostats and control hardware. Now that we’ve got some quantum energy science going, perhaps next time we will see some quantum energy engineering!