Contributed article by Parham Pashaei, Director of Quantum Solutions, Strategy & Partnerships, Quantum Algorithms Institute

It’s no secret that the use of Artificial Intelligence (AI) is increasing worldwide, but that comes at a cost that not many of us are thinking about: energy consumption and CO2 emissions. For example, each ChatGPT query consumes nearly 10 times as much electricity as a Google search and produces approximately 4.32 grams of CO2. As the AI industry searches for more sustainable and efficient ways to consume energy, quantum computers offer several solutions which the world should consider tomake AI energy consumption more efficient and sustainable.

AI needs data centres in order to function and, to support AI, data centres are consuming more energy now than ever. Barclay’s Research estimates that data centres account for 3.5% of US electricity consumption today. As the use of AI increases, data centre energy consumption is also predicted to grow. US data centre energy use could be above 5.5% in 2027 and more than 9% by 2030.

This trend is also being observed globally. According to the International Energy Agency (IEA), in 2022, data centres consumed 1.65 billion gigajoules of electricity — about 2% of global demand.  That is a lot of power being used! By 2026, the IEA projects that data centres’ energy consumption will have increased by between 35% and 128%.

There is a clear power consumption challenge to be addressed. Within this power consumption challenge, quantum computers are emerging as a potential avenue to help achieve more efficient power consumption for AI computation.

But first, what are quantum computers? They are a fundamentally new type of computers that operate based on the principles of quantum mechanics. While classical computers use bits—represented as 0s and 1s—for computation, quantum computers use quantum states, known as qubits. Once we have a large number of these isolated quantum states connected well together, we may be able to solve some problems that cannot be solved by traditional computers. Quantum computers also provide a computing platform that is attractive for solving AI models.  

So, given quantum computers’ unique characteristics, can they help solve the problem of AI’s high-power consumption?

In some research examples, quantum computers were years faster than today’s fastest supercomputers, but, surprisingly, they did not consume way more power than a supercomputer.  A classical supercomputer consumes a lot of energy, typically several megawatts (MW) of power. For context, this could be enough to power a small town of about 1,000 people. A quantum computer, however, generally consumes energy in the order of kilowatts, about the amount of power used by a domestic electric oven. Now compare a supercomputer taking thousands of years and megawatts of energy to solve a problem with a quantum computer resolving the same problem in minutes and consuming just kilowatts!

Most of quantum computers’ power consumption is spent on the computer’s infrastructure:   the quantum circuit itself consumes little energy. As quantum computers get rapidly more powerful their power consumption scales at a much lower rate.

So, how could quantum computers improve power consumption at AI data centres? Researchers at Cornell University discovered a new quantum computing-based framework specifically for reducing energy consumption at AI data centres. The framework looked at what would happen if classical computing methods and quantum computing were combined into one framework to power AI data centres while being more versatile and efficient. The research found that the framework could lower energy consumption by up to 12.5% and reduce carbon emissions by up to 9.8%.  While these results are promising, AI energy use problems may not be completely solved by quantum computing alone. Currently, quantum approaches can potentially improve the performance of certain AI tasks.

As quantum computing technology advances, it creates opportunities for more ways to address the rapidly increasing energy demands of AI data centres. If both the AI and quantum sectors work to minimize energy needs, our future technology ecosystem could be sustainable.

March 28, 2025