El Capitan Supercomputer: Power Play in Quantum Computing Credit: LLNL/Garry McLeod
The advancement of large quantum computers offers the potential to solve complex problems beyond the reach of today’s most powerful classical supercomputers. However, this leap in capability may come with increased energy demands.
Currently, most existing quantum computers are limited in size, with less than 1,000 qubits. These fragile qubits are susceptible to errors, hindering their ability to tackle significant issues, like aiding in drug discovery. Experts agree that to reach practical utility, a Fault-Tolerant Quantum Computer (FTQC) must emerge, with a much higher qubit count and robust error correction. The engineering hurdles involved in this pursuit are substantial, compounded by multiple competing designs.
Olivier Ezratty, from the Quantum Energy Initiative (QEI), warns that the energy consumption of utility-scale FTQCs has been largely overlooked. During the Q2B Silicon Valley Conference in Santa Clara, California, on December 9, he presented his preliminary estimates. Notably, some FTQC designs could eclipse the energy requirements of the world’s top supercomputers.
For context, El Capitan, the fastest supercomputer globally, located at Lawrence Livermore National Laboratory, draws approximately 20 megawatts of electricity—three times that of the nearby city of Livermore, which has a population of 88,000. Ezratty forecasts that FTQC designs scaling up to 4,000 logical qubits may demand even more energy. Some of the power-hungry designs could require upwards of 200 megawatts.
Ezratty’s estimates derive from accessible data, proprietary insights from quantum tech firms, and theoretical models. He outlines a wide energy consumption range for future FTQCs, from 100 kilowatts to 200 megawatts. Interestingly, he believes that three forthcoming FTQC designs could ultimately operate below 1 megawatt, aligning with conventional supercomputers utilized in research labs. This variance could significantly steer industry trends, particularly as low-power models become more mainstream.
The discrepancies in projected energy use stem from the various strategies that quantum computing companies employ to construct and maintain their qubits. For instance, certain qubit technologies necessitate extensive cooling to function effectively. Light-based qubits struggle with warm light sources and detectors, leading to heightened energy consumption. Similarly, superconducting circuits require entire chips to be housed in large refrigeration systems, while designs based on trapped ions or ultracold atoms demand substantial energy input from lasers or microwaves to precisely control qubits.
Oliver Dial from IBM, known for superconducting quantum computers, anticipates that his company’s large-scale FTQC will need approximately 2 to 3 megawatts of power, a fraction of what a hyperscale AI data center could consume. This demand could be lessened through integration with existing supercomputers. Meanwhile, a team from QuEra, specializing in ultracold atomic quantum computing, estimates their FTQC will require around 100 kilowatts, landing on the lower end of Ezratty’s spectrum.
Other companies like Xanadu, focusing on light-based quantum technologies, as well as Google Quantum AI, centered on superconducting qubits, have opted not to comment. PsiQuantum, another light-based qubit developer, was unavailable for a response. New Scientist has made multiple attempts for their insights.
Ezratty also pointed out that traditional electronics responsible for directing and monitoring qubit operations could result in additional costs, particularly for FTQC systems where qubits need further instructions to self-correct errors. This complexity necessitates understanding how these algorithms contribute to energy footprints. The operational runtime length of quantum computers adds another layer, as energy savings from fewer qubits might be negated if longer operation times are needed.
To effectively measure and report the energy consumption of machines, the industry must establish robust standards and benchmarks. Ezratty emphasizes that this is an integral element of QEI’s mission, with projects actively progressing in both the United States and the European Union.
As the field of quantum computing continues to mature, Ezratty anticipates that his research will pave the way for insights into FTQC energy consumption. This understanding could be vital for optimizing designs to minimize energy use. “Countless technological options could facilitate reduced energy consumption,” he asserts.
Topics:
Source: www.newscientist.com
