Abstract
The rapid advancement of quantum hardware brings a host of research opportunities and the potential for quantum advantages across numerous fields. In this landscape, quantum circuit simulations serve as an indispensable tool by emulating quantum behavior on classical computers. They offer easy access, noise-free environments, and real-time observation of quantum states. However, the sustainability aspect of quantum circuit simulation is yet to be explored. In this paper, we introduce for the first time the concept of environmental impact from quantum circuit simulation. We present a preliminary model to compute the $CO_2e$ emissions derived from quantum circuit simulations. Our results indicate that large quantum circuit simulations (43 qubits) could lead to $CO_2e$ emissions 48 times greater than training a transformer machine learning model.
Introduction
Quantum computers are currently scarce, expensive, and "noisy" (prone to errors). To bypass these hurdles, researchers use quantum circuit simulators—classical software that emulates quantum behavior. While these simulators are typically evaluated based on fidelity and speed, their energy consumption is rarely discussed. Because simulating quantum states on classical hardware is computationally expensive—scaling exponentially with the number of qubits—the energy required grows at an alarming rate. This study aims to bring environmental impact into the conversation of quantum research, providing a framework to quantify the carbon footprint of these simulations.
Methodology
The authors propose a preliminary mathematical model to calculate Carbon Dioxide Equivalent ($CO_2e$) emissions. The formula factors in the number of processors used ($n$), average power draw per processor ($Pp$), total simulation duration ($T$), Carbon Intensity ($CI$), and Power Usage Effectiveness ($PUE$).
The study applies this model to two primary scenarios:
- Small-scale simulations: Running Variational Quantum Circuits (VQC) versus classical Convolutional Neural Networks (CNN) on the MNIST dataset.
- Large-scale simulations: Analyzing existing performance data from high-performance runs (40–43 qubits) on AWS and NVIDIA hardware.
Results and Analysis
Quantum vs. Classical Machine Learning
Even at small scales (8–16 qubits), Variational Quantum Circuits (VQCs) are significantly more carbon-intensive than classical CNNs with a comparable number of parameters. As shown in Figure 1, the emissions for VQCs rise sharply compared to the relatively flat energy consumption of classical counterparts.
Comparative Environmental Impact
The study reveals a stark contrast between the carbon footprint of quantum simulations and other common activities. A single 43-qubit simulation run (568.77 kg $CO_2e$) is more damaging to the climate than a transatlantic flight for one passenger (313.90 kg $CO_2e$).
As the qubit count increases, the impact skyrockets due to the exponential nature of the simulation task. For instance, emissions roughly double with each additional qubit:
- 40 Qubits: 54.23 kg $CO_2e$
- 41 Qubits: 122.91 kg $CO_2e$
- 42 Qubits: 274.74 kg $CO_2e$
- 43 Qubits: 568.77 kg $CO_2e$
Discussion and Limitations
While this study highlights a critical oversight in quantum research, the model is preliminary. It focuses primarily on dynamic power consumption during active processing and does not fully account for embodied emissions (hardware manufacturing) or granular cooling and memory energy costs. Furthermore, the results are based on State Vector Simulators; alternative methods like tensor network simulators might yield different footprints.
Conclusion
Sustainability should be a standard metric for evaluating quantum simulators, alongside traditional metrics like fidelity and resource usage. Future research should focus on developing "green" quantum simulation strategies, such as optimizing algorithms to reduce circuit depth or utilizing hardware in regions with lower carbon intensity.
Summary updated