Artificial intelligence ( AI ) is often framed as a driver of efficiency and progress. But it also presents a profound sustainability challenge. While high-performance computing ( HPC ) enables critical advances in climate modelling, drug discovery and renewable energy development, it consumes staggering amounts of energy.

The tech sector already accounts for more than 3% of the world’s greenhouse gas emissions, and the rapid rise of generative AI is poised to expand its carbon footprint and push power grids to their limits. Training Grok-4 alone reportedly required 310 gigawatt-hours of electricity – enough to power a town of 4,000. And Google’s AI could eventually consume as much electricity as all of Ireland.

In response to surging energy demand, the industry has largely favoured incremental measures, such as recovering heat from HPC cooling systems, experimenting with early “frugal AI” approaches, or promoting the expansion of nuclear capacity. While useful, these efforts fail to address the underlying issues.

The challenge is twofold: cutting the industry’s environmental footprint by curbing low-value AI workloads, and developing computational architectures that advance the United Nations Sustainable Development Goals ( SDGs ). One promising approach, known as Sustainable Quantum AI ( SQAI ), seeks to harness quantum technologies to reduce HPC energy consumption.

Despite their ubiquity, quantum technologies have not realized their full potential. For example, ammonia synthesis – a triumph of quantum chemistry that helped feed billions of people – now urgently needs reinvention to support truly regenerative agriculture.

Quantum science could advance sustainability by transforming not only agriculture, food and health systems, but also cutting-edge technologies like AI. Quantum computers rely on information units known as qubits, which enable them to make certain calculations far more efficiently than conventional HPC systems, and this has been shown to have a direct effect on energy use.

A central focus of these efforts is the development of quantum processing units ( QPUs ) that consume far less energy than conventional supercomputing hardware. Operating essentially at room temperature, neutral atom QPUs eliminate the need for power-intensive cooling.

Neutral-atom devices such as the Orion system, developed by Pasqal ( one of us is CEO ), consume no more than 3.5 kilowatts of electrical power, whereas a single classical supercomputer like Frontier requires roughly 22 megawatts. While the time needed to solve some problems still varies, and many applications will rely on a combination of quantum and conventional processors, energy use could still be reduced to a small fraction of today’s levels.

The contrast is clearest when viewed through lifecycle analysis ( LCA ), an essential tool for transparency and accountability. Since a full assessment must include emissions from hardware manufacturing, which can account for up to 80% of digital technologies’ carbon footprint, the comparison is striking.

An HPC system like the French Joliot-Curie supercomputer emits about 0.2 metric tonnes of CO2-equivalent per hour, including hardware amortization and electricity use, even with France’s low-carbon power supply. By contrast, the neutral atom-based Orion system produces around 2.1 kilogrammes of CO2-equivalent per hour. Taken together, these figures suggest that AI-related emissions will start to decline before the arrival of fully reliable, fault-tolerant quantum computers.

But energy efficiency is only part of the story. Neutral atom architectures are particularly well suited to tackle complex sustainable development challenges that remain beyond the reach of conventional HPC systems, especially large-scale combinatorial optimization problems, such as those associated with molecular design, energy systems and transportation logistics.

To test this idea, Pasqal launched two SQAI challenges in 2023 and 2025. Notably, both the company and the challenges were named after the 17th-century French mathematician Blaise Pascal, a pioneer of probability theory and inventor of one of the earliest mechanical calculators, the Pascaline. Less well known, but no less significant, was his final endeavour: the creation of the Carrosses à Cinq sols, among the first public transportation systems.

With that legacy in mind, the SQAI challenges required each applicant to demonstrate both a concrete sustainability-related use case and a lower energy footprint than a conventional HPC approach, using technologies available today. The response was remarkable, drawing more than 2,000 participants from over 60 countries.

Early results point to meaningful progress across multiple SDGs. Clean and affordable energy emerged as a natural focus, as participating teams used hybrid quantum-AI methods to optimize power grids and wind-farm layouts. Some teams explored quantum-based approaches to wildfire management and typhoon modelling, while others worked on methods for accelerating antimicrobial discovery and improving the sustainability of food production.

To be sure, today’s quantum computers are not yet capable of addressing real-world, complex problems. Optimizing public-transport networks or simulating molecular dynamics requires handling thousands of interconnected variables, far beyond the encoding capacity of current QPUs. Even so, the rapid growth of qubit capacity highlights the largely untapped potential of quantum computing.

For SQAI to fulfil its promise, the industry must commit to clear criteria, transparent metrics, and rigorous LCA methodologies that account for emissions across the entire lifecycle, from design and manufacturing to operation and disposal. Greater collaboration between companies, researchers and institutions, particularly when fostered by independent bodies, is crucial to ensuring that resources are directed towards applications that deliver genuine, measurable sustainability benefits.

SQAI represents a chance to steer AI away from its unsustainable trajectory, challenging scientists and entrepreneurs to develop and scale energy-efficient quantum solutions. As confidence in carbon-neutral commitments continues to erode, sustainability may be driven less by corporate pledges than by the ambitions of a new generation of innovators and the massive electricity bills that will soon hit data centres around the world.

Bertrand Badré is the chair of the Project Syndicate advisory board, CEO and founder of Blue like an Orange Sustainable Capital and a former managing director of the World Bank; Loïc Henriet is the CEO of Pasqal; and Etienne de Rocquigny is the co-founder of the Sustainable Quantum AI World Challenge and Rerum Novairum.

Copyright: Project Syndicate