The first event was Bizkaia BBVA. Banks in Quantum Days, organized by the Government of Biscay and BBVA, together with BIQAIN, which showcased the strong integration of quantum solutions in the financial sector. The second was a conference titled Quantum Computing and Hydrogen: The Future of Green Energy, organized by the Basque Hydrogen Corridor and Ibermática Fundazioa, where the potential of quantum computing as a tool to advance renewable energy development, specifically in the hydrogen sector, was discussed. At first glance, these two conferences may seem unrelated, but they are intertwined by a key concept: digital sustainability.

THE NEW STAPLER IN THE OFFICE

The enormous importance of AI in accelerating productivity and knowledge is undeniable. The AI systems we knew 20 years ago are now becoming obsolete. A decade ago, a company could differentiate itself from the competition by implementing AI systems. Today, that opportunity no longer exists: having a generative AI system is as essential as having email or PowerPoint. The reason is clear: AI has become an indispensable technology, just as steam engines were during the Industrial Revolution.

At the same time, as customers and citizens, we demand that our companies, financial institutions, and governments uphold reasonable environmental, social, and governance (ESG) responsibilities. However, with the current graphic processor-based AI systems, some Sustainable Development Goals (SDGs) (such as climate action and responsible production and consumption) are completely at odds with these technologies.

PROBLEM OR SOLUTION?

Ironically, AI can be a key tool for achieving the United Nations' Sustainable Development Goals (SDGs). Through automation and big data analysis, AI can contribute to climate action by optimizing resource use and improving energy efficiency across industries. However, the energy consumption required by today’s large foundational AI models has a significant, if not dramatic, environmental impact.

For example, it is estimated that the training of current ChatGPT models consumes as much energy as the annual consumption of 40 households. Generating a 100-word text in ChatGPT consumes, on average, 519 milliliters of water, equivalent to a standard water bottle, per query. The reason such high energy is needed is simple: we must model knowledge from a chaotic data environment with the least possible uncertainty, and in information theory, this means "cooling" complex data environments filled with high entropy (in computing jargon)or "very high temperatures"in a physical analogy. The higher the temperature, the greater the disorder, which translates to extracting less information. Current classical computing systems are not efficient in this task; they require many bits and millions of parameters to mitigate uncertainty and select the correct response among billions of possibilities. And this is computationally very expensive.

In this process, another major issue is the energy supply to these analytical systems. Data centers, where the servers operating AI models are housed, are responsible for nearly 1% of global electricity consumption, according to the International Energy Agency (IEA). The expansion of AI and other digital services contributes to the growing problem, although companies like Google and Microsoft have managed to power their data centers with 100% renewable energy. However, new energy sources like solar and wind are not always available, which has led to the consideration of modular nuclear reactors for power supply, ultimately creating yet another challenge to solve.

0S AND 1S ARE FALLING BEHIND

Regarding computational complexity, if a technology capable of evaluating all possible states of a system (the answers to a complex problem) in parallel and within just a few computing cycles existed, computational complexity would radically decrease. As a result, uncertainty would be resolved more efficiently, essentially "cooling down" the system better.

And this property is precisely what quantum systems offer us. This hardware, based on the exponential growth of quantum storage capacity and the advantages of quantum parallelism (where we can apply a function to all states in a single cycle), results in computationally more efficient systems that, consequently, demand less energy.

Theoretically, many quantum functions are even reversible, maintaining the quantum "coherence" of information. Since this information is not lost, entropy does not change, meaning energy would not be consumed; although, in contrast, cooling these systems, error control, and measurements do require energy.

On the other hand, energy supply to data centers also requires new technology. And this is where hydrogen comes into play. Hydrogen is considered one of the key solutions for renewable energy storage and can play a crucial role in the transition to a cleaner and more sustainable energy system.

On one hand, it acts as an energy reservoir for electricity generated by intermittent renewable sources such as solar or wind power, allowing excess energy to be converted into hydrogen via electrolysis and stored for use when renewable generation is low, thus providing flexibility to the energy system.

On the other hand, hydrogen is a key alternative for sectors that are difficult to electrify, such as heavy transport and industry, where it can replace fossil fuels and contribute to decarbonization (especially in steel production, chemical industries, and refineries) helping to reduce emissions and meet global climate goals.

However, hydrogen presents several challenges, such as its production through electrolysis using renewable energy, which requires large amounts of energy and has suboptimal efficiency compared to other, cheaper forms of hydrogen. Its storage and transportation are also problematic, as they require high pressure or extremely low temperatures—processes that are energy-intensive and costly, as well as posing safety risks due to hydrogen's high flammability.

Finally, the overall efficiency of the hydrogen life cycle is low, as energy losses occur at each stage (production, storage, and reconversion), making it less efficient than other storage technologies, such as batteries based on alternative molecular components.

In this context, we face a new computational challenge: we need an efficient, comprehensive, and sustainable computing system that facilitates the simulation and design of more efficient materials for the electrolysis process, a highly complex problem for current computational systems. For example, optimizing chemical processes and helping to find more efficient methods for hydrogen production using less energy requires simulations of multi-physical systems, which today take months to run on classical computing systems, even on current supercomputers. The development of new materials to improve hydrogen storage optimization, as well as designing more efficient compression and liquefaction systems, would reduce energy consumption in these processes.

All these problems stem from the difficulty of modeling analog systems of particles, waves, and fluids, which, in reality, are nothing more than quantum systems. And as Poplavsky famously stated: "Classical computers cannot efficiently simulate quantum systems," to which Feynman responded, however, that "since quantum computers are quantum systems, they can (by definition) simulate quantum systems. Therefore, quantum advantage exists." Clear as day. Quantum computers are also analog systems.

THE NEW ERA OF AI

Quantum computing is an area experiencing rapid progress, completely transforming how we understand computing and problem-solving in optimization. Chemical, fluid, and gas simulations (in short, particles and waves) related to the generation, improvement, storage, and processing of hydrogen are extremely difficult. Advancing in the simulation of chemical and physical systems based on quantum computing models is necessary to deepen our understanding of interactions between complex molecules and their implications in the development of new products, processes, and materials.

If we want a sustainable computational technology, we need to move forward with other paradigms, such as quantum computing or neuromorphic computing, and undoubtedly direct our efforts toward quantum AI. The key is to develop quantum AI technologies that are more efficient than current classical AI systems while simultaneously hybridizing the use of renewable energy with new technologies like hydrogen. This approach will help promote responsible governance, supporting the evolution of these two interlinked technologies, which complement each other and are essential for a sustainable future for our country.