Interview with Prof. Mateo Valero, Director, Barcelona Supercomputing Center – Centro Nacional de Supercomputación (BSC-CNS)

13 Jan Interview with Prof. Mateo Valero, Director, Barcelona Supercomputing Center – Centro Nacional de Supercomputación (BSC-CNS)

What factors have enabled Barcelona to achieve such a high level of activity in emerging technologies like supercomputing, gaming, ICT and health sciences?

There are several factors that contribute to this. First, the universities in the region are of a very high standard, then it’s essential to focus on specific areas. Catalonia, for example, has been focusing on the health sector for some time. This includes establishing specialized centers and launching programs that attract researchers from outside Catalonia to conduct their investigations here. Research centers have been created to support these initiatives and this effort has grown significantly. Today, Catalonia is one of the most important hubs for health research.

Additionally, there are other fields where we excel. For instance, we have the synchrotron, the Institute of Photonic Sciences and the Barcelona Supercomputing Center. Each of these centers has a unique origin. The Barcelona Supercomputing Center (BSC), for example, is a collaboration between the Spanish government, the Catalan government and the Polytechnic University of Catalonia. It began 40 years ago as a smaller center and 20 years ago, it was proposed to elevate it to the next level, becoming the national center. Starting with 70 staff, we now have over 1,200 people, making it the largest supercomputing research center in Europe.

What steps will be necessary to establish BSC as one of Europe’s artificial intelligence (AI) super-factories in 2025 and which specific areas of the AI ecosystem should BSC prioritize to maintain Spain and the EU’s position as a frontrunner in this technology?

AI and supercomputing are deeply interconnected. AI began advancing significantly when sufficient computational power and data became available. Once this happened, we started seeing remarkable developments, like computers winning chess matches. Perhaps the moment AI became more evident to society was with the introduction of ChatGPT. While many people were already using AI in their phones, they weren’t fully aware of it. This progress has been possible thanks to the combination of data and computational power. To use AI effectively, it’s necessary to train a network with calculations and data. Essentially, the more parameters a model has, the more powerful the network becomes. Once the model is trained, it’s capable of performing various tasks.

Supercomputing supports AI, but AI also benefits supercomputing. A prime example is solving the protein folding problem—predicting a protein’s structure based on its amino acid sequence. This breakthrough, which was awarded the 2024 Nobel Prize in Chemistry, is a major scientific milestone that will help more than 50,000 researchers working on new drug discoveries. It demonstrates how AI enhances supercomputing’s capabilities.

At our center, we’ve been working with AI and supercomputing for the past 10 years. Traditional supercomputing centers no longer exist as standalone entities; now, the most important centers are those combining AI and supercomputing. There is a synergy between the two and we are moving further in that direction.

At BSC, we have large supercomputers and projects designed to develop massive models, like those equivalent to ChatGPT. These help us improve digital twins, whether for climate change or the human body and we actively work on such projects. There’s no scientific progress without collaboration. For instance, we are part of a global initiative called the Trillion Parameter Consortium, which includes more than 100 partners around the world. Together, we are working to build foundational Large Language Models that will accelerate the development and use of generative AI in order to improve science and engineering thanks to supercomputing.

In Europe, there is also an initiative to create AI factories. These won’t compete with the supercomputing work we already do. Instead, their goal is to help small and medium-sized enterprises and public administrations use their data with AI to improve productivity, workflows and organization. These machines won’t replicate what we’re already doing; they are designed to bring AI to SMEs. It’s a significant challenge, but also a great opportunity.

What specific areas is BSC currently focusing on in relation to digital twin technologies and what efforts are being made to implement these innovations in Catalonia and globally? 

Of the 950 researchers we have, we are organized into four departments. One of the largest is the computer science department, where we specialize in designing computers. The other three departments focus on applications, which we chose to center around earth sciences, life sciences and engineering. These application departments work on creating digital twins, which are essentially simulations of something we want to understand better, either for the first time or in increasingly precise ways.

In the area of cities, we are working with Bologna, Italy, and Kobe, Japan, to develop urban digital twins that support science-based public policy in the areas of mobility, energy, urban planning and greenhouse gas reduction. The more information and data you have, the better insights you can provide to help city leaders make decisions. We also created a spin-off company that has developed a digital twin of the heart, focused on improving heart simulations that can be used, for example, to test drug treatments or support surgical preparation. If you have an individualized heart model, you can achieve a lot. The biggest research challenge is to create a digital twin of the entire human body, as this would be incredibly helpful, but it’s a long-term project that needs to be tackled step by step. For example, we’re studying the influence of genes on cancer. Digital twins in medicine are a growing area of focus for us.

One major European project in this area is “Destination Earth.” Thanks to faster computers, we can create more accurate models. Currently, we are working on models with a resolution of three miles which is highly precise. This requires extremely fast computers, teams incorporating physical models and AI-guided programs that recognize patterns and help direct simulations. We are very excited about these developments.

The European Chips Act, introduced in 2022, aims to double the EU’s share of the global semiconductor market from 10% to 20% by 2030. How has the European Chips Act driven the growth of the semiconductor industry in Catalonia and what role is the BSC playing to make this vision a reality?

The topic of chips is geopolitically one of the most important in the world, second only to wars and climate change. Chips govern every aspect of life. AI data and computing are transforming nearly every facet of our existence. Chips are the most strategic element, but there are many types of chips, from simple to highly complex. The complexity is defined by the size of the transistor.

New chips are being designed and we are approaching the physical limit of how small silicon transistors can get, as quantum phenomena emerge when there are too few silicon atoms. Transistors, the basic building blocks of chips, are essential for computation. There are chips today with 200 billion transistors, operating at 2 billion cycles per second and millions of these processors work together in a computer.

Chip design and manufacturing are two distinct processes. The most advanced chips are manufactured in only two places: Taiwan and South Korea. The US has Intel and is trying to regain ground. Although 90% of the world’s most advanced chips are manufactured in Taiwan, they are often designed in the U.S. The design depends heavily on the materials available.

Europe is capable of designing and manufacturing chips with larger transistors, which are used in certain applications. However, Europe lacks the capability to produce the high-performance chips needed for AI, supercomputers, high-tech computing and future vehicles. While the goal to increase Europe’s global share from 10% to 20% sounds ambitious, it currently focuses on simpler chips. The focus should shift toward producing 10% of high-quality, cutting-edge chips. To boost this effort, countries are investing heavily. Spain, for example, has launched the PERTE Chip program, with €12.4 billion allocated to enhance chip design capabilities.

Currently, European supercomputers do not use European-made chips. For this reason, we persuaded the European Union to create a program dedicated to developing chips using RISC-V technology, an open-source architecture. This is a significant challenge, as while we design the chips, manufacturing still happens in Taiwan.

Our latest efforts include creating companies to drive chip production. The goal of the BSC is to activate this ecosystem, aiming to produce European chips within the next four years. These chips may not yet compete with NVIDIA’s, but it’s a starting point. We were the ones who initiated the push to convince Europe of the importance of this endeavor. We are optimistic about producing European chips.

What major breakthroughs has BSC achieved in earth sciences and climate change research? What ongoing efforts are shaping its contributions to this critical area?

There are three aspects in which we are trying to advance the impact of technology in climate modelling and the provision of climate information to society. The first is the development of a digital twin for climate adaptation by developing a whole chain of information systems based on the most advanced global climate models running at unprecedented resolutions on several supercomputing platforms in Europe. This chain is designed to provide solutions mainly for policy makers to decide what are the best policies for climate adaptation strategies for European citizens.

The second aspect has to do with emulating the past and future climate. So, given that the current climate simulators based on global climate models are terribly expensive in terms of computing time, we are working on developing an emulator to obtain machine-learning based climate models that we can use to provide climate data for the future much faster than what we can currently do with the physical climate models. We are also using these simulations to explore solutions that correspond to possible climate futures that we cannot explore with the physical climate models because of their cost.

And the third way in which we are working hard to increase the penetration of technology into climate-related problems has to do with the provision of climate information to society, based on the best sources of climate data available at BSC and elsewhere in the world. We can provide solutions for sectors that are vulnerable to climate change, such as agriculture, energy, hydrology and water resources, tourism, urban planning and many others. We do this in collaboration with organisations with whom we have developed recommendations for standardising the climate services that underpin the provision of the climate information needed to reduce our exposure to climate change and mitigate its impacts.

How vital is international cooperation for the development of technologies like those at BSC? What attractive opportunities exist for foreign involvement in BSC and Barcelona’s broader tech ecosystem in 2025?

Without financial support, nothing can be achieved. Our European programs are the main source of funding for many of our projects. It’s not easy; you have to collaborate with other countries and out of every seven proposals, only one is accepted. It’s an extremely competitive process. Without this support, there wouldn’t be supercomputers.

In Europe, half of the funding for computers comes from European initiatives, with co-financing arrangements. These projects allow us to hire people, carry out research and foster collaboration. Without collaboration, there’s no research.

One of our projects involved hospitals in Canada, the United States and Japan and focused on studying the influence of genes and their role in cancer susceptibility. We’re currently involved in the Trillion Parameter Consortium, as I mentioned earlier, without which it would not be possible to program foundational large language models, which require huge human and computational resources. From the beginning, we’ve collaborated with companies like Intel and Microsoft. Collaboration has always been at the core of our efforts.

How significant do you believe supercomputing will be for the development of the global industrial ecosystem? What major challenges could it help humanity overcome and what positive impacts might it have for Spanish and global citizens?

In everything we do, AI plays a role. Supercomputing is the most influential tool because it powers the execution of the most critical AI applications, such as protein folding simulations and advancements in green energy. This influence stems from science — without science, there is no progress. Every time a new tool emerges, it sparks a wave of scientific and engineering advancements around it. The computer has been the tool that has transformed the world and driven progress the most. It is fueling a movement where new developments emerge every day.

To conclude, do you have any final thoughts or messages for the readers of USA TODAY?

Collaboration, not only between countries but also in science, serves as a small yet significant antidote to prevent wars. When people collaborate, we understand each other better and gain a different perspective on the world compared to when we remain isolated. At the Barcelona Supercomputing Center, we aim to be a meeting point for Spanish and Catalan administrations, as this institution has been one of the few places where both governments have converged.

BSC is also a hub for the exchange of ideas and societal progress. While we strive for excellent research published in top-tier journals, our ultimate goal is not just to write papers but to solve real societal problems. Our mission is to create a happier world and a better society.