Ten years ago, few entrepreneurs would have sat down to listen to what a supercomputer does. It was seen as something too distant for the day-to-day of their companies, perhaps too complicated. Things have changed, and this is reflected in the increasing presence of the Barcelona Supercomputing Center - Centro Nacional de Supercomputación (BSC-CNS) in public debate, no longer just to talk about the many complex calculations it is capable of performing, but also about its applications in issues that concern everyone, such as health or climate change. It has even become a source of pride for the city as a scientific infrastructure that positions it on the world stage.
It is not surprising, then, that the associate director of the BSC, Josep Maria Martorell, was the first guest of a new series of meetings organized by the Barcelona Chamber of Commerce, sponsored by Indra and with The New Barcelona Post as a collaborating media outlet. Nothing like the typical tie-and-suit lunch. It's early morning and it's time for a fork breakfast, starting with cold cuts and plenty of bread, fesols omelet and botifarra in between, and carquinyolis to finish. Half past eight in the morning is not a time when anyone would imagine thinking about petaflops, but Martorell, with a degree in Physical Sciences from the UB and a doctorate in Computer Science from Ramon Llull, does not miss any opportunity to try to make a theory that often sounds like science fiction more accessible.
To understand why things are now about supercomputing, but also about the famous artificial intelligence (AI), one must understand a first idea: "They are two sides of the same coin." And go back. In the beginning, science was essentially experimental. "The world was a laboratory, things fall, planets revolve, stars give light and fire burns. Based on this experimentation, the human species creates theories and ideas, even religions," says Martorell. From the 17th century, theoretical science appeared thanks to names like Newton or Kepler, who began to look for models to explain why things happen in nature. That is, theories that explain the world and laboratories to test them began to emerge. "This dialogue between theory and experimentation has led the human species to where we are. Everything is explained with this," he emphasizes.
Between the mid and late 20th century, this model also became obsolete because mathematics and sciences evolved so much that methods for finding solutions to the equations they posed were exhausted. So, they began to be solved numerically with computers. This is the era of computational science. Science takes another leap, but it's not the last. About ten years ago, it encountered disciplines like medicine that are not based on formulas but on data, a lot of data, to which AI knocks on the door to process it all, find behavioral patterns, and make new discoveries. Another important point: AI is not new; it was invented in the late 50s, it has been training for decades, but now the three elements it needed to explode have coincided in time (abundant data, thanks to mobile phones and communication networks, better computing, and more efficient algorithms).
"Nowadays, no one in the world does frontier science without the use of large computers and artificial intelligence. If you don't have them, you fall behind," emphasizes Martorell, which in English is who doesn't compute, doesn't compete. The litmus test is to look for the academic article of any recent discovery. Words like numerical simulation, AI, high-performance computing, or data analysis will always appear. This is where equipment like the BSC comes into play, but it doesn't stop there. Because industry is already testing these new tools to rethink its business and open new markets, which leads to being more competitive. In this context, initiatives such as AI factories that the Barcelona supercomputer will host are promoted to extend their use among companies, where it will work with the Barcelona Chamber of Commerce: "Within a year, any European company, but particularly those close to us, that wants to do R&D will have access to a spectacular machine, at almost zero cost, accompanied by training and support services. This will allow them to reach a very important competitive stage compared to their colleagues."
All of this makes it inevitable that governments start to look at what is happening here and geopolitics comes into play, as has become so clear this week with the emergence of DeepSeek's Chinese AI and the stir it has caused in the US economy. And Europe in the middle. It may have missed the train in many battles, but in something it has managed to agree with the EuroHPC initiative, sharing the budget of the 27 to acquire supercomputers that could look eye to eye with their counterparts from other more advanced countries. One of them, in Barcelona, with its latest version launched at the end of 2023, MareNostrum 5, and already preparing the next one, MareNostrum 6, for 2029. "It is a very clear European success," defends its associate director, in office since 2016; previously, he was general director of Research in the Generalitat.
There are more battles to fight, such as quantum computing, where it is not yet too late and there is a game to play, especially with the announcement a few days ago that the BSC will have a new quantum supercomputer, built by two Catalan startups, Qilimanjaro Quantum Tech and Do it Now. Without forgetting the role played in this field by the Institute of Photonic Sciences (ICFO) in Castelldefels.
"We are not children of chance, we are children of a man named Mateo Valero, who since the 80s was at the Polytechnic building computers, but also of Andreu Mas-Colell, who in the 2000s managed to convince President Pujol to implement a specific policy so that this country would have world-class research. And behind that came 20 years of governments of different colors that maintained it. Thanks to this, the city can play a fundamental role in this field," Martorell points out. The next game: having local technology within these devices, where, now, at most Europeans almost only manufacture the screws. For this to be possible, the budgets will be very substantial. "We are not bad in Europe, but there is a way to go," he emphasizes, without underestimating a more polarized and armed world, with data as a new tool for wealth and defense.