The impact of artificial intelligence (AI) is one of the main topics of debate in Europe. And, in relation to AI, Europe is in a very similar situation, perhaps worse, than it was with respect to aeronautics in 1970.

In 1970, two commercial aircraft manufacturers dominated the market. Boeing and McDonell Douglas had 50% and 35% market share respectively, while other aircraft manufacturers shared the remaining 15%. At that time, some European governments made a momentous decision: to create a European commercial aircraft manufacturing company capable of competing with the big two. Today, 50 years later, two commercial aircraft manufacturers also dominate the market. Boeing has about 40% of the market share, and Airbus has about 60%. With permission from China, which in the near future will have its own commercial aircraft manufacturer.

A similar situation occurred in the late 1990s. The United States had developed a global positioning system for military purposes that only offered limited precision for civilian uses. During the 2000s, Europe decided to launch its own global positioning system, Galileo, and did so in an open way and compatible with the American GPS, but offering maximum precision. This project led to the US’s subsequent decision to eliminate the limitations of its GPS system and offer the same precision. Today, GPS and Galileo operate simultaneously, and from the European decision to design an open system, the countless geoinformation applications currently available have been developed.

Probably in 1970 most economic analysts would have predicted that the two large manufacturers, Boeing and McDonell Douglas, would share the commercial aviation market for many years. However, Boeing absorbed McDonell Douglas in 1996, precisely because of the rise of Airbus. If European governments had not taken the decisive action of creating an airplane manufacturer in 1970, your next trip would probably be in a Boeing or a McDonell Douglas. And if they had not decided to design the Galileo system as an open system in the 2000s, your car would not have maps to guide you through all the cities in the world.

Let’s go back to artificial intelligence. In hardware and artificial intelligence, Europe is in a very similar situation to what it found in aeronautics in 1970 or in GPS in 2000. We Europeans have the same feeling as then: there will never be a European company capable of competing with the market leaders. . Or if?

The problem of artificial intelligence is somewhat more complex, because it is based on the combination of hardware and software. And in both fields, Europe is lagging behind significantly. The development of the most advanced models of artificial intelligence requires very advanced processors such as those developed by Nvidia, AMD and Intel. In Europe, we neither design nor manufacture this type of processors; We depend entirely on the acquisition of these chips at a very high cost over which we have no control.

On the other hand, the design of the AI ​​software required by current and future systems with more than a few billion input parameters is very difficult to replicate in Europe, even though we have the most advanced chips of the moment. And to train those systems we need data. In Europe we have data, although many of them, personal data, are kidnapped by the large American technology companies, the GAFAM (Google, Amazon, Facebook, Apple and Microsoft), to develop their business model with our unconscious and essential contribution.

Given this situation, Europe takes the initiative in regulating Artificial Intelligence, but, in view of our technological delay, we could become the referees of a match in which we do not have players to compete. However, Europe can catch up on the technological gap in Artificial Intelligence if it is able to follow the Airbus and Galileo models.

Europe is beginning to be aware, with its Chip.act initiative, of the challenge that the development of new chips represents, and, on the other hand, we must recognize the success of the vision of general director Roberto Viola of establishing in Europe the AI factories in strategic locations such as large supercomputing centers and others. They will try to make the most of the use of European data to increase the competitiveness of our companies and researchers. But this is not enough. As in the case of Galileo, Europe must commit to the development of free technology and international collaboration. From Barcelona, ​​the BSC defends that Europe develops all hardware, from IoT chips to those necessary in large supercomputers and artificial intelligence machines, using the idea of ​​Open hardware enabled by RISC V technology.

On the other hand, we consider that it will be very difficult to develop the following generative AI models by researchers from public institutions; We need global collaboration. And we need the competitiveness that the private sector offers. This is where we need the Airbus model. Europe must create a large European company to offer artificial intelligence software and services following the Airbus model. A company capable of offering all the services offered by GAFAM: internet search engine, office suite, cloud storage services and artificial intelligence services. It will take time, it is clear. Resources will be needed, of course. It won’t be profitable for a few years, no doubt. But we should not think about 2030. We should think about 2070. What will be the main technology companies in 2070? Will any of them be European?

Creating this company may seem risky, but what is the cost of not creating it? Absolute and permanent technological dependence. Here the fact that Europe leads the regulation of the ethical aspects of AI is an advantage. A European AI company can develop its products under these regulations, consequently becoming the first option for European companies and governments to comply with European legislation. The European Commission and European governments have the resources. We have a European High Performance Computing Network (The European High Performance Computing Joint Undertaking) that leads research in supercomputing and can be an excellent foundation on which to develop this company. Competition regulation may have to be adapted, but it is time to recognize that, as in the case of aeronautics, competition regulation must be applied to the global market, not just the European one if it prevents a European company from competing. with the companies that dominate the market.

Finally, a fact. Baidu was created in 1999, and in 2024 it is the dominant search engine in China. Without denying that the Chinese market is much more closed than the European one, the truth is that Baidu offers the same features as GAFAM. China has done it. Why not Europe?

We need European chips, and we need to create AIbus.

Daniel Crespo He is rector of the Polytechnic University of Catalonia.

Mateo Valero He is director of the Barcelona Supercomputing Center.

You can follow The USA Print in Facebook and x or sign up here to receive our weekly newsletter.