On Attaining European Technological Sovereignty

What first comes to mind when talking about Europe and tech is likely the strong forms of regulation that the European Union (EU) has legislated in recent times. Most prominent of this is of course the General Data Protection Regulation (GDPR) that governs the handling of data on European users and has had global impacts through the Brussels effect. Thinking of American or Chinese tech on the other hand likely awakes images of more empirical accomplishments — of companies, products and services that actually make up an industry.

This invites the conclusion that because Europe has strong regulation it has killed off its nascent tech industry. I do not hold the opinion that this is the case. GDPR in particular mostly harms larger firms that collect large amounts of data on individuals and therefore have to build large elaborate systems to try and manage this data collection. Small startups are instead incentivized to use data collection systems that anonymize data to begin with.

“But if European startups can’t collect as much data they won’t make us much money right?” There is an element of truth to this, but the large American tech companies have shown that it is still profitable to run social media platforms in a post-GDPR Europe — otherwise they would have left the European market.

There are also those who say that the reason why Europe does not have much AI, or a strong tech industry more broadly, is because there are those who percieve such sectors as “authoritatian”, “promethean madness” or “techno-fascist”. To me this seems absurd. Even if these people have any actual influence on policy they are still not the people who would have otherwise built a tech industry. The question should be focused on the people on the margins — why do people who sit in their bedrooms and code all day not found successful software companies?

Building a social media platform is comparatively easy². By far the biggest issues for any new player entering the field is network effects, that everyone you know is already on the preëxisting platform. The second hardest thing is keeping people continuously hooked on your platform (to use a colloquial term). This is important since the more users have their attention directed at your service, the more advertisements you can show them — the lifeblood of most social media platforms.

The provider of the digital platform does not do most of the work, that is done by content creators³. Hosts are incentivized to make it hard to move your creations to other platforms to keep the aforementioned network effects in place. To successfully break out into this space they need to provide a sufficiently different experience to create their own network effects before their inventions are copied by the dominant players.

But why do we need to have this focus on social media to begin with? Would it not be a better idea for Europe to “skip the line” and start building European Artificial Intelligence (AI) straight away? This sounds reasonable, but it still remains to be seen what the business model for AI will be. So far it has been bankrolled almost entirely from the spending on more traditional “Web 2.0” social media and advertisement companies. Meta and Google combined have an almost complete monopoly on digital advertisement, and are using this income from the “real economy” to finance these large bets on AI. Social media is the tech industry at this point.

China has managed to build its own tech industry through the application of this logic. The famous Great Firewall of China does not just keep information out of the country (like about the Tiananmen Square massacre) but just as importantly it has kept foreign tech companies out. Chinese tech companies have then filled this void and built competitive platforms. Mainland émigrés do not delete their Wechat accounts when moving abroad, abandoning them for superior American alternatives, they keep them due to their network effects.

It is possible to replicate this approach in Europe. A total block would likely be infeasible, owing to fact that European nations generally want friendly relations with the USA, but some sort of system that gives American companies a handicap for their monopolistic nature would not be unwarranted. This would likely be some sort of digital services tax (DST) that taxes larger platforms more. This would allow European companies to compete with entrenched rivals from the other side of the Atlantic.

The downside of such an approach is obvious, especially with the recent American administration. The tech industry has considerable influence in American decision-making, and so any move to supplant them will likely receive backlash. The Americans themselves consider Google to be a monopoly, so reducing their influence here would likely improve the state of the digital advertisement market.

I do not know if this is the path forward. To me it is easy to see the status quo as a wealth transfer from Europe (and the rest of the world) to Silicon Valley, data as a form of digital oil gathered cheaply in the orient before being sold to the domestic market. The problem may genuinely just be one of competition — of a few number of companies (that just happen to be American) gaining a disproportionate share of the economy due to their monopolistic practices.

What is the benefit genuinely gained from Europe not possesing their own tech companies? Cheaper electricity prices (than would otherswise be the case) may be one, but data centres still need to be present in Europe to efficiently provide Europeans with services (and content). To me it seems that smaller European companies are more than capable to providing the same services as the American tech giants, the only exception being the sheer scale (and that is itself perhaps problematic).

This has been discussed ad nauseam, but it bears repeating. The European Union needs a systematic banking and capital markets reform. This would allow startups to raise capital in Europe without moving to Silicon Valley.

Here in Sweden we are quite proud of what we see as an innovative and digital culture and economy, spawning things like Spotify, Minecraft, and Skype. But all of these have moved to the USA in some capacity⁴. That the American financial system is so dominant is of course a given, and we should therefore expect European companies to make use of this to raise additional capital — but Europe also needs to more effectively make use of the resources in its possession.

A deeper union for banking (perhaps paired with some form of combined European debt) would allow for better coördinated form of spending. This appears to be to be the biggest initial problem for European startups, a problem that right now is only solved by moving to the United States of America.

Data centres do not matter. That is an exaggeration of course, but investing more in data centres will not create a tech industry. They employ very few people following their construction, and can be used remotely from anywhere on the planet. A tech startup will not decide where to relocate to depending on whether or not they get slightly lower ping⁵. They will be located where they can get skilled staff, attract investment, and carry out their work effectively. There is no reason that this can not be in the European Union.

Data centres are still necessary, but constructing them does not drive innovation (except seemingly in AI). The state should be more focused on funding research and enabling innovative companies to actually enter the market rather than building infrastructure that will be used by American tech companies instead.

The entire point of the internet is to be able to separate computing from where it is needed and used. Data centres allow devices at the edge (laptops, smartphones, et cetera) to perform feats of computation that would otherwise be too slow — like running LLMs⁶. Having domestic data centre capacity is not critical for building a tech industry, you can just rent capacity in northern Virginia if you really need it. Building local data centres is then only a good idea if demand holds steady, and even if that is the case they still need to be refitted with newer, faster computers every so often.

This calculation looks different for China however. They are actively hostile to the US and therefore need to attain their own data centre capacity in case they come into armed conflict over Formosa. While America has shown its inability to be trusted in recent times, I still believe that the American tech lobby is strong enough to maintain European market access for the foreseeable future. The European market is simply too alluring for those companies to avoid.

Recently there has been increasing talks of tech stacks and the of the costs and advantages of building our own compared to using a preëxisting one. Tech stacks are sets of software (and/or hardware) that can be deployed in combination with each other. Most famous of these is perhaps the LAMP stack (Linux, Apache, MySQL/MongoDB, Perl/PHP/Python) using free (as in freedom) software to serve websites or applications. A newer example might be the use of Nvidia AI chips running pytorch written in Cuda. We all stands on the shoulders of giants, and it is needless to reinvent the wheel for all of your infrastructure.

A recent report from the Centre for European Policy Studies has broadened this perspective to the entire supply chain for digital services, all the way from the end-user application to the physical materials used in the manufacturing of the integrated circuits themselves. The proposal is that all of this (or at least a significant portion) should be supplied by EU institutions is not only protectionist to an absurd degree, it is unrealistic. Attaining digital sovereignty should not mean abandoning non-EU digital solutions entirely, and any attempt at doing so will likely fail.

Instead Europe should do what has been proven to work, to rid itself of proprietary solutions that exert control over their users (regardless of if they are the American government or a Lithuanian small business). Europe is already strong in this regard; The Linux kernel was first developed in Helsinki and is today developed globally. The World Wide Web (WWW) was developed at the pan-European CERN.

Free software is not just a part of building European technology, it is the only path forward. Supporting initiatives such as RISC-V is a good idea, but it is still problematic if European companies build proprietary systems on top of those free solutions, like Nvidia has done with its Cuda.

Building AI has not made much money so far, but I do not believe it will in the future either — just as being an internet-service provider is not hugely profitable (but not loss-making either). Open models have shown how LLMs are almost fungible in practice; Switching models can be done quickly and quite easily. AI companies have no “moat”. Instead the money is likely to be made by those who can effectively apply LLMs to new industries and applications. While OpenAI has generally performed best out of all the major model providers competitors are usually not far behind. Why use one company’s model when you can get 90% of the performance for a far lower price?⁷

At the Swedish Defence University — perhaps one of the places you would most likely expect technological independence to play a major role — we are still (and increasingly) reliant on Microsoft software. Switching to free and open alternatives (that are likely significantly more secure) should be easy, there is just a lack of urgency surrounding switching to alternatives that would improve European sovereignty.

Benjamin Bratton says that while Europe needs to build its own stack to achieve sovereignty, it can not rely on free software because it is “anti-sovereign”. But this is exactly what Europe needs. To crush the hearts of European federalists everywhere: Europe is not a state. It can not rely on solutions that allow other states to pressure each other. This is the reason that EU countries are limited in their ability to orchestrate industrial policy. The solution is not merely to move this decision-making to the EU-level. Instead Europe needs anti-sovereign solutions, because they work in the same spirit of European coöperation and integration as the EU itself.

Much like Europe itself free software is subject to intense debate on what it actually is, comes from a wide breadth of different places and people, sometimes has issues with coming up with unified plan, and yet is responsible for some of the most important improvements in recent history. What could be more fitting than for a European tech stack to reject a centralized form of industrial policy and instead adopt a form of technology that prioritizes freedom, efficiency, and diversity? In varietate concordia. ❦