Is EU’s Regulation Holding It Back in AI?

The shortest answer is: no.

(Image by Thijs ter Haar)

I sometimes meet people and see posters online who confidently claim, in varying terms, that the EU’s regulation is holding back its potential in AI technology. In my own words, the typical argument goes something like this:

“The US is so significantly ahead in tech and innovation that the EU has resorted to becoming a regulator - simply because it cannot compete. The EU cannot develop a mature IT sector of its own and will remain dependent on exports from its friendly neighbor across the Atlantic Ocean as long as it maintains a regulatory environment that is hostile towards innovation.”

There may be some truth to this statement but if there is, it’s only one small piece of a much larger jigsaw puzzle. The causality between EU’s digital regulation and its lag behind the US and China when it comes to AI cannot survive closer scrutiny. 

Let’s just look at the most controversial bit of EU’s AI Act; its compliance requirements for providers of “general-purpose models”. As I wrote about in my post about the new AI laws in California,  California opted for only regulating certain potentially harmful applications of AI rather than placing any compliance burden on AI model developers. A widely touted view among tech entrepreneurs, especially in the US, is that regulating models before they are placed on the market is harmful to innovation. 

Nonetheless, the AI Act’s obligations to providers of general-purpose AI models listed in Article 53, do not seem unfairly burdensome but reflect what a responsible company should already do. That is my view. The same view is reflected in an opinion piece in Fortune by Bob Goodson, president and founder of the San Francisco-based AI company Quid Inc. We can also consider that the compliance costs incurred from living up to the AI Act’s requirements are measly compared to the costs of training and operating state-of-the-art general-purpose models.  

The EU is trailing the US in AI development, not because of its regulation but due to the fact that the EU does not have as strong a capital market or a comparable BigTech sector. The well-respected economist and former European Central Bank President, Mario Draghi, shared in his recent report, The Future of European Competitiveness a few key facts about EU’s global position in tech, computing, and AI:

• There is no EU company with a market cap over €100 billion that has been set up from scratch in the last fifty years. Six US companies with a valuation above €1 trillion (Alphabet, Apple, Amazon, Meta, Microsoft, and Nvidia) have been created during the same period.

• Among leading companies in software and internet, EU firms represent only 7% of R&D expenditure, compared with 71% for the US and 15% for China.

• The EU is home to only four of the fifty largest digital marketplaces worldwide, while the ten largest platforms serving EU citizens are owned by the US (Alphabet, Amazon, Meta, Apple, Microsoft, X) and China (Tencent, Alibaba, Byte Dance and Baidu).

• The three US-based cloud 'Hyperscalers' (Amazon Web Services, Microsoft Azure, and Google Cloud) account for 65% of the EU’s cloud market. EU’s competitive disadvantage will most likely only widen in the years to come as operating costs, including real estate and energy costs, are substantially higher in the EU than in the US and China.   

• 73% of foundational models (general-purpose modes) developed since 2017 are from the US and 15% from China.

• In 2023, an estimated $8 billion in venture capital investment was made in AI in the EU, compared to $68 billion in the US and $15 billion in China

• Of the top global AI startups worldwide, 61% of global funding goes to US companies, 17% to Chinese companies, and only 6% to those in the EU.

(On a more positive note, the EU has a strong international position in high-performance computing (HPC). Three of EU's supercomputers (Lumi in Finland, Leonardo in Italy, and Mare Nostrum 5 in Spain) are in the top ten worldwide.  Two exascale computers are planned to launch in the near future. The HPC centers are now opening up to AI startups, small – to medium enterprises (SMEs), and the broader AI community. )

How can we explain the EU’s lack of tech innovation and its conspicuous second runner-up position in the race with the US and China, if it’s not because of its laws?  

We will take a closer look at this question below and finally ask whether a global leadership position in AI is even advantageous without a strong legal framework to govern the technology. 

Anu Bradford, scholar at Columbia University Law School, identifies four reasons relating to features of the EU’s tech ecosystem in her paper, “The False Choice Between Digital Regulation and Innovation,” which helps to explain why the EU is at such a disadvantage in the global AI arms race and lacks a BigTech sector.