Hello. It looks like you�re using an ad blocker that may prevent our website from working properly. To receive the best Tortoise experience possible, please make sure any blockers are switched off and refresh the page.

If you have any questions or need help, let us know at memberhelp@tortoisemedia.com

Europe falls behind in the AI race

Europe falls behind in the AI race

A weak startup ecosystem is holding the EU back from unleashing its full potential in the AI space, while the US and China still reign supreme, according to the latest release of the Tortoise Global AI Index

EU nations are falling further behind the US and China in the race to develop artificial intelligence, according to the most comprehensive measure of global AI capabilities. 

Published today in its third iteration, the Tortoise Global AI Index measures AI performance across more than 62 countries using 150 indicators, including quality of research, activity on coding platforms and level of investment and government spending. 

In the last six months, three quarters of EU nations represented in the Index have seen their AI performance fall relative to rivals elsewhere. Despite leading on AI regulatory proposals, the bloc is struggling to compete with the intense speed of development in China and the US.

AI is making far-reaching changes to the way that we work and it is expected to unleash enormous efficiencies across all sectors, including everything from customer service to defence. Governments across the world have set out strategies on how best to maximise its potential rewards, and manage its risks. 

Tortoise Intelligence launched the Global AI Index in November 2019 to help policymakers, analysts and journalists track how nations’ AI ecosystems emerge over time, and to determine who is truly leading in this space. The Index’s rankings are refreshed every six months with a major update once a year in December. 

The rankings

Today’s update reveals that three quarters of the 22 EU member states featured in our Index dropped at least one position in our overall rankings. The European countries with the biggest falls were Luxembourg, which slipped 7 positions, Greece, which dropped by 8 places, and Slovenia, which dropped 3 places.

Despite this slippage, the EU is leading the charge to regulate the AI sector. In April, the European Commission proposed a regulatory structure that sets out to ban certain high-risk uses of AI. Countries such as Japan and Canada are already taking a close look at the proposals, which will not become law for around two years. 

The move to regulate the space comes against a backdrop of fears about potential misuse of AI, particularly in areas such as live biometric data collection and social scoring. 

Why is the EU slipping back?

The EU is struggling to maintain the same level of financial momentum for its AI startup ecosystem as China and the US. The share of venture capital funding going to AI companies in EU countries is declining, according to a Tortoise analysis of Crunchbase, an extensive database of funding rounds.

The analysis, which is incorporated into our Index, covers funding in 22 of the EU’s 27 states. In 2020, these countries cumulatively pulled in $1.7 billion in venture capital for AI companies, down slightly from $2 billion in 2019. In 2021, venture funding flowing into these EU states amounted to just 2.9 per cent of the total across all 62 index countries, a figure that has declined from 4.4 per cent in 2020 and 5 per cent in 2019.

The EU states analysed do not include the UK. From 2019-2020, the UK alone raised almost as much as the 22 EU states combined. So far this year, UK AI companies have already raised $1.6 billion – 70 per cent more than the $950 million raised by the EU states in the same period. 

Indisputably, it’s the US that attracts the lion’s share of AI-focussed venture capital funding. Since 2019, its AI startups have pulled in $65 billion, 14 times the amount raised by the 22 EU nations. Chinese AI companies are also receiving far more venture capital than those in the EU, pulling in $24.1 billion since 2019 – five times more than the amount invested into EU companies. 

This isn’t to say that EU startups aren’t making their mark on the industry. Konux, a German-based company making railway sensors, and Paris-based Prophesee, which develops neuromorphic vision technology, are among the European startups to receive substantive funding rounds.

The US leads the way on coding

As well as following funding flows, our Index tracks the lifeblood of AI activity: code. On these metrics, too, we find that when EU countries are combined, they tend to sit substantially below the US. 

For instance, the US leads on contributions to the most widely used open-source code libraries, such as Pytorch and Tensorflow, according to our analysis of Github, a platform for hosting and sharing code. We found that, over the past year, US-based coders have made over five times more code contributions than the 22 EU countries.

We also monitor Kaggle, a major online platform that organises challenges relating to computer science, where the highest achievers in competitions are awarded with the title of “grandmaster”. On this measure, the US is outpacing both China and EU nations, having doubled its count of grandmasters from 52 to 103 since July last year.

When it comes to Python package downloads, another measure of coding activity, EU countries have marginally increased their share of the total. 

China often comes below the EU on coding metrics, though we’re aware that some of our indicators tend to favour Western countries. For instance, while Python is a commonly used programming language in China, the site that hosts it has been targeted by Chinese state censorship, and Kaggle has also received the same treatment. This said, Github has remained accessible in China since an attempt by the government to block it in 2013 resulted in protests.

The EU’s potential: research

The EU’s real strength lies in its research. Academics in the region are known for their scientific standing, and the EU is home to a quarter of the top 100 Times Higher Education Universities for Physical Sciences (this does not include the UK, which is home to a further 11). 

Our data also shows that the bloc is home to nearly as many AI researchers per capita as the US, a measure where it also far outstrips China.

EU countries perform well when it comes to publishing, too. Between 2017 and 2021, 605 European academic AI papers were accepted by top AI journals managed by the IEEE – a body which sets AI standards and also publishes research. The US still outperforms the EU in this area, publishing 638 papers over the same time period.

On this measure, both the EU and the US lag far behind China. Last year alone, Chinese academics published 1,008 research papers to the IEEE – that’s three times as many as EU and US academics combined. The trend remains clear when we adjust the figures for population sizes – China is far ahead, with the EU nations and US fighting for second place.

Overall, however, China ranks second in our Index for research; it is outstripped by the US, which has a stronger overall academic ecosystem. The Index leader is home to more researchers per population and top 100 science universities than any other nation, as well as numerous AI labs and societies. On these measures, both China and the US have plenty of catching up to do.

Our data suggests that the real strengths – and potential – of the EU lie in developing its research base. Cooperative clusters – such as the Nordic Artificial Intelligence Institute, the Benelux Association for Artificial Intelligence and Robott-Net – allow academics to collaborate on projects and accelerate the development of AI.

The future of European AI

Individual EU countries come high in our rankings: Germany, The Netherlands and France all place in the top 10 spots of the Index thanks to their strong talent, research and government strategy pillars. As a bloc, however, the EU has struggled to unite these strengths. When it comes to investment into AI and its implementation, European talent and funding is scattered across different nations. 

European venture capital markets are also less mature, with far less investment flowing into the region than the US. Challenges around funding have been accelerated by Brexit, particularly as the EU is also lacking centralised investment funding on the scale of that of the Chinese government.

The EU has gaps to fill when it comes to AI capacity and implementation. But the bloc has taken the lead in another important area: regulation. 

The publication of a first-of-its-kind framework for AI regulation is significant. Though the draft regulation strategy must now go through a lengthy legislative process, the final version may soon become a global standard for the technology. Europe may have work to do to catch up to its competitors, but the evidence suggests that its influence will be felt across the sector.

Elsewhere in the Index: Who’s up, who’s down?

The US remains firmly ahead 

China has made unmistakable progress in specific areas of AI. The recent announcement of the “Wu Dao” AI system, built by the Beijing Academy of Artificial Intelligence, dramatically outstrips the US-built GPT-3 model in terms of the size of its training data. Other Chinese strengths include AI chip manufacturing and quantum computing, where Chinese tech giant Baidu has emerged as a driving force

However, these huge leaps have not fed through to our Index. Since November 2019, China’s score out of 100 has remained the same. This is for a number of reasons. First, our index captures the overall strength of a country’s ecosystem, across seven pillars. It does not capture distinct or individual breakthroughs (e.g. the GPT3 rival). In this sense, China’s position may be undervalued. However, in terms of an analysis of broad national AI ecosystems, the gap between China and the US does not seem to be narrowing (yet). 

The UK hangs on to third place – just

From its position in fourth place, Canada is nipping at the heels of the UK. In the latest iteration of the Index, the gap between the two has narrowed for the third time, and by 35 per cent between December 2020 and today. Brexit may pose significant challenges to the UK’s AI sector. This said, the funding landscape in the UK appears to be picking up pace, and the UK government is ramping up its focus on AI. It has already earmarked a budget of £950 million via the AI Sector deal, announced in 2018, but is now planning a spending review, which may unleash new funds for the field. It is also due to update its national AI strategy in the fall. In a separate development, last week the UK announced a five-year £210 million research initiative in partnership with IBM, designed to encourage new discoveries in artificial intelligence and related fields like quantum computing.

South Korea, Taiwan and Hong Kong up

In this round of the Index, strong performance was observed across many Asian countries, in particular Taiwan, Hong Kong and South Korea. Traditionally these nations have important export strengths, with each being major players in the computer chip industry. Hong Kong leads the way for its number of integrated circuit exports, while Taiwan ranks 4th and South Korea 5th. Research is another growing strength for these nations: they all saw large increases in the number of published papers at the IEEE, the global standards-setting and research organisation. 


A fuller description of our methodology is available on our Intelligence website.

What is an index?

An index is a ranking built from a careful selection of different measurements around a central topic or theme. Here, the index ranks countries on the basis of their capacity for artificial intelligence.

Which countries are we covering?

We limited our analysis to 62 countries, most of which had published some sort of national strategy on AI setting out future plans.

What’s included?

We grouped these data points into three clusters: innovation, implementation and investment. Within those are seven key categories: research, development, talent, operating environment, infrastructure, commercial ventures and government strategy – all of which contribute to overall AI capacity. We weighted the factors by their significance, and gave a preference to human and intellectual capital, as well as investment, as these as critical drivers to building capacity.