Europe has started taking steps to fix the immense problems it believes have hampered its competitiveness and left it with an under-developed artificial intelligence (AI) industry that cannot compete on the world stage.
Leaders of European Union (EU) nations, having called for urgent measures, set terms for political negotiations to simplify onerous rules governing financiers and firms angling for public funds to support investment deals in key sectors including AI.
A newly formed European administration, having made mending EU competitiveness the main task for its term, published its flagship plan to do that – the AI Continent Strategy, which promised a handful of the vast array of ambitious reforms that Mario Draghi, former EU Central Bank chief, set out in his report on European competitiveness for Commission president Ursula von der Leyen last year. She pledged to implement them upon starting her second term in January.
The AI plan promised to fulfil a long-held ambition to build a European cloud computing infrastructure, without which Europe would not have the capacity it would need to build an AI industry to be reckoned with. US dominance of the sector has been a rallying cry in Brussels.
It promised to get traditional EU industries to actually use AI to their advantage as well, and to build a workforce capable of producing not only the AI apps industry would need to do that, but to build European AI systems the apps would run on. Only 13.5% of EU firms were using AI, Eurostat said in January. Most of those were large tech firms in wealthier North and West European countries.
The Commission also pledged a single market for data – years in construction already – to corral data sources that AI developers needed to make credible systems.
It also pledged tentative steps towards simplifying the vast statute of AI and digital regulations that Von der Leyen, in her last term as Commission president, spent four years erecting, only for Draghi, who she commissioned to determine why European tech was ailing, to tell her the regulations were one of the problems.
Lack of choices
Revisiting the issue in a recent speech, Draghi pleaded urgency in recognition that the world’s top 10 large language models (LLMs), such as those of market-leading OpenAI, included none from Europe. The symptom was evident in the choice innovative EU AI startup firms faced when seeking growth, he said: “Stay small or move to the US.”
EU investors were being enticed to the US by light regulation, low taxes and a threefold saving on energy prices, said Draghi. EU startups were meanwhile further hindered by burdensome red tape, a fragmented capital market and barriers in a malformed single European market that hindered growth beyond their home nations.
As for Europe’s infamous digital regulations, they prioritised precaution over innovation, and added cost for EU firms – 20% from its General Data Protection Regulation alone. Persistent barriers in Europe’s single market meanwhile amounted to a 110% tariff on EU firms.
Europe wanted for a computing infrastructure to support an AI industry moreover, because it lacked investment to build one. Its fragmented capital market and bank-dominated financial system were causing €300bn a year of EU savings to go overseas. Energy prices were high and Europe was suffering a brain drain of scientists to the US as well.
The EU still had a chance of becoming a world leader in AI, Henna Virkkunen, the EU commissioner responsible for the AI strategy insisted, upon introducing it. “The global race for AI is far from over,” she said.
The big issues undermining EU competitiveness – its fragmented capital, energy and single markets – form the rest of an agenda the Commission declared for its term in February.
The AI plan meanwhile proposes raising money to build a European industry by offering €50bn of guarantees for private investors nervous about risking their money on EU firms. Leyen declared €150bn that private investors had pledged in return. But the association of AI firms and financiers that offered it did so on the condition that the EU drastically cut the statute of over 100 tech regulations that are deterring innovation, cut bureaucracy that hindered investment, boost skills, and harmonise rules on datacentres, networks and energy needed to build computing infrastructure. It amounted to the Commission’s four-year plan.
Europe lagging behind
Europe’s ailment is evident in official data that, by Computer Weekly’s analysis, shows 80% of private investment in AI in the past decade was made in the US. Europe managed barely 6% of that.
Lacking private investment, the Commission has found €30bn to fund 18 AI factories and gigafactories – public supercomputers it intends to nurture by providing computer infrastructure. A Cloud and AI Development Act would later this year prepare to treble datacentres housing EU computing capacity.
Europe has produced less AI not only because it had less investment, but because its firms were more inclined to buy US technology, said Lex Avstreikh, head of strategy for Swedish AI data storage startup Hopsworks.
“[Americans] tended to be more risk-takers, so they have more startups and more technology,” he said. “We are a bit more risk averse. We have a strong culture of software engineering [but] a less strong culture of monetising it. It’s just a matter of motivation.”
Ian Andrew, chief revenue officer of US AI chip pioneer Groq, said it planned to invest in Europe after raising $640m from investors last year, but was delayed by regulatory hurdles.
“The US has been the dominant region for deploying AI because of the cost of power, largely,” said Dan Scarbrough, chief of staff at UK AI infrastructure startup Stelia, which has been expanding in Europe. The US had more abundant energy as well.
The AI plan pledged an office dedicated to helping firms understand Europe’s AI regulations, but deferred the greater task of simplifying its tech statute until after a public consultation. Its AI Factories would meanwhile be overseen by another office, enforcing the AI Act.
At pains repeatedly to answer critics of the regulation, Virkkunen insisted recently that 85% of AI systems were not obligated to it. That AI innovators are nevertheless deterred by fear and incomprehension is a view frequently heard in the European Parliament.
