The European Ai challenge passes through start-ups: the cases of Italy, France and Lithuania

According to Harvard University's Critical and Emerging Technologies Index, explained by the authors in this article, our country 'is among the few in Europe with a first-class supercomputing infrastructure and actively participates in the main global fora on AI. The scientific quality is there, as is an advanced regulatory framework thanks to the European AI Act, of which we are among the most convinced promoters. Yet, all this is not enough without a real bridge between research and industry. While France and Germany attract billions in venture capital investments, Italian start-ups struggle to scale. Our ecosystem still lacks players capable of developing founding models, such as LLMs (Large Language Models), and there are no national public-private consortia for the training of competitive models on an international scale'.

According to the latest report produced by the Artificial Intelligence Observatory of the Politecnico di Milano, the number of AI start-ups has increased from 697 in 2023 to 796 in 2024, with 156 new companies founded in the last year alone. However, this quantitative growth is not accompanied by a structural or dimensional strengthening of the ecosystem.

Italian AI start-ups remain mostly small, with average revenues in 2024 of around EUR 140,000, a level in line with previous years (EUR 177,000 in 2023, EUR 134,000 in 2022). These values highlight the difficulty for many realities to get out of the early stage and reach a sustainable scale, thus limiting their competitiveness, especially at an international level.

Mixed signals are also observed on the funding front. The average funding per AI startup in Italy grew in 2024 compared to the previous year, from 606 thousand euros to 1.77 million euros, but still remains lower than observed in other European contexts. In 2023, for example, Italian AI startups had only raised a total of EUR 59 million, compared to Germany's EUR 1,910 million, France's EUR 1,690 million and Spain's EUR 363 million.

Despite the overall critical situation, positive signs are being seen thanks to structural initiatives promoted at the institutional level. Prominent among these is the commitment of CDP Venture Capital, which has launched a EUR 1 billion fund dedicated to artificial intelligence, with the aim of financing start-ups, research projects and internationalisation paths.

France is experiencing an explosion of artificial intelligence-related initiatives. According to data from the Governmental General Agency for Businesses, there will be over a thousand AI-related French start-ups in 2025, up sharply from 502 in 2021. The leading sectors are biotechnology and health (13%), followed by data and cloud (8%), as noted by France Digitale.

The ecosystem has also developed thanks to massive support from the public sector. In 2022, AI startups benefited from €1.5 billion in public funding, as part of the 'France 2030' strategy, launched back in 2018 to position the country at the head of the European race on AI.

In parallel, the 'French Tech 2030' programme selected 125 promising start-ups, including six specialised in AI. For these, support includes not only financial advice and international visibility, but also an exclusive network and a dedicated manager, as well as support in security and business development.

However, the picture is not without its criticalities. The main obstacle for AI start-ups in France, according to a recent survey conducted by France Digitale, is the difficulty in attracting investment and building sustainable business models. Historical problems such as the shortage of talent and computational power remain, but they are less pressing today than in other European countries.

The real challenge, however, seems to be access to data: one in four start-ups consider this to be the number one problem for the next 12 months, due to strict European privacy regulations, which make comparison with non-European counterparts particularly difficult. There is also growing concern over access to energy, often a priority for 'big players' such as Google or Microsoft.

"France is the European centre of AI start-ups. But this national leadership is no longer enough: Europe must build a unified ecosystem capable of offering turnkey solutions for businesses and public administrations,' said Maya Noël, Director General of France Digitale.

One of the symbolic stories of the French moment is that of Mistral AI, founded in April 2023 by Arthur Mensch, Guillaume Lample and Timothée Lacroix, former engineers at Meta and Google. It is the most important AI reality in Europe. Their open source model quickly gained international acclaim for its high level of performance, able to compete with those of OpenAI, Meta and Google. President Emmanuel Macron has publicly praised 'Le Chat', the conversational application developed by Mistral.

Bloomberg recently reported on Apple's possible interest in acquiring Mistral AI, but according to industry experts, these rumours are unfounded.

Lithuania is accelerating on artificial intelligence, determined to carve out a leading role for itself on the European technology scene. Thanks to a national strategy launched as early as 2019 - the second in Europe after France - and a strong synergy between the government, the public sector and start-ups, the Baltic country is set to become the future AI hub in Europe.

The 'Brand Lithuania Team', a governmental structure for the promotion of the country, emphasises the efforts underway: from the creation of a AI sandbox to test innovative solutions in a protected regulatory environment, to the implementation of the European AI Act with the direct involvement of the Innovation Agency Lithuania and the National Communications Authority.

The investment is also economic: EUR 15 million has already been earmarked for deep tech start-ups, while another EUR 110 million will be used to digitise public services, with AI at the core. Key sectors include life sciences, CleanTech and AgriTech, where AI solutions are already being applied to early diagnosis, renewable energy and smart agriculture.

There is no shortage of success stories: healthtech Spike raised $3.5 million to develop systems based on data from wearable devices; Unmanned Defense Systems received $3.2 million for autonomous military technologies; Pixevia is experimenting with automated shops. The US giant CAST AI has also chosen Vilnius as its base for development outside the US.

Lithuania now boasts over 800 startups, an ecosystem valued at EUR 16 billion (+39 times since 2014), with giants such as Vinted, Nord Security and BCG already in the Olympus of unicorns. According to Dealroom, it is the fastest growing country in Central and Eastern Europe.

Deep tech also finds its place in this ferment: medical robotics, artificial intelligence for cardiac diagnostics and even DNA data storage solutions. But the obstacle remains European fragmentation. According to the Baltic Deep Tech Report, only 20% of academic patents reach the market, a sign of a lack of integration between research and business.

To bridge the gap with the US and Asia, Brussels has created the EUR 10 billion European Innovation Council, of which EUR 391 million has already been allocated to deep tech start-ups. And from this fund come important resources for innovative Lithuanian companies such as Ligence, Genomika and Sentante.

*This article is part of the European collaborative journalism project "Pulse".