Last week the European Union unveiled an the “AI Continent”action plan, proclaiming that there is “a unique opportunity for Europe to act swiftly to shape the future of AI.”
It could be Europe’s moment, say tech industry veterans, if it takes the right approach. But some believe it is heading in the wrong direction.
Among other things the action plan aims to:
- Strengthen Europe’s AI and supercomputing infrastructure with a network of AI factories. Thirteen of these factories are already being deployed around Europe’s supercomputers. They aim to support EU AI startups, industry and researchers in developing AI models and applications.
- Set up AI Gigafactories. These will be large-scale facilities equipped with approximately 100,000 state-of-the-art AI chips, four times more than current AI factories. They will integrate massive computing power and data centers to train and develop complex AI models. The aim is to stimulate private investment in Gigafactories through InvestAI, which will mobilize €20 billion investment for up to five AI Gigafactories across the Union.
- Create data labs to facilitate access to high-quality data
- Cultivate AI skills and talents and increase means to attract foreign talent
“This screams of another half-hearted attempt at Europe trying to play catchup via its lethargic and bloated public sector,” veteran European venture capitalist Michael Jackson said in a LinkedIn post. “Even the terminology used – AI factories – shows they’re still stuck in industrial era thinking. Europe should be obsessed right now with making itself the best place in the world to start, build, and scale private businesses. This ain’t that.”
The plan does not effectively address Europe’s bottlenecks in AI resources, compute and energy, notes Giorgos Verdi, a tech policy fellow at the European Council on Foreign Relations in a policy alert. “To match its aspirations for AI leadership, the EU will need to develop an approach that feeds into its existing strengths—and breaks down artificial silos.”
Europe is starting from way behind:
- According to the 2025 AI Index report by Stanford Institute for Human-Centered Artificial Intelligence (HAI) only three notable AI models were developed in Europe last year – all of them in France. This represents a downward trend from 2023, as training runs are getting more expensive and AI development is getting more complex. In 2024, U.S.-based institutions produced 40 notable AI models, significantly surpassing Europe but also China’s 15. In the past decade more notable machine learning models have originated from the United States than any other country.
- $19.42 billion worth of AI private investments were made in the EU last year. While this represents a significant increase from 2023, the gap with the United States is huge: $109 billion was invested in AI on the other side of the Atlantic.
- 306 AI companies were newly funded in Europe last year, including startups in the UK and Switzerland. More than triple that number, 1073 AI companies, were newly funded in 2024 in the U.S.
Read on to find out what Verdi and tech industry veterans think Europe needs to do to turn things around and become a global leader in AI and other areas of deep tech.
Weaponize Open Source
While the AI Continent plan mentions open source is does not make it a key pillar of Europe’s strategy. That’s a mistake, says French serial entrepreneur Yann Lechelle, CEO of Probabl, a spin-off of French research center Inria that has been financing a global open source data science library called scikit-learn, a tool widely used for performing complex AI and machine learning tasks. (See The Innovator’s story about Probabl). “In a world where exponential technologies have shifted the concentration of power and value capture to a handful of non-European companies, mostly through proprietary and lock-in techniques, openness is the only radical and non-conflicting public policy that can reverse the trend immediately,” he says. “For the European Commission to stimulate, support and adopt more open science, open data, open source and open weights, open standards and open hardware , may be the strongest path to transforming the economic landscape. It’s a weapon that can be used for a massive leveling of the playing field.”
The recent AI Action Summit saw the emergence of a new AI bloc comprised of Europe and much of the rest of the world, built around open source. Some 58 of the countries attending the summit – which together represent one half of the global population – signed a statement committing to promoting AI accessibility to reduce digital divides; ensuring AI is open, inclusive, transparent, ethical, safe, security and trustworthy; avoiding market concentration; and making AI sustainable for people and planet. Countries embracing open source are rejecting the dominance of U.S. and Chinese players. Europe needs to capitalize on that, he says.
“Africa was late to digital because it did not invest in infrastructure so to catch up it adopted mobile tech to leapfrog ahead,” Lechelle said in an interview with The Innovator. “I would say Europe is now in the position that Africa was in pre-adoption of mobile. We have not invested in the Cloud so how do we catch up and leapfrog ahead – the way forward is to massively leverage open source and to weaponize it the way China did.”
By building on open source China’s DeekSeek developed AI reasoning models that appear to be on par with U.S. companies OpenAI and Anthropic, at significantly lower cost. This is fueling optimism that the closed large language models that U.S. AI players are spending billions on are not necessarily the future and U.S. AI domination in not inevitable.
Big Tech’s own growth stories illustrate the power of openness, says Lechelle. He notes that Microsoft once called Linux “cancer,” but now Azure thrives on open source. Google’s global reach is built on open web standards. Although Meta’s Llama 2+ isn’t entirely “open,” it still signals a recognition of the strategy’s importance. Amazon Web Services runs largely on OSS foundations, benefiting from open innovation and standards. NVIDIA success is due in part to blockchain and open science in deep learning, while Apple’s fortress, too, is built on OSS tools and code. “Simply put, each tech giant stands on the shoulders of open-source contributors—often without full acknowledgment,” says Lechelle. “Openness isn’t just a footnote; it’s the engine behind success.”
Better accessibility of open-source AI would lower the barrier to market entry for Europe’s low-resource actors and enable greater AI diffusion through customization, says Verdi. Without open-source software, businesses would need to spend 3.5 times more on software, he says. Simultaneously, open-source AI could enable independence for companies and mitigate the risk that the market concentrates with a few corporate gatekeepers. And, instead of focusing on building specialized models, developers can dedicate resources to creating specialized applications, unlocking the power of AI to solve real-world problems.
Create European AI Chips
“Europe’s dependency on the U.S. for the supply of AI chips is worrisome for its AI infrastructure plans,” says Verdi. Most EU member states already face restrictions on the amount of AI chips they can import from US-based providers. This gives U.S. President Donald Trump another bargaining tool—he could further lower these thresholds, he argues.
To ensure certainty for investors and avoid coercion, the EU will need to secure unhindered supplies of cutting-edge AI chips by investing in fabrication facilities and launching a Chips Act 2.0 to commit subsidies to AI chips manufacturing, Verdi says.
Collaboration with Asian manufacturers is essential, particularly by incentivizing them to establish fabrication plants (fabs) in Europe, Peter Wennink, a former CEO of European chip company ASML said in an interview published in the March 2025 European Deep Tech Report produced by dealroom. LakeStar, Hello Tomorrow and Walden Catalyst. “This requires a unified approach, with major corporates such as Siemens, Bosch, and the automotive sector acting as collective customers and strong advocates for local manufacturing,” he said. “The 2023 European Chips Act falls short in facilitating a focused access to capital and fostering targeted innovation,” said Wennink. What is needed, he said, to drive meaningful progress is to establish an innovation fund specifically for the AI stack and create dedicated innovation centers across Europe.
It took China just three years to develop its own AI chips, says Andre Loesekrug-Pietri, the chairman and scientific director of the Joint European Disruptive Initiative (JEDI), a precursor to a European Advanced Research Projects Agency (European ARPA), which brings together more than 6000 leading scientists, startup founders and industrialists from 29 European countries -to accelerate the continent’s leadership in disruptive innovations,. Europe should do the same and there is no time to lose, he said in an interview with The Innovator.
Develop Leaner AI Models
Building AI factories is Europe’s answer to Stargate, an AI joint venture that aims to build data centers in the U.S. for OpenAI, says Loesekrug-Pietri. “We are fighting yesterday’s battle.”
OpenAI, Softbank Group, Oracle and MGX are the initial equity funders and Arm, Microsoft and Nvidia are technology partners in Stargate. The companies are committing $100 billion initially, but plan to invest up to $500 billion over the next four years. “
Since the announcement of Stargate, the AI landscape has changed with DeepSeek proving it is possible to do more with less. Europe should instead be focusing on developing leaner models and AI applications, says Loesekrug-Pietri.
That view is shared by Rasmus Rothe, Founder and General Partner of Merantix Capital. “This year, and specifically the recent DeepSeek news, showed that the focus—and obsession—within Europe to compete at the LLM [large language model] game was wrongheaded,” Rothe was quoted as saying in the 2025 European Deep Tech report. “As many of us have argued, just like Europe didn’t need to waste time and resources on creating a European search engine against Google in the last tech revolution, we don’t need a ‘European OpenAI’ now, as these models continue to become commoditized and open source. What we need to do is focus on our strengths: applying AI at the application layer to specialized domains in industries where Europe has the talent and proprietary data sets, from manufacturing to health to finance to logistics and more. Europe stands to really benefit from this new reset in AI, but we need act fast and think really big this year.”
Resolve Compute And Energy Limitations
The Continent isn’t expected to catch up with the U.S. on compute anytime soon. The U.S. built more data-center capacity in 2023 than the rest of the world combined, excluding China, according to recently released report by the Tony Blair Institute For Global Change (TBI) TBI’s data-center investment indicator shows that the U.S. has announced that 5,796.2 megawatts’ (MW) worth of data centers are to be built over the next three years. This is a 164.4% rise from 2023 and shows no sign of slowing.
These data centers are consuming more and more energy. That’s a problem for Europe, where energy is considerably more expensive than in the U.S.
The EU plan commits to triple data-center capacity in Europe by 2035, through a “Cloud and AI Development Act”—but there are already significant challenges to accommodate growing demand, notes Verdi. In Dublin, for example, authorities have halted data-center buildouts until 2028. Europe needs to address bottlenecks swiftly if it is to realize its objectives, says Verdi.
In the short-term, he suggests establishing special AI zones in European regions which would benefit from accelerated permit for buildouts, and which have a high potential for renewable energy generation. In the long-term, the EU should consider the role of small modular reactors as a complementary clean-energy source and create an EU AI energy council comprised of national authorities, civil society and experts, he says.
Don’t Focus on AI Factories
AI factories are not what startups want, says Stef Van Grieken, a biotech and AI founder who previously worked as a Senior Product Manager for Google AI (Brain). He says he finds the fact that AI Factories are front in center of the European AI strategy concerning. “It’s old wine in new bottles,” says Grieken, who is currently CEO of Cradle, a Swiss-Dutch startup that applies generative AI to biology to reduce the time and cost of research and development in designing protein-based products. (See The Innovator’s story on Cradle). “These ‘AI Factories’ are essentially EuroHPC supercomputers being rebranded for AI,” he says.
In a LinkedIn post Grieken argues that there is a fundamental mismatch between startups and AI factories: He cites three reasons why they are not fit for purpose:
- -AI factories are built for physics simulations, not AI workloads.
- They’re inflexible compared to modern Cloud infrastructure and lack a bunch of other relevant services to build products.
- Depending on the age, in most cases they will require significant rework of models and infrastructure software to make them even work. “Which is not something you should be spending time on as a startup,” he says.
Manage Cloud Sovereignty Risk
If Europe wants to compete globally in AI, it should focus on asking founders of scale-ups what they need to succeed, says van Grieken. Existing Cloud solutions work well, he says, and they’re built for purpose.” If we worry about EU Cloud Infrastructure sovereignty, we should try to find ways to be able to decouple the sovereignty risks current cloud providers pose,” he says in a LinkedIn post. “For example, we could put European Cloud provider infrastructure under separate legal entities with strict legal and technical controls to manage sovereignty risk.”
Fix The Demand Problem
EU startups are not interested in accessing Graphics Processing Units in Europe because demand for their services is in the U.S., according to tech industry executives. To succeed in AI the EU needs to stress the importance of European governments and corporations buying from local AI providers, says Lechelle. “There is a need to reorient the demand.”
Create A Real Single Market
One of the most important things the EU could do is to create a true single market, Loesekrug-Pietri said in an interview with The Innovator. European startups have been hampered for decades by the lack of one, but little progress has been made. This makes the risk reward ratio much less compelling for investors in European startups.
Don’t Put All It Of The EU’s Eggs Into An AI Basket
Europe has multiple deep tech strengths; it should focus on a range of them instead of just emphasizing AI, says Finnish tech industry veteran Mikko Suonenlahti, President of the Digital Security & Trust Group of the EIC Scaling Club and a founder of Tyrion Partners, which specializes in finding capital for tech founders and deals for fund managers.
Europe is, in fact, a a hub of world-class innovation in deep tech, excelling in areas like quantum computing, energy, and robotics but despite exceptional technical talent and groundbreaking research, the region has produced few global leaders in these transformative fields, notes the 2025 European Deep Tech report. “To bridge this gap, increased late-stage capital, a focus on global leadership—not just regional excellence— and a strong support ecosystem are crucial to translate cutting-edge research into market-leading solutions,” it says.
Suonenlahti, who is Chairman of the Board of Cubbit, a Europe Innovation Council-backed Italian Cloud storage business. says the answer is increasing public and private sector investment in deep tech funds and startups. He thinks the objective of the EU’s AI Continent plan should be the building of several European deep tech startups with valuations in the billions that become global market leaders.
“But it is not just capital that is needed,” Suonenlahti, who is in his fifth decade of working with founders of tech businesses in Europe and the U.S., said in an interview with The Innovator. “For growth to happen European startups need strong boards.” At least 60% of European deep tech companies do not have a formal board, according to Barcelona-based IESE Business School analysis and other sources, he says. What’s more there is a strong need for independent directors of Europe’s deep tech companies because “most of them do not effectively implement a go-to-market strategy,” he says.
Address Problems That Have Been Recognized For Years
Some of the biggest challenges that remain in Europe are to encourage more entrepreneurs to move into deep tech, to harmonize university spin-out terms, to form more dense talent / excellence clusters, to increase the limited partner base and involvement of institutional investors, to strengthen the exit channels, and to promote diversity across founders and investors, says the 2025 European Deep Tech report. It notes that progress on these issues has been limited.
Cultivate A Sense Of Urgency And Move From Talk To Action
“In this tense, volatile geopolitical landscape we can no longer constrain our definition of impact investing,” European serial entrepreneur Niklas Zennström, Founding Partner and CEO of London-based venture firm Atomico, said in a LinkedIn posting. “Yes, it’s still about climate, health, and sustainable solutions,” he says. “But it’s also about something broader: investing in Europe’s ability to operate on its own terms. Our security, our energy systems, our healthcare, our food supply, our digital infrastructure. The companies that will create jobs, drive GDP growth, and make sure we aren’t dependent on others to thrive.” There is now a shared understanding that Europe needs to support sovereign innovation and strategic infrastructure with serious capital, long-term thinking, and global ambition, says Zennstrom. “But we need to do more to make sure this translates from conversation into action.”
Promote Top Tier Research And No Bros Or Oligarchs
Europe has plenty of things going for it. For starters it boasts top-tier research and is home to six of the top 20 universities and nine of the top 25 research institutes. And the quality of life in Europe is better than most place in the world. European Commission President Ursula von der Leyen recently summed it up this way during an interview with the German newspaper Zeit. “We don’t have bros or oligarchs making the rules. We don’t invade our neighbors, and we don’t punish them. On the contrary, there are twelve countries on the waiting list to become members of the European Union. That’s about 150 million people,” she said, referring to the accession process. “In Europe, children can go to good schools, however wealthy their parents are. We have lower CO2 emissions; we have higher life expectancy. Controversial debates are allowed at our universities. This and more are all values that must be defended, and which show that Europe is more than a union. Europe is our home. And people know that.”
Attract The Best And The Brightest
The turmoil caused by U.S. President Trump is making Europe more alluring for other reasons. American scientists are starting to ask for asylum in Europe and foreign students and researchers, who are now less likely to ask for or receive U.S. visas, might be enticed to come to Europe, says Suonenlahti.
While skeptics say it will be impossible for Europe to match U.S. salaries, JEDI’s Loesekrug-Pietri argues that the Continent can attract the best people by offering them the chance to work on the most exciting projects. “We can pose the most challenging, complicated problems and offer them the chance to help crack them with AI,” he says. “If we get it right this really could be Europe’s moment.”
This article is content that would normally only be available to subscribers. Sign up for a four-week free trial to see what you have been missing.
To access more of The Innovator’s Innovation Strategy articles click here.
