Europe’s velocity and direction of artificial intelligence policy contrast with that of the new Trump administration. Agree or disagree with Trump’s politics, the tempo, and momentum are remarkable: Within two weeks, a first executive order removed Biden-era AI safety restrictions and made federal land and resources available for data center construction.
The European Commission’s new AI Continent Action Plan comes up short, particularly when compared to the rapid and aggressive moves coming out of Washington and Beijing. Europe needs to be bold: concentrating on open source, boosting R&D funding, and injecting AI into its military buildup. Importantly, it should work with US tech, not scorn it.
Below, the European plan’s five pillars are unpacked — along with my humble suggestion of the steps to take if the continent is serious about AI.
Infrastructure
The Action Plan spreads €2 billion across university-run supercomputer clusters, €20 billion to five “AI gigafactories” and to triple data-center capacity by 2032. That barely registers against the trillion-dollar spend under way in the United States and China.
If Brussels were serious, it would:
- Redirect gigafactory funds to the cloud. Underwrite commercial compute providers with long-term contracts or subsidized demand, aiming to pull the 2030-32 target of tripling data center capacity forward to 2028.
- “Europeanize” US hyperscalers. Offer a clear path to full market access (insulated from potential trade retaliation) if the US cloud providers localize data, site R&D in Europe and comply with EU security rules—turning a dependency into domestic capacity rather than cutting the continent off from the best tools.
Data
Everyone agrees data is AI’s rocket fuel, but only Europe insists on fueling rockets through a thicket of red tape. In the US, AI developers tap rich troves of data from social media, e-commerce, and cloud services with few restrictions. In 2023, China established a National Data Bureau to coordinate data policy and has been building data trading markets.
If Brussels were serious, it would:
- Rewrite GDPR privacy rules for an AI era. Introduce safe-harbor provisions for clearly anonymized, high-public-value datasets and accelerate legitimate-interest processing.
- Double down on open data. Rigorously enforce existing rules opening public sector and high-value datasets, push national governments to go beyond EU-level requirements and publish a continental data map listing every significant public- and private-sector dataset, its legal status and a contact point, updated by AI crawlers.
- Reward sharing. Offer grants or tax credits to organizations that release high-quality data under open or public-good licenses.
AI Adoption
Only 13 % of European companies say AI is core to their business, versus 49% in the US and 83% in China. Europe’s usual fix – strategies, hubs, pilots – is too slow.
If Brussels were serious, it would:
- Set moon-shot missions with hard KPIs. For instance: Health: AI reads 90 % of cancer screenings by 2030. Mobility: 25 % of EU urban traffic lights are adaptive by 2027. Agriculture: real-time crop analytics on 50 % of farms by 2030.
- Procure outcomes, not paperwork. Make outcome-based public contracts—“cut patient wait-times 20 %”—the default above €5 million, and let AI-native challengers compete.
Skills and Talent
Europe’s proposed AI Skills Academy and modest migration tweaks will not stem the AI talent hemorrhage. Silicon Valley already employs an outsized share of Europe’s top machine-learning graduates; China is minting domestic PhDs at industrial scale.
If Brussels were serious, it would:
- Endow 1,000 AI Chairs, modelled on the successful European Research Agency Chairs. Fund these new positions for 10 years and lure senior researchers back from the US.
- Pivot Erasmus and the European Social Fund. Carve out 10-20% of Europe’s impressive continent-wide scholarship program to offer AI-rich joint degrees and domain-specific programs—AI & medicine, AI & law, AI & energy and re-direct 5-10% of Europe’s current €140 billion social spending programm to AI reskilling.
- Launch an AI Talent Scoreboard. Publish annual metrics on MSc and PhD output, median salaries and net migration for every European country.
(De)Regulation
The European Commission made waves when it withdrew its plans for an AI Liability Directive and appeared to signal a willingness to amend the AI Act (and GDPR privacy rules) to facilitate AI adoption. But the actual steps for simplification taken look insufficient. The Commission has proposed an AI Act Service Desk to assist in implementing the AI act. There’s no actual simplification of the legislative rules.
If Brussels were serious, it would amend the AI act to:
- Carve-out B2B safe harbors. Turn the AI-Act’s risk tiers into actual gates: low-risk uses pass with minimal obligations; high-risk uses face accelerated (30-day) audits via pre-certified templates.
- Introduce passporting. One national authority, one decision—valid across the Single Market. Remarkably, unlike much EU regulation, the AI Act presently allows any EU AI regulator to investigate any AI solution provider.
- Whitelist mature, low-risk applications. Route-optimization for public transport, visual inspection in manufacturing, etc., should be exempt outright.
Perhaps most noteworthy are the topics on which the AI Action Plan says little:
- Defense and Security. The plan is mute on military or intelligence AI – though this has been crucial to Ukraine’s military success. As Europe rearms, it should built its capabilities for future wars with outcome-driven R&D that mirror how the Pentagon funds through its DARPA agency.
- Open-source, Open Weights, Open Data. Open Source AI represents last year’s breakthrough surprise (e.g., the “DeepSeek moment”). Collaboration overcomes secrecy and silos, yielding models and products that are functionally superior, cheaper and with less lock-in. A preference for open source makes particular sense for an underdog challenger. China has embraced this approach in its AI competition with the US. The EU’s open source mandate (and funding) should be strengthened.
- Global leadership. With the US pivoting inward and China’s governance model unpalatable to many democracies, Europe could lead a “rest of the world” coalition of countries building open, rules-based AI. That means building and investing alongside the UK, Canada, Japan, South Korea, Ukraine and many other partners, not building Fortress Europe.
Intelligent minds will disagree on the specific merits of these proposals. What should not be a matter of debate, is that Europe needs to taking concrete, bold steps. Europe has the talent, market and even financial heft to catch up with China and the US – but only if it aims high.
Luukas Ilves is non-resident fellow at the International Center for Defence Studies in Tallinn and an advisor to the Vice Prime Minister of Ukraine. He was previously Estonia’s Undersecretary for Digital Transformation and Government Chief Information Officer. This piece reflects his personal views.
Bandwidth is CEPA’s online journal dedicated to advancing transatlantic cooperation on tech policy. All opinions expressed on Bandwidth are those of the author alone and may not represent those of the institutions they represent or the Center for European Policy Analysis. CEPA maintains a strict intellectual independence policy across all its projects and publications.
2025 CEPA Forum Tech & Security Conference
Explore the latest from the conference.
