EU's AI ambitions are like a mighty oak: dreaming of touching the sky, but tripping over its own roots

Do you know how a €200 billion question sounds? Try this: Can Europe vault from regulator to innovator?

In the global AI arms race, Europe has traditionally been the cautious third wheel—strong on principles but light on execution. But February 2025 marked a turning point in Brussels' approach: the unveiling of 'InvestAI', a €200 billion initiative to transform Europe from AI rule-maker to market-maker.

This moonshot comes as the U.S. advances its $500 billion 'Stargate' AI project and China celebrates breakthroughs like DeepSeek. Europe, sensing existential technological irrelevance, is now trying to secure a seat at the AI table. However, a critical question remains: Is InvestAI the visionary initiative that bridges Europe's AI gap, or is it another well-intentioned bureaucratic endeavor destined for the growing graveyard of EU tech ambitions?

The Challenge: Overcoming "Regulate, Not Create"

Europe's position in global technology has become almost a punchline: "America innovates, China replicates, Europe regulates." While unfair to Europe's considerable research strengths, the caricature contains uncomfortable truths about execution.

Europe's tech companies invest only a fraction of what their American and Chinese counterparts commit to AI capital expenditures. Although there are positive examples, such as large European firms like Novo Nordisk and Spotify partnering with AI startups and investing in their own AI capabilities, as well as Europe-born startups like ElevenLabs and DeepL leading in their respective fields, these cases are exceptions in a continent where regulation often surpasses innovation.

The Draghi Report crystallized this challenge, advocating for a fundamental shift in Europe's approach:

• Coordinated industrial policy across the EU

• Targeted protectionist measures for strategic sectors

• Expansion of public-private partnerships

• Massive investment in computational infrastructure

InvestAI represents the most ambitious manifestation of this new industrial strategy. Yet, without concrete plans for harnessing Europe's talent and research resources, legitimate questions surround the program's ability to deliver on its lofty ambitions.

The Public-Private Puzzle

The headline €200 billion figure comes with an important caveat: three-quarters must come from private investors under the 'European AI Champions' initiative, a coalition of more than 70 firms across the continent.

This highlights Europe's conventional approach to technological advancement: using public funds to reduce risks tied to private investment. However, this model has produced mixed outcomes thus far: while advancements have occurred, returns haven't materialized to the extent needed to create self-sustaining ecosystems. In many instances, private investors remain cautious despite government efforts.

Perhaps most concerning is the composition of the European AI Champions coalition:

• Over 40% consists of AI startups and tech firms

• Traditional industrial players – manufacturing, agriculture, healthcare – remain underrepresented

• Limited participation from data-rich industries hampers the potential for AI development

It's exciting to see so many European tech players involved. However, this imbalance threatens the initiative's success. Without participation from Europe's traditional economic engines, access to the large datasets critical for AI development will likely remain a significant bottleneck.

Europe's Computing Gambit

The centerpiece of InvestAI's strategy is the development of "AI factories" – massive computational infrastructure hubs equipped with 100,000 high-power chips each. For context, these facilities would be four times larger than Jupiter, currently Germany's largest supercomputer.

The economics are sobering:

• Current prices for cutting-edge Nvidia GPUs hover around $40,000 each

• A single AI factory would require multiple billions in investment

• The EU's access to advanced AI chips faces restrictions from U.S. export controls

For comparison, Meta is currently investing $10 billion to build a 1.3 million GPU facility in Louisiana, U.S., which significantly exceeds the capabilities of an ‘InvestAI’ factory. Additionally, the useful lifespan of these computing facilities is approximately 18 months. In the absence of European counterparts to OpenAI, Google, or Amazon driving demand for this scale of computing, the long-term prospects for these investments remain uncertain.

As indicated above, we believe the initiative's success heavily relies on significant demand from sectors such as manufacturing, agriculture, and public services, which are essential to the European economy but are likely to fall behind in developing AI capabilities because of outdated systems, CAPEX constraints, and resistance to change. Furthermore, we are currently hearing disturbing feedback from some large European corporations expressing that, due to the current economic climate, they are intentionally delaying new, research and development-intensive projects while they await better economic conditions. While this last point offers only anecdotal evidence, it is nevertheless disconcerting.

Location, Location, Location

Let's take a closer look at the planned AI gigafactories across the continent. Some of the selected sites already have operational supercomputer centers:

• Athens (Greece) – Pharos AI factory with the DAEDALUS supercomputer

• Barcelona (Spain) – Integration with the 'MareNostrum 5' supercomputer

• Bologna (Italy) – CINECA - Bologna Tecnopolo

• Bissen (Luxembourg) – Meluxina-AI factory with 'LuxProvide'

• Kajaani (Finland) – LUMI AI Factory alongside Europe's most powerful supercomputer

• Linköping (Sweden) – MIMER AI factory at Linköping University

• Stuttgart (Germany) – 'HammerHAI' consortium with the University of Stuttgart

The most critical constraint for these hubs isn't real estate—it's energy. Each facility requires approximately 0.5 GW of stable power, a demand that exceeds the available capacity in most locations.

Currently, only five EU member states – Finland, France, Germany, Ireland, and Sweden – can meet the infrastructure, research, and energy requirements. France, utilizing its nuclear capacity, is particularly well-positioned, with state-owned utility EDF in discussions to power three data centers that require 1 GW each. Finland, Germany, Ireland, and Sweden can manage such loads through grid interconnections and controlled energy resources, but they may lack the necessary excess power locally for factories.

Other announced sites in Vienna, Sofia, Poznań, and Maribor will require substantial energy investments before becoming viable. This reality threatens to widen the innovation gap within Europe rather than close it, as AI capabilities cluster around a handful of energy-rich locations.

All sites of AI factories have direct access to AI research talent, and are located in urban areas, where significant investment in AI infrastructure development can nurture further growth and competitiveness of vibrant ecosystems with spill-overs into the economy. AI factories will further raise the attractiveness of the mentioned locations for venture capital, with aggregation of investors focused on AI and government programs targeting AI capability development in early-stage firms. For instance, Barcelona AI Factory’s development is already in progress, leveraging the benefits of the ‘MareNostrum 5’ supercomputer at the Barcelona Supercomputing Center (BSC), talent resources, quantum computing initiatives, and strong government support that the project has received, with potential broader benefits for evolving Barcelona’s start-up ecosystem.

It seems that the sites announced in March 2025 were chosen for reasons other than energy resources. Without a stable energy supply of over 0.5 GW, these factory sites will likely face significant delays. The goal of the ‘InvestAI’ initiative is to promote equal growth in AI infrastructure across EU Member States by providing financial support and improving policy coordination. However, without a clear strategy for executing complex projects, 'InvestAI' may worsen the innovation gap in the EU. AI development might cluster in a few locations that receive support and have high-power AI infrastructure. Meanwhile, Eastern EU member states are lagging in developing this infrastructure, leading to decreased competitiveness and a loss of talent and resources, which will widen existing gaps in the EU.

Member State Initiatives: Four Approaches

While InvestAI aims to unify Europe's AI strategy, individual member states continue pursuing their own initiatives. These national approaches reveal differing priorities and capabilities:

France: The Infrastructure Leader

France has emerged as Europe's most ambitious AI investor. In February 2025, President Macron announced a €109 billion initiative to develop computing and data center infrastructure.

• 'France 2030' plan includes €2.5 billion for AI development

• Nine AI clusters with research and training centers (€360 million commitment)

• Goal to train 40,000 AI-ready students annually

• 81 AI hubs across the country, many hosting American tech companies

• Station F and other accelerators have received over €1 billion in support

Italy: The SME Champion

Italy has focused its more modest resources on integrating AI into its small and medium enterprise ecosystem:

• 'Italian Strategy for Artificial Intelligence 2024-2026' aligns with EU objectives

• Prioritizes the broader deployment of venture capital to AI startups

• Ensures local firms can access high-performance computing from InvestAI

• Partnership between OpenAI and state lender CDP to promote AI adoption

• Istituto Italiano per l'Intelligenza Artificiale (AI4I) serves as the hub for coordinating research and private sector collaboration

Poland: The Defense-First Approach

With concerns about EU program benefits and significant defense priorities, Poland has made more targeted investments:

• €2.5 billion 'East Shield' initiative for defense technologies, including AI systems

• €220 million 'National AI Development Plan' for infrastructure and research

United Kingdom: The Post-Brexit Question?

Brexit has led to corporate investment stagnation, limiting the UK's AI ambitions:

• Pledged future increase in AI computing capacity by 2030

• 'AI Opportunities Action Plan' focused on public sector adoption

• Earlier 'AI Sector Deal' allocated £300 million to research and co-investment schemes

The Road Ahead: Three Critical Questions

Like a centuries-old oak tree, Europe's AI ambitions demonstrate both strength and vulnerability. Oak trees can live for hundreds of years and grow to impressive heights, much like Europe's long-term technological vision. However, oaks often develop shallow, surface-level root systems that, despite their spread, make these mighty trees surprisingly vulnerable to strong winds and storms.

This perfectly mirrors Europe's current AI predicament: while its ambitions reach skyward with impressive scope (€200 billion InvestAI initiative), its foundational elements - energy infrastructure, private sector commitment, and fragmented national strategies - remain surprisingly superficial. Just as an oak's surface roots can be its undoing during severe weather, Europe's inability to deeply embed its AI initiatives into its industrial and technological bedrock could leave its grand ambitions exposed to global competitive pressures. Furthermore, like an oak's extensive but shallow root system that competes with nearby plants for resources, Europe's disparate national AI strategies might actually compete with each other rather than create a unified, robust ecosystem.

As Europe moves forward with its AI ambitions, three questions will determine whether InvestAI succeeds or joins the continent's collection of well-intentioned but underwhelming tech initiatives:

1. Energy security or AI security?

Europe's computational ambitions directly collide with its energy constraints. Without massive investments in power generation and distribution, many planned AI factories will remain blueprints.

2. Public investment or private execution?

The reliance on private capital for 75% of funding creates inherent fragility. Historical evidence suggests Europe's de-risking approach rarely generates the aggressive private investment needed for technological leadership.

3. Unity or fragmentation?

While InvestAI represents a unified approach, member states continue pursuing independent strategies. Italy's tight alignment with EU objectives contrasts with France's more autonomous approach.

Europe possesses the talent, research capacity, and economic heft to become an AI powerhouse. However, its complex approach to private sector mobilization, fragmented national strategies, and infrastructure limitations threaten to undermine its ambitious goals.

For Europe to transform from regulator to innovator, it must reconcile its grand visions with practical execution. The mighty oak must learn not just to dream of touching the sky, but to strengthen its roots.