Europe is trying to break its dependence on U.S. Big Tech—and it’s doing so with a blend of industrial-scale investments, public‑private partnerships, open‑source strategies and new legal frameworks that together aim to build what policymakers and advocates call a “EuroStack.”
Background: why Europe sees a problem
Europe’s sense of urgency is not rhetorical. For more than a decade the continent has applied tough regulation to Big Tech—GDPR, the Digital Markets Act (DMA) and the Digital Services Act (DSA) among them—but regulation alone did not create European alternatives at scale to the cloud, AI models, or hyperscale infrastructure that power modern digital services. The result is a structural dependency: the three U.S. hyperscalers dominate the cloud market in Europe, and a disproportionate share of compute and platform value continues to flow offshore.
That dependency has strategic consequences. When critical infrastructure, datasets or compute live under foreign providers’ control, policy levers and geopolitical friction can translate into economic vulnerability. Recent tensions on the transatlantic relationship—from trade frictions to renewed, high‑profile rhetoric about sovereignty—have hardened European political will to act. Whether that rhetoric is focused on trade, national security, or even dramatic headlines about territorial ambitions, the political effect is the same: European governments now treat digital infrastructure as a strategic asset, not merely a commercial market.
Overview: what “breaking up” means in practice
“Breaking up” is a blunt phrase. Practically speaking, Europe is pursuing a multi‑track strategy combining:
- Build: Fund and host large‑scale AI compute, data centres and chip supply chains in Europe.
- Buy/localize: Encourage European enterprises and governments to shift procurement toward European providers and sovereign clouds.
- Open: Lean on open‑source, federation and interoperability rather than trying to clone a single Silicon Valley heavyweight.
- Regulate smartly: Recalibrate regulations to protect privacy and values while removing choke points that make scale‑ups impossible.
- Partner: Recruit industry champions across the semiconductor, software and enterprise ecosystems to co‑invest.
Each of these elements is in motion, but the balance of ambition and realism differs by sector. The EU’s public plans—under umbrella efforts such as the
AI Continent Action Plan and the InvestAI facility—are concrete attempts to convert ambition into capacity.
The core problem: hyperscalers, market share, and the “digital colony”
The numbers you need to know
European analysts and policy reports have repeatedly quantified the cloud imbalance. The three U.S. hyperscalers—
Amazon Web Services (AWS),
Microsoft Azure and
Google Cloud—account for a large majority of Europe’s cloud infrastructure market; estimates commonly cluster in the mid‑60s to high‑60s percent range for the early 2020s. European cloud suppliers, by contrast, hold a low‑teens share. These are not marginal gaps: they capture where the continent’s compute, data and many commercial services reside today.
Those market shares explain why commentators such as economists and policy advocates describe Europe today as a “digital colony”: revenue and value accrued to U.S. centralised platforms, while European capital often flows abroad rather than building regional champions. Prominent voices from industry and economics have used that phrase to stress the political and economic urgency of re‑rooting core infrastructure in Europe.
What this concentration means practically
- Vendor lock‑in for European companies and governments.
- Regulatory friction when U.S. platforms clash with EU rules (e.g., data access disputes).
- Capital outflows as subscription and license dollars enrich non‑European firms.
- Difficulty for domestic startups to compete when hyperscalers can bundle compute, storage and AI services at scale.
Those aren’t theoretical risks; they alter procurement decisions, industrial policy and defence planning, and they help explain the political impetus behind the EuroStack conversation.
Building the EuroStack: pragmatic choices, not cloning Silicon Valley
EuroStack as an idea and network
EuroStack is less a single company and more a movement and set of policy proposals advocating a European digital stack built around sovereign infrastructure, open standards and industrial cooperation. It has roots in academic and industry writing and is championed publicly by economists and policy figures who argue that Europe must stop relying on regulation alone and start investing in the physical and software infrastructure that underpins the internet. The idea has traction in think tanks, industry consortia and among several national governments.
Cristina Caffarra, one of the visible public proponents, frames the problem bluntly: Europe has ceded much of the underlying infrastructure and now must
build rather than simply
police. That line captures why the conversation has shifted from antitrust and privacy to procurement, capital deployment and infrastructure.
Why not just build “European AWS”?
European policymakers have consciously avoided the singular goal of recreating a single Silicon Valley giant. Instead they emphasize federated solutions—shared, interoperable infrastructure—plus targeted industry champions in software, chips and data centres. The logic is pragmatic: Europe’s political, financial and market structure makes recreating one vertically integrated hyperscaler difficult and possibly unnecessary. Open‑source, federation (think Gaia‑X and the EU digital commons concept) and public‑private consortia are seen as more realistic paths to scale while preserving values like data protection and interoperability.
The heavy lifting: AI factories, gigafactories and InvestAI
The EU’s infrastructure play
In 2024–2025 the Commission launched an
AI Continent Action Plan focused on establishing multiple “AI Factories” (specialized, accessible compute hubs) and a smaller set of larger
AI Gigafactories—massive public‑private computing facilities designed to host frontier training and development. The EuroHPC Joint Undertaking and the Commission have enumerated plans for at least a dozen AI Factories and the procurement of additional AI‑optimised supercomputers to increase Europe’s shared compute capacity.
InvestAI: what the official paperwork actually states
The InvestAI facility—part of the Commission’s package—is expressly designed as a public‑private financing vehicle to support major AI infrastructure, including up to five AI Gigafactories. The most consistent official figure is an InvestAI Facility target of around
€20 billion to catalyze large‑scale public‑private investment for AI infrastructure (the Commission and EuroHPC documentation use this figure). Other reporting has discussed larger aggregate mobilization targets (figures up to €200 billion appear in secondary analysis as possible total mobilized capital via associated industry pledges), but those higher totals mix private pledges, national contributions and extrapolations; they are not identical to the InvestAI Facility’s initial, legally framed commitment. Readers should therefore treat the €20 billion figure as the formal EU facility target and larger aggregate numbers as aspirational or partially pledged totals reported by analysts.
Why scale matters — and what Europe actually needs
Training frontier AI models requires concentrated compute, interconnect, and energy. The EU’s plan to fund shared
AI Factories and a limited number of
AI Gigafactories is a recognition that small, scattered compute budgets won’t support frontier training runs. But even a handful of gigafactories will not instantly close the gap with U.S. hyperscalers and major cloud providers; building and operating these facilities requires time, skilled labour, and secure, affordable power. The Commission makes clear this is a multi‑year, multibillion‑euro infrastructure agenda, not a short sprint.
Private industry muscle: chips, machine learning startups and data centers
ASML + Mistral: a European axis
Europe has unique industrial strengths:
ASML (Netherlands) is the only company that has commercialized
extreme ultraviolet (EUV) lithography—critical for the most advanced chips—and wields outsized strategic importance in the semiconductor value chain. Recent, high‑profile investments such as ASML’s large funding into the French AI startup
Mistral AI underscore how European industrial champions can align hardware and AI expertise—combining lithography mastery with model development capacity to keep parts of the value chain in Europe. Those partnerships are a clear strength in any sovereignty strategy.
European cloud and data‑centre bets: Schwarz Digits and StackIT
Private players are stepping up. Schwarz Digits—the IT division of the Schwarz Group (owner of Lidl and Kaufland)—announced a major data‑centre campus investment in Lübbenau, Brandenburg: roughly
€11 billion committed to a modular facility with initial power allocation in the hundreds of megawatts (around
200 MW) and capacity to host large numbers of GPUs. Projects like this are explicitly pitched as contributions to Europe’s digital sovereignty, offering locally controlled compute for enterprise and sovereign workloads. That is the very kind of large private project the Commission hopes to catalyze with its regulatory and financing packages.
Enterprise software: SAP and sovereign AI
Large European enterprise software companies also see an opportunity. SAP has publicly expanded partnerships with European AI firms to offer sovereign AI stacks and cloud services tailored for European compliance and enterprise needs. This is important: enterprise software providers have customer relationships, sectoral reach and the credibility to deliver compliant stacks to governments and regulated industries. Those partnerships make the idea of a European sovereign AI stack more realistic for corporate and public customers.
Energy and geography: the physical constraints that matter
Power is not an afterthought
Building gigafactories and hyperscale data centres requires
cheap, reliable energy—and that is where Europe faces real constraints. The continent’s energy prices spiked after the Russian invasion of Ukraine in 2022, pipeline flows were reduced dramatically, and although renewables have accelerated, the shift to a low‑carbon grid and faster permitting for grid upgrades takes time. High electricity and gas prices raise the operating cost per GPU or per petaflop, and that affects the economics of domestic AI infrastructure compared to U.S. peers with cheaper power in many locations. European planners are explicitly factoring this into siting decisions and the design of public support mechanisms.
Location, permitting, and timelines
Data‑centre projects in Europe also hit non‑energy constraints: grid connection queues, environmental permitting, land availability and local community concerns. Analyses repeatedly show multi‑year waits for grid connections in major European hubs; those delays matter because compute demand cannot be decoupled from the pace of infrastructure build‑out. The Commission’s Cloud and AI Development Act seeks to reduce bureaucratic friction, but legal and social constraints cannot be waived overnight. This is a structural hurdle to the EuroStack timetable.
Strengths of the European approach
- Industrial depth in specialized niches: Europe hosts world‑leading equipment and precision engineering firms (ASML) and powerful enterprise software players (SAP) that can anchor value chains.
- Public legitimacy and values: The EU’s regulatory model gives European providers a comparative advantage in data protection and compliance for regulated sectors (healthcare, public administration).
- Open-source and federation model: Emphasizing open standards and shared infrastructure lowers barriers for SMEs and public bodies to collaborate cross‑border, instead of forcing a winner‑takes‑all hyperscaler model.
Key risks, trade‑offs and unintended consequences
1. Energy and economics
If European power remains structurally more expensive than alternative locations, Europe will build fewer gigafactories for the same money. High operating costs could make European facilities unattractive except for the narrowest use cases (regulated public services, defence, or firms that pay a premium for sovereignty). The industry must solve energy sourcing and grid upgrades in parallel with compute investments.
2. Over‑reliance on national champions and supply chains
ASML’s unique position is a strength—and a single point of criticality. Geopolitical export controls and fragile supply chains for ASML components expose the broader strategy to second‑order risks. A EuroStack that relies on a handful of corporate champions is vulnerable if those companies face sanctions, trade restrictions or technical setbacks.
3. Regulatory contradictions: protect vs. promote
Europe’s regulatory DNA is protective and rights‑oriented. That is a competitive advantage for trust, but the same rules can slow product development and hinder venture capital flows. The Commission is already moving to simplify some rules for AI and cloud access, but balancing
safety with
speed to market is politically fraught. Diluting protections to chase scale would be dangerous; clinging to them without a robust industrial policy will be ineffective.
4. Capital markets and scaling challenges
European capital markets remain fragmented compared to the U.S. ecosystem that underpins Silicon Valley’s runaway scale. Building a hyperscaler‑like firm requires patient, large scale capital and regulatory tolerance for loss‑making growth. The EU’s new funds and national commitments are meaningful, but creating the dynamic that produced companies of the scale of AWS, Google or Microsoft is not guaranteed.
5. Geopolitical friction and retaliation
A sharp pivot away from U.S. providers—if enforced by procurement rules or protectionist measures—could provoke political pushback and even reciprocal measures. Europe’s strategy must therefore be surgical and cooperative where possible; solidarity rhetoric should not morph into closed markets that hurt industrial competitiveness. Recent diplomatic strains and headline risks make the politics sensitive.
What success would look like — and how to get there
Success for the EuroStack is not a single, monolithic European hyperscaler swallowing AWS; it is a resilient, interoperable ecosystem that meets European public‑policy needs and hosts a substantial share of continental compute and data. Specific measures to move toward that outcome include:
- Prioritize a pragmatic portfolio: fund a small number of gigafactories and many more AI Factories, ensuring access for startups and research consortia.
- Tie public support to energy commitments: require gigafactories to demonstrate long‑term green energy procurement and grid stabilization plans.
- Incentivize private capital: create blended finance vehicles that de‑risk long‑horizon infrastructure investments for pension funds and institutional investors.
- Double down on open‑source, federation and interoperability to lower switching costs and stimulate multi‑vendor markets.
- Strengthen European industrial cooperation: link semiconductor initiatives, lithography capacity, and AI infrastructure plans in coordinated supply‑chain roadmaps.
These are sequential and interdependent steps; none works in isolation. Europe’s ambition must be matched by fiscal patience and operational coordination.
The short‑term outlook: realistic expectations
Expect incremental wins rather than immediate parity. The EU’s investments in AI Factories and a €20‑billion InvestAI Facility are substantial and meaningful; they will deliver capacity to academic researchers, startups and some corporate users. High‑profile private investments—Schwarz Digits’ Lübbenau campus and ASML’s strategic funding into leading AI startups—show the private sector is willing to co‑invest. But major systemic rebalancing of the cloud market will take years and require sustained political and industrial alignment. The timetable is measured in half‑decades, not months.
Final assessment: hopeful realism
Europe’s push for a EuroStack is a credible strategic response to genuine vulnerabilities. It blends the continent’s engineering strengths with new funding mechanisms, and it has the political momentum of a bloc increasingly aware that digital sovereignty is both an economic and a security imperative. At its best, the approach leverages Europe’s values (privacy, openness, competition) and its industrial base (equipment makers, enterprise software firms, major corporations) to create a more resilient, interoperable ecosystem.
But ambition meets reality: energy prices, permitting bottlenecks, capital market friction and global supply‑chain realities will slow progress and impose costs. Europe must therefore avoid two traps: (a) thinking regulation alone can rebuild supply chains, and (b) swinging to protectionism that damages competitiveness. The most plausible path to meaningful sovereignty is a mixed strategy that coordinates public funds, private capital, industrial partnerships and open architectures—an approach that builds capacity while retaining access to global innovation.
Europe’s break‑up with U.S. tech is not a divorce decree. It is the start of a multi‑year effort to rebalance where value is created and where critical infrastructure sits. If managed intelligently, it can deliver a more plural, resilient internet architecture for Europe—one that preserves the bloc’s values while giving its companies and governments the tools to compete on the global stage.
Source: Sherwood News
Europe wants to break up with US tech. Here’s how it plans on doing that.