EU unveils €20bn plan for AI gigafactories to close gap with US and China

The European Commission has put a €20bn price tag on its ambition to turn Europe into what it calls an "AI continent", unveiling plans for up to five gigafactories, massive sites housing supercomputers with more than 100,000 advanced AI processors each, while simultaneously opening the door to rewriting parts of the AI Act barely twelve months after the Parliament gave it final approval.

Executive Vice-President Henna Virkkunen presented the strategy on 9 April, framing the technology as central to European competitiveness, security and technological sovereignty. The global race for AI is far from over, she said. The numbers behind that claim are stark. Stanford University's 2025 AI Index found that US institutions produced 40 notable AI models in 2024, China produced 15, and Europe produced three, all of them French. A separate Stanford vibrancy ranking, measuring private investment, patents and research, placed no EU member state in the top five; the United Kingdom came third, France sixth, Germany eighth.

From factories to gigafactories

The EU is already funding 13 AI factories, sites where researchers develop and test models on supercomputers equipped with up to 25,000 advanced processors. The largest of those cost roughly €600m. Gigafactories are a different order of magnitude. The Commission envisions three to five of them, each costing €3bn to €5bn and exceeding 100,000 processors. The target is what the strategy document calls moonshots: breakthroughs in healthcare, biotechnology, industrial robotics and scientific discovery that require compute capacity far beyond what current European infrastructure can deliver.

The scale of the gap is not merely academic. Training frontier models now demands clusters of graphics processing units running for months. Without domestic capacity at that scale, European researchers and companies must rent time on US or Chinese clouds, creating strategic dependence and data governance risks. The Commission argues that gigafactories would also serve as anchor tenants for a European semiconductor supply chain, a point reinforced by a parallel push to develop home-grown AI chips.

The energy constraint

Power is the binding constraint. A single gigafactory could consume electricity on the scale of a mid-sized city. The Commission says facilities should run as much as possible on green energy, with water recycling for cooling. Yet 47% of EU electricity came from renewable sources in 2024, and the trajectory is not yet steep enough to absorb a wave of new industrial demand without extending the life of fossil generation.

John Hyland, a Greenpeace spokesperson, put it bluntly: Europe has made some strides towards fully renewable electricity, and still has a long way to go, but this could all be undone by ballooning energy consumption. Every power-guzzling datacentre risks being a lifeline to polluting power plants, well illustrated in Ireland where they consume over a fifth of the electricity. Ireland's experience is instructive: Central Statistics Office data shows datacentres accounted for 21% of metered electricity consumption in 2023, up from 5% in 2015. The Commission's own impact assessments acknowledge that without additional renewables deployment, new datacentre load will be met by gas.

Funding model and the chips gap

The €20bn is not a direct budget line. The Commission intends to use public funds to de-risk private investment through a facility managed by the European Investment Bank. That model, blending grants, guarantees and equity, has been used for the existing AI factories and for the Chips Act, but the sums involved here are larger and the revenue models less proven. Gigafactories are not commercial cloud providers; they are intended to serve public-good research and strategic industrial projects. Who pays for operations after the capital expenditure remains an open question.

The semiconductor side of the strategy is equally ambitious. Europe has no volume producer of the advanced logic chips, GPUs, TPUs, custom accelerators, that power modern AI. The Chips Act, adopted in 2023, targets 20% of global semiconductor production value by 2030, but that figure includes automotive and industrial chips where Europe is already strong. The gap in leading-edge AI silicon is wider. The Commission's new push for European-made AI semiconductors is essentially a second phase of the Chips Act, but without a confirmed budget or a clear path to competing with Nvidia, AMD or the Chinese firms now designing their own accelerators.

The AI Act simplification controversy

The same day the gigafactory plan was published, the Commission launched a consultation on possible simplification of the AI Act. The regulation, the world's first comprehensive AI law, was approved by the European Parliament in March 2024 and enters full application in August 2027. Most of its provisions are not yet in force. Virkkunen said the Commission remains very committed to the risk-based approach and the main principles, but is examining whether administrative burdens or reporting obligations could be cut to make life easier for businesses.

Critics see a contradiction. Frederico Oliveira da Silva, senior legal officer at the European Consumer Organisation (BEUC), said much of the AI Act does not even apply yet but the commission is already signalling that it is open to cutting some of its requirements. When we know that consumers worry about how AI could be used to harm them, the commission should instead be trying to reassure consumers. BEUC's position reflects a broader tension: the Act was designed to be technology-neutral and future-proof, yet the Commission is now treating its own regulatory framework as a potential drag on the very industry it is trying to subsidise.

Why the rush to rewrite?

The simplification drive is part of a wider competitiveness push led by Commission President Ursula von der Leyen's second term. Former ECB President Mario Draghi's 2024 report on European competitiveness argued that regulatory accumulation is stifling innovation and that the EU needs a radical simplification agenda. The AI Act, with its conformity assessments, post-market monitoring and documentation requirements for high-risk systems, is a natural target. But the timing, before a single enforcement action has been taken, suggests political pressure from member states and industry lobbyists has outweighed the case for waiting for evidence.

There is also a transatlantic dimension. The US has no federal AI regulation, relying on voluntary commitments and sectoral rules. The Trump administration has signalled a deregulatory approach. European firms argue they are competing with one hand tied. The Commission's consultation, open until June 2025, will test whether the simplification argument withstands scrutiny from the Parliament, which must approve any amendments, and from civil society groups that view the Act as a hard-won baseline.

Geopolitics and the sovereignty argument

Behind the industrial policy lies a geopolitical calculation. The US leads in models, talent, capital and compute. China leads in data, state-directed deployment and specific applications such as computer vision. Europe leads in neither. The gigafactory plan is an attempt to secure a seat at the table by controlling the infrastructure layer. If European researchers cannot train frontier models domestically, they cannot shape their values, safety properties or linguistic and cultural coverage. That argument resonates in capitals that remember the cloud wars of the 2010s, when European data migrated to US hyperscalers with limited legal recourse.

Yet sovereignty cuts both ways. The AI Act applies extraterritorially to any system placed on the EU market. If the Commission weakens its own rules to help domestic champions, it undermines the regulatory leverage that gives the EU influence over US and Chinese providers. The Parliament's rapporteurs on the AI Act have already warned that simplification must not mean deregulation. The next eighteen months will reveal whether the gigafactory investment and the regulatory revision are complementary or contradictory.