On 30 July, the European Union launched what it described as its “latest major push” to accelerate Europe’s technological sovereignty and realise its ambition of becoming the “AI Continent.”

The plan is to establish up to seven AI Gigafactories. Supported by as much as €10 billion in EU and national funding, the initiative is expected to attract at least another €20 billion from private investors.

Applications close on 12 November 2026. The proposals will then undergo a competitive evaluation before the successful projects are selected.

Thirty billion euros would be a major intervention in almost any policy field. In artificial intelligence, it must be considered against a level of corporate investment that has changed the meaning of scale.

OpenAI and its partners intend to invest $500 billion in Stargate infrastructure over four years. OpenAI Amazon expects approximately $200 billion in capital expenditure during 2026. Alphabet plans between $175 billion and $185 billion, while Meta’s latest range is between $130 billion and $145 billion.

These numbers are not directly comparable. Stargate is a multi-year infrastructure programme. Europe’s figure combines public support with anticipated private investment. The company figures cover annual capital expenditure across businesses extending beyond AI.

Even with those qualifications, the distance is revealing. Amazon, Alphabet and Meta together expect to spend between $505 billion and $530 billion in one year. Europe is seeking to mobilise €30 billion across several countries and as many as seven projects.

The weakness is not the size of the European number in isolation. It is the industrial position from which that money must work.

A Factory Does Not Create an Ecosystem

The planned Gigafactories will combine what the Commission describes as “advanced AI processors, software and cloud technology stacks, high-speed connectivity and energy-efficient data centres.”

They are intended to give European companies, researchers and public institutions the computing power required for the “training, inference and fine-tuning” of advanced AI models.

Europe needs that capacity. An organisation without reliable access to compute cannot develop a competitive frontier model, however strong its researchers may be.

Compute is nevertheless only one part of the AI economy.

The largest American technology companies connect infrastructure directly to models and products. They can place a new capability inside a cloud platform, search engine or workplace application and make it available to millions of existing customers.

This produces a reinforcing commercial cycle. Revenue from established products finances infrastructure. That infrastructure supports more capable models. The models improve the products and create new sources of revenue.

Europe’s Gigafactories intervene at the infrastructure stage. Whether they produce a similar cycle will depend on what happens above them.

A European company still needs access to talent and patient capital. It must turn a model into a product that solves a problem customers will pay to address. It then needs sufficient distribution to grow without relocating or being acquired before reaching scale.

A building full of processors cannot complete those steps for it.

Can Public Funding Become a Catalyst?

The Commission explains the logic behind the initiative in explicitly catalytic terms:

“Public funding acts as a powerful catalyst for the initiative, driving the deep private investment required to construct and operate these facilities.”

This is an important distinction. The EU is not claiming that €10 billion alone will finance the entire programme. Public institutions will procure computing access from the selected Gigafactories, providing consortia with guaranteed demand and reducing some of the commercial risk.

But the quality of a catalyst is measured by what it activates.

If public procurement encourages sustained private investment, attracts model developers and helps European companies reach commercial scale, the intervention could have an impact considerably larger than its initial budget.

If it primarily finances infrastructure that depends on continued public purchasing, the multiplier will be weaker. Europe could end up with well-funded facilities searching for enough competitive companies to use them.

The €20 billion in anticipated private investment is therefore not merely an additional number. It is an early test of whether European public policy can stimulate a durable AI market.

Sovereignty With American Hardware

The Commission says the Gigafactories will strengthen Europe’s “technological leadership, resilience and strategic autonomy.” Operating major computing facilities within Europe would improve resilience and give European organisations greater control over where sensitive workloads are processed.

Its ambition is stated clearly:

“The initiative will ensure that Europe can develop advanced AI on its own infrastructure, in line with EU rules and values.”

The programme also exposes the present limits of that autonomy.

AMD, Nvidia and Qualcomm have signed letters of intent intended to ensure that European consortia have “seamless access to necessary hardware.” All three companies are American.

European operators may therefore control the facilities and determine who can use them while remaining dependent on processor designs and software environments developed elsewhere. Manufacturing represents another dependency, because many of the most advanced chips are produced outside both Europe and the United States.

European control over the location and operation of a data centre has real value. It can improve access, security and resilience. It does not provide control over every technology required to keep that facility competitive.

A stronger form of sovereignty would require European capability across more of the system. That includes processor design and manufacturing, cloud platforms, foundation models and the products through which AI reaches customers.

The Gigafactories could help stimulate those capabilities by creating reliable demand. They cannot be treated as evidence that Europe’s dependencies have already been resolved.

The Concentration Question

Supporting up to seven projects gives more Member States and regional ecosystems the opportunity to participate. It could widen access to compute and prevent European AI capacity from becoming concentrated in a single country.

The design also creates a trade-off.

Frontier AI benefits from concentration. Large computing clusters require substantial and dependable electricity supplies. They need specialised engineers nearby and close operational relationships between infrastructure providers, model developers and commercial users.

Spreading resources too thinly could leave Europe with several capable facilities that lack the scale or surrounding expertise of the strongest American clusters.

Concentration carries its own risks. Placing most European capacity in one location would increase regional inequality and create a larger operational point of failure.

The relevant decision is therefore not simply whether Europe should build one facility or seven. It is whether the selected projects can operate as a connected European system while preserving sufficient scale at each location to remain technically and economically useful.

A Different Model of AI Development

The United States has allowed a small number of companies to accumulate extraordinary financial and technological power. Those companies can invest rapidly because they already possess global products, large customer bases and substantial cash flows.

That model produces speed. It also concentrates infrastructure and decision-making inside a small group of private organisations.

Europe is pursuing a more coordinated approach. Public support is intended to widen access and ensure that smaller companies and research institutions can use infrastructure they could never finance independently.

The Commission also promises that AI developed in the Gigafactories will follow European standards on “data protection, safety, security and ethics.”

There is no reason for Europe to abandon those commitments in an attempt to imitate the United States.

But a different model must still produce results. Democratic oversight does not reduce the electricity required to operate a computing cluster. European values do not remove the need for specialised talent, competitive products or paying customers.

Regulation can establish the conditions under which an industry develops. Infrastructure can give that industry essential capacity. Neither guarantees that competitive companies will emerge.

What Gets Built Above the Factories

Europe’s AI Gigafactories could strengthen research and give promising companies access to resources that are currently difficult to obtain. They could also help those companies remain in Europe during the costly transition from experimentation to commercial scale.

That would make the programme valuable even if its expenditure never approaches the American figures.

Its success should be judged by what the infrastructure produces.

By 2030, Europe should be able to identify the advanced models trained in these facilities. It should be able to name companies that remained and expanded because European computing capacity was available. Those companies should have products, customers and revenue rather than another sequence of subsidised pilot projects.

Europe should also be able to demonstrate which technological dependencies have been reduced. Operating seven facilities with imported hardware and lightly used capacity would not meet that test.

The applications will close on 12 November. The decisions that follow will determine whether Europe is financing seven large computing facilities—or laying the foundations of an AI industry capable of using them.