The Government has approved a public investment of €719 million to boost Spain’s bid for a future European AI gigafactory. The operation will be channeled through the Spanish Society for Technological Transformation (SETT), via the Next Tech facility and with funds from the Recovery, Transformation, and Resilience Plan. The goal is for Spain to present a solid proposal when the European Commission is expected to launch the call under InvestAI.
The Spanish bid will be multisite, including Móra la Nova in Tarragona and San Fernando de Henares in Madrid. This is not yet an official European award or a confirmed project, but a strategic positioning move: the state is entering the public-private consortium aiming to develop one of Europe’s major advanced AI computing infrastructures.
This decision comes at a time when AI no longer depends solely on models, talent, and data. The new bottleneck is physical: GPUs, energy, data centers, networks, cooling, and operational capacity. Without this infrastructure, Europe can regulate, research, and apply AI, but will still rely on U.S. hyperscalers or closed Chinese capabilities for training and deploying the most demanding models.
What is an AI Gigafactory?
An AI gigafactory isn’t just a conventional data center with more servers. It’s a facility designed to house hundreds of thousands of GPUs or specialized accelerators, capable of training, fine-tuning, and running large-scale language models, multimodal systems, scientific models, advanced computer vision, and industrial applications that require massive computing power.
The difference lies in scale and purpose. Traditional cloud data centers host diverse workloads: storage, databases, business applications, web services, virtualization, or backups. An AI gigafactory is built around dense clusters of accelerators, ultra-low latency internal networks, advanced cooling, high-capacity power supplies, and software capable of distributing training and inference at large scale.
| Infrastructure | Main Use | Distinctive Trait |
|---|---|---|
| Conventional cloud data center | Applications, storage, digital services, and enterprise workloads | Versatility and general capacity |
| HPC supercomputer | Science, simulation, research, and intensive computing | High performance for scientific workloads |
| AI Factory | Access to supercomputing and services for startups, SMEs, and research | Technical support and AI capacity access |
| AI Gigafactory | Training and inference of advanced models at massive scale | Hundreds of thousands of GPUs and industrial focus |
The European Commission intends to fund up to five AI gigafactories as part of its plan to make Europe more competitive in artificial intelligence. This initiative is part of InvestAI, which aims to mobilize €200 billion for AI across the European Union and €20 billion specifically for gigafactories.
Móra la Nova and San Fernando de Henares: A Multisite Candidate
The Spanish proposal features two locations. Móra la Nova, in Tarragona, offers significant regional and energy advantages within a developmental industrial environment. San Fernando de Henares, near Madrid, provides proximity to the main connectivity hub, companies, cloud infrastructure, government, and tech talent.
A multisite approach can make sense if infrastructure is distributed according to different regional strengths: energy, land, connectivity, talent, business demand, and proximity to decision centers. However, it also adds complexity. An AI gigafactory isn’t just racks and servers; it requires electrical coordination, fiber networks, physical security, permitting, equipment supply, technical operations, and access for both public and private clients.
The government notes that the formation of the public-private consortium is being finalized. This point is crucial. A project of this scale cannot rely solely on public funding. It needs industrial partners, technology providers, data center operators, energy suppliers, user companies, universities, research centers, and private financial capacity.
The participation of SETT aims to provide strategic guidance and public coordination capacity. It also allows the state to hold an equity stake in the entity submitting the Spanish bid. In such a sensitive area as AI infrastructure, this public presence is not just financial—it also concerns sovereignty, access control, and alignment with national and European priorities.
Technological Sovereignty, But With Conditions
The argument for technological sovereignty is strong. Europe aims to avoid complete dependence on external infrastructures for training its models, processing sensitive data, or developing critical applications in health, industry, energy, defense, public administration, or science. Computing capacity has become a strategic resource, comparable in importance to energy, semiconductors, or telecommunications networks.
Spain is already part of the European supercomputing map with MareNostrum 5 at the Barcelona Supercomputing Center and capabilities like CESGA in Galicia. Additionally, EuroHPC has supported AI Factories across various European countries, including projects linked to Spain, to provide access for startups, SMEs, research, and industry. The gigafactory would represent a significant step up from these existing AI facilities.
But sovereignty isn’t just about capacity. It requires decisions about who can access it, at what cost, under what conditions, with what security guarantees, which models are prioritized, and how to prevent infrastructure from being captured by a few large clients. To strengthen Europe’s AI ecosystem, startups, universities, SMEs, and research centers need genuine access pathways, not just institutional references.
The relationship with large tech providers must also be managed. Europe can build gigafactories, but will still depend on global supply chains for GPUs, networks, software, cooling systems, and components. Autonomy remains relative if infrastructure relies on chips, licenses, or services controlled outside Europe.
Energy, Water, and Network: The Project’s Challenging Aspects
AI infrastructure consumes substantial energy and requires meticulous electrical planning. A gigafactory isn’t placed purely based on land availability. It needs firm power supply, grid connection, energy agreements, stability, cooling, and scalability. In Spain, where discussions around grid saturation and rising electrical demands are ongoing, this will be a central issue.
Sustainability can’t be just a label. If the project aims to be European and aligned with energy transition goals, it must demonstrate renewable energy sourcing, operational efficiency, responsible water management, heat reuse potential, territorial integration, and transparency regarding its environmental impacts.
The government highlights the sustainability focus of the bid, but the details will be decisive. Metrics such as PUE, WUE, liquid cooling, rack density, thermal recovery, and electrical location have tangible impacts on costs and social acceptance for AI data centers.
| Challenge | Key Question |
| Energy | Is there sufficient, guaranteed power to grow? |
| Water | What cooling methods will be used, and what will be the actual consumption? |
| Electric Grid | What reinforcements will be needed, and in what timeframe? |
| Connectivity | How will it connect to Madrid, Europe, and cloud clients? |
| Access | Who will be able to use the capacity, and under what conditions? |
| Governance | What roles will the state, private partners, and Europe play? |
Addressing these points could determine the difference between a transformative project and an overly ambitious promise. AI deployment isn’t abstract—it occurs in specific municipalities, with concrete networks, electricity consumption, and economic effects.
An Opportunity for Industry, Science, and Business
If Spain secures a gigafactory, its impact will extend beyond the tech sector. Large-scale computing infrastructures can support industries needing custom models, simulation, digital twins, process optimization, material design, biomedicine, robotics, energy, logistics, or large-scale data analysis.
For startups and SMEs, access to computing can be a critical barrier. Training or fine-tuning advanced models is costly, and reliance on external providers can limit margins, privacy, availability, or regulatory compliance. A European infrastructure with well-designed access mechanisms could lower these barriers.
Universities and research centers could leverage the gigafactory for projects that currently depend on grants, hyperscaler partnerships, or limited resources. For public administration, it could open pathways to sovereign models for public services, healthcare, justice, education, security, or document management—always under legal and ethical safeguards.
The risk is that the infrastructure becomes just a symbol without widespread actual use. To prevent this, the project should include access programs, training, technical support, quality data, cloud-native tools, and a user experience simpler than traditional supercomputing. Business AI deployment also happens outside HPC, in Kubernetes, APIs, objects, pipelines, notebooks, MLOps, and continuous deployment.
Spain Is Playing a Candidacy, Not a Victory
The €719 million announcement is significant, but it must be interpreted precisely. Spain hasn’t yet secured a European gigafactory. The government has approved this investment to participate in the consortium applying when the European call is launched. The final decision will depend on the European process, competition from other proposals, and the Spanish project’s ability to demonstrate technical, financial, energy, and territorial viability.
The European Commission has received dozens of expressions of interest from various member states for AI gigafactories. Competition will be fierce, as all countries see these infrastructures as means to attract investment, talent, and companies. In this context, Spain must show it offers not just funds but an actionable plan.
Spain has arguments: renewable energy, supercomputing capacity, a strong tech ecosystem, connectivity, strategic location, research centers, and a public strategy to reinforce AI and semiconductors. Challenges include administrative slowness, network saturation points, talent shortages, and EU-wide competition for providers and clients.
This effort marks a shift in discourse. Spain no longer only discusses AI regulation, adoption, or training. It aims to get involved in heavy infrastructure—key for training cutting-edge models, controlling capacity, and offering alternatives to the global giants.
While an AI gigafactory alone won’t eliminate Europe’s technological dependence, without such infrastructure, reducing that dependence will be much harder. The current question is whether Spain can turn this announced investment into a winning bid and, subsequently, into a functional, efficient, and accessible facility for the ecosystem.
Frequently Asked Questions
Does Spain already have an awarded European AI gigafactory?
No. The government has approved €719 million to participate in the consortium that will submit Spain’s candidacy when the European call is issued.
Where would the gigafactory be located?
The Spanish bid will be multisite, including Móra la Nova in Tarragona and San Fernando de Henares in Madrid.
What is the purpose of an AI gigafactory?
It serves to train, test, and deploy large-scale advanced AI models, with hundreds of thousands of GPUs or specialized accelerators.
Why is this important for Europe?
Because it lessens dependence on infrastructures controlled by U.S. hyperscalers or closed Chinese ecosystems, and can facilitate access to advanced AI for businesses, research, and public agencies.
via: planderecuperacion
