Spain wants to speed up the adoption of artificial intelligence in public administration, even as large-scale rollout remains one of the public sector’s main challenges. The report Data and AI in the Public Sector 2026 from Capgemini finds that only 28% of the public organizations surveyed have successfully scaled their AI initiatives, a figure that contrasts with the ambition of the Spanish government, which is pursuing new infrastructure, training, security, and administrative services built on this technology.
Spain’s public-sector AI plan in 30 seconds
- Capgemini puts at 28% the share of public organizations that have successfully scaled their AI projects.
- Spain is planning an AI gigafactory with €5 billion in public-private investment.
- The official goal is to have 100,000 AI accelerators operational by 2028/2029.
- The IA360 plan includes a €600 million business voucher and a network to bring AI to small and medium-sized businesses.
- The administration expects AI assistants in the Citizen Folder and a new WhatsApp 060 channel in 2027.
The gap between experimenting with artificial intelligence and using it on a routine basis stands out as one of the biggest challenges for public administrations. Capgemini surveyed 600 senior public-sector organizations and concludes that adoption is moving forward, but the ability to turn pilot projects into operational services remains limited.
The 28% figure is global and doesn’t specifically refer to Spanish public administration. So it doesn’t support the claim that only that share of Spanish public bodies has scaled AI. It does, however, help gauge the difficulty the public sector faces when trying to move from isolated tests to systems used on a stable basis.
Spain is in fact putting forward a strategy to get past that barrier. The government’s IA360 plan, presented as a program to move the country into the group of the world’s five AI leaders, lays out actions for 2026 and 2027 covering infrastructure, security, education, talent, technology, business, public administration, and international partnerships.
A €5 billion gigafactory to boost compute capacity
One of the largest projects is an artificial intelligence gigafactory with €5 billion in public-private investment, of which €719 million would be public money. The goal is to build a European-scale compute infrastructure for training and deploying advanced models and applications.
The document sets a target of 100,000 AI accelerators operational by 2028/2029. Construction is slated to begin in February 2027, with the facility opening in 2028.
The project is detailed on pages 4 and 5 of the IA360 document, alongside a proposed data center regulation that aims to account for energy consumption, water use, land, resilience, and digital sovereignty. The government plans to approve that framework in October 2026.
The scale of this investment shows that Spain’s strategy isn’t limited to rolling out AI tools in public bodies. It also aims to boost the compute capacity available in the country and attract projects that need large amounts of processing power.
Capgemini specifically identifies data and infrastructure as two of the conditions administrations need to scale their projects. Legacy systems, fragmented data, and a lack of interoperability continue to make that leap difficult.
From pilots to an administration with AI assistants
Spain’s plan also lays out concrete changes in how citizens interact with public administration. One project calls for adding an artificial intelligence assistant to the Citizen Folder (Carpeta Ciudadana) to guide people through their paperwork.
The assistant is expected to roll out during 2027. The government also plans to launch a new WhatsApp 060 channel in the first quarter of that year for inquiries and certain procedures, subject to the planned security and data protection safeguards.
The goal isn’t to replace public oversight with automated systems. The document itself states that AI use in public administration must remain under human supervision, public control, and the protection of citizens’ rights.
There’s also a plan to use artificial intelligence to cut red tape for freelancers and small and medium-sized businesses. The plan includes a so-called AI Marathon for simplification, set for the first quarter of 2027, which will look for documents that administrations repeatedly request, outdated or duplicate procedures, and unjustified differences between regions.
The proposal aims to use data and supercomputing capabilities to produce verifiable changes in administrative procedures, rather than simply rolling out a new piece of technology.
Data, talent, and security remain the bottleneck
Capgemini’s findings line up with several of the problems the Spanish plan itself acknowledges. Public administrations see data readiness as a prerequisite for rolling out AI, but legacy systems, fragmented architectures, and insufficient interoperability make it harder to access quality information.
Data management thus turns out to be just as important as choosing the AI model. An administration can have access to advanced models and still struggle to use them if the information it needs is spread across systems that don’t communicate properly with each other.
Talent is another problem flagged in IA360. The document ranks Spain 23rd worldwide on the talent pillar and 25th out of 32 countries in concentration of AI professionals, a gap also visible in Spain’s AI Atlas, the government’s own tracker of AI adoption, talent, and infrastructure. The government plans regulatory changes to attract and retain researchers and make it easier to transfer knowledge to businesses.
Security is another of the plan’s eight areas. Planned measures include creating an Artificial Intelligence Security Institute within the Spanish Agency for the Supervision of Artificial Intelligence (AESIA), involving bodies such as the National Cybersecurity Institute (INCIBE), the National Cryptologic Center, the Spanish Data Protection Agency, and the Barcelona Supercomputing Center.
There’s also a plan for an advanced cybersecurity program with an initial budget envelope of around €300 million and measures tied to the transition toward post-quantum cryptography.
The business challenge: getting AI out of the lab too
Capgemini’s findings aren’t limited to the public sector. Its analysis points to the most advanced organizations being those that combine strategy, data governance, talent, and technology capabilities to turn projects into real operations.
The Spanish government identifies a similar difficulty among businesses. According to data gathered in IA360, 21.1% of Spanish companies with 10 or more employees were using artificial intelligence in the first quarter of 2025, compared with 44.2% of the working-age population using generative AI, according to the indicator cited in the document.
To close that gap, the plan proposes the NEURONA Network, a network of demonstration and deployment centers that would let companies test AI solutions with their own data. The first pilot is expected to reach 100 small and medium-sized businesses across two autonomous regions before the end of 2026, with plans to extend the network to 10 regions and 5,000 companies during 2027.
On top of that, there’s an artificial intelligence business voucher with a planned endowment of €600 million. The design wouldn’t simply fund software licenses, but services that bring AI into production processes, along with training, prior diagnostics, and later impact measurement.
The stated goal is to transform 25,000 companies and multiply private investment fivefold by 2030 compared with 2024 levels.
The gap between these goals and the reality Capgemini describes marks the main challenge. Putting money into infrastructure, models, and projects can speed up adoption, but the real leap comes when systems stop working as demos and become part of everyday processes.
For Spain’s public administration, that means the gigafactory, the citizen assistants, the security programs, and the training initiatives will eventually need to connect to concrete services and procedures. The 28% of public organizations that, according to Capgemini, have already managed to scale their initiatives shows that this shift still hasn’t been resolved, even among the most advanced administrations.
Frequently asked questions
What share of public organizations has managed to scale AI?
According to Capgemini, 28% of the public organizations surveyed say they have successfully scaled their artificial intelligence initiatives. It’s a global figure, not one specific to Spain.
How much will Spain invest in its artificial intelligence gigafactory?
The IA360 plan calls for €5 billion in public-private investment, of which €719 million would be public money. The goal is to reach 100,000 AI accelerators operational by 2028/2029.
When will AI reach the Citizen Folder?
The government plans to roll out the Citizen Folder’s artificial intelligence assistant during 2027. It also plans a WhatsApp 060 channel in the first quarter of that year.
What is the artificial intelligence business voucher?
It’s a planned €600 million aid program to fund AI services and solutions that transform business processes. The design includes training, a maturity diagnosis, and impact measurement, and isn’t meant to cover software licenses alone.
