Spain’s AI Adoption Gap Starts With The Employer

Spain’s government says its people have taken up AI faster than its companies have.

The government’s new Plan IA360 sets two numbers side by side. According to a Microsoft report for the first quarter of 2026, 44.2% of Spain’s working-age population uses generative AI. According to the national statistics office, only 21.1% of companies with ten or more employees used AI in the first quarter of 2025.

The two figures come from different sources, different years and different units, one counting people and one counting firms. That is precisely why the pairing is useful. The AI adoption gap it describes does not run between Spain and its rivals. It runs between the workforce and the organisations that employ it.

What Spain announced on 21 September

Pedro Sánchez presented the plan at the Moncloa as a twelve-month roadmap across eight areas. Closing the event, he told the employers and union officials in the room that self-regulation does not work, as TNW reported.

Three measures carry the people side of the plan.

First comes a Tripartite Agreement. The government will convene unions and employers in October 2026. Sectoral tables will cover education, the labour market, the financial system and culture, among others.

The second is a proposed permanent observatory on the labour market. Its stated job is to analyse how AI is changing employment and to guide decisions on jobs, training and regulatory adaptation. The OECD AI Principles already ask governments to build human capacity and prepare for labour market transformation. An observatory is a concrete way of answering that request.

Finally, there is a business AI voucher with a budget of €600 million. Its conditions say more about adoption than the headline figures do.

The voucher separates a licence from AI adoption

The voucher will not fund a licence subscription. Instead, it pays for AI services from European technology firms that add their own value on top of foundation models. That value can come through development, integration, sector knowledge, data work or process redesign.

Its design would also include milestone payments and conditions on training workers in AI and on bringing AI into internal processes. The rollout starts with a maturity diagnosis and a concrete use case. It ends with a measurement of the productivity impact. A controlled pilot is planned for the first half of 2027, with a general call before the end of that year.

Read as a definition of what closing the AI adoption gap means, that is a useful one. A licence buys access. Adoption means changed processes, trained people and a measured result, and the voucher pays only for the second.

Where the AI adoption gap shows up at work

Translated into a single organisation, the AI adoption gap would look like this: staff using AI tools on their own initiative while the organisation is still deciding what it has approved.

In that situation the use does not wait for the decision. It simply happens outside the policy, which is why a policy that treats everyone the same misses where the exposure actually sits. Closing the AI adoption gap from the employer’s side therefore starts with an honest picture of current use, not with another tool rollout.

The plan makes a related point about skills. It notes that use is not the same as competence, and it proposes a free AI literacy programme for all adults, to deploy in 2027. Inside an organisation, the AI adoption gap is therefore also a training gap: use has run ahead of instruction.

Why unions now sit at the adoption table

The more distinctive choice is who Spain wants in the room. Spain is opening social dialogue at the start of its adoption push. In effect, the government treats the AI adoption gap as a subject for negotiation between employers and workers, not as a procurement problem.

For organisations operating in Spain, workforce consultation may therefore arrive inside the adoption plan rather than after it. That is a change in sequence, and sequence matters. An employer that has already rolled out a tool and then meets its works council is defending a decision. An employer that brings the council in at the use-case stage is shaping one.

None of this is binding yet. The plan is a set of proposals with milestones, and Spain’s national AI bill is not expected to pass before the end of 2027, according to TNW. Until then, binding rules come from EU law.

What the AI adoption gap asks of the AI lead

The Spanish plan is national, but the AI adoption gap it names is a question any employer can ask of itself. Any EU-operating organisation can run the same comparison internally, and the result is a governance question as much as a technology one.

Three questions follow from the voucher’s own logic. What share of your staff already uses AI at work, and how much of that use has the organisation actually sanctioned? Which processes have changed because of AI, as distinct from which licences have been bought? And what was measured before and after, so that any claimed gain survives scrutiny?

That last question matters more than it looks. Productivity claims made without a baseline are hard to test after the fact, and evidencing AI gains comes down to what was recorded at the start.

Spain has decided that closing the AI adoption gap is a social project as well as an economic one. Employers elsewhere do not have to agree with that framing to borrow its method. Train the people, change the process, measure the result and involve the workforce early.

Future Prep follows how these national plans turn into obligations and expectations for EU-operating organisations. The FP website carries the analysis as it develops.