Spain's IA360 AI voucher: what it will pay for and how to prepare your project

What the plan says

The IA360 Plan, presented by the Spanish Government on 21 September, includes a business AI voucher among its projects. The official document describes it as follows:

• Budget: €600 million.
• Goal: 25,000 companies transformed by 2030.
• What it will fund: AI services and solutions provided by European technology companies that build on foundation models and add their own value through "development, integration, sector knowledge, data processing or process redesign".
• What it won't: "The voucher will not fund mere licence subscriptions".
• How: milestone-based payments, conditions on staff AI training and on embedding AI in internal processes, with a prior maturity diagnosis, a specific use case and subsequent measurement of the productivity impact.
• Timeline: design, regulatory preparation and a controlled pilot in the first half of 2027; general call before the end of 2027.

The voucher is linked to the Neurona Network, another measure in the plan: centres where small businesses can try AI solutions for their sector with their own data, starting with a pilot of 100 SMEs in two regions before the end of 2026.

The plan aims to raise the share of companies using AI from 21.1% to 55% by 2030. Spain's statistics institute has since updated that figure: according to its survey published on 22 October, 33.6% of companies with ten or more employees used AI in the first quarter of 2026.

What it funds and what it doesn't

The plan itself draws the line: the voucher pays for work that changes how a company operates, not standalone tools. Autónomos y Emprendedor sums it up with examples:

For a service business, that means an agent that reads incoming documents, sorts them and records them in your management software fits; giving the whole team an AI assistant account doesn't.

What we still don't know

As Economist & Jurist and Autónomos y Emprendedor point out, the announcement doesn't entitle anyone to anything. Still unknown:

• The amount per company and whether there will be bands by size.
• The requirements for beneficiaries and providers.
• The body that will manage it and how to apply.
• The State aid regime it will use, which may limit how much each company receives if it has already had other aid.

Until the rules are published, any specific figure per company is a guess.

How to prepare without waiting for the call

The plan already says what will be required: a diagnosis, a use case and measurement. You can prepare that now, and it's useful even if you never apply:

1. A diagnosis of how you work today. Which processes you have, which software you use, where your data lives and which tasks are repeated by hand.
2. One specific use case. Just one, with volume and a clear result. If in doubt, see which processes to automate first.
3. Baseline figures. Time per task, cases per month, errors and cost. Without them you can't show the productivity impact the plan asks for. The ROI calculator helps you estimate it.
4. Milestones. Split the project into phases you can verify: integration, supervised pilot, go-live and measurement. We explain it in how to implement a digital employee.
5. Team training. Who will supervise the system and what they need to know. The plan makes it a condition, and the AI Act already requires companies to take AI literacy measures for staff who use these systems.

Beware of anyone offering to process it now

There's nothing to apply for today. Be wary of anyone offering to "reserve" the voucher, asking for an upfront fee to process it or guaranteeing an amount. The sensible move now is to prepare the project; the application will come with the call.

Nor does it make sense to put a project that already pays for itself on hold while waiting for aid that, at the earliest, will arrive with a pilot in 2027. If the use case has a return, it's worth doing with or without the voucher.

How we do it at Amud

What the voucher asks for (a diagnosis, a use case, integration with your software and measurement) is how we always work. We start by measuring a process, build the AI agent or automation that solves it, with human oversight wherever there are consequences, and measure the result. If your project fits once the rules are published, the groundwork will already be done.