• H2 2025 – Launch: Advanced computing / AI Factory (EuroHPC); Gigafactory; National Data Center Plan; AMALIA; AI for SMEs (IFIC).
• H1 2026 – Implementation: Sectoral data spaces; international network collaboration; AI Center of Excellence for the public sector; National AI competitions for the public sector; Practical Guide to Public Procurement of Innovation; Accelerated AI training plan for the public sector; National Smart Skills Framework; Implementation of the EU AI Act; Implementation guides and risk- assessment tools.
• H2 2026 – Scale and replication: Support for fundamental research; Sectoral AI Centers; “PME.IA” platform; regulatory sandboxes; National AI Week; international cooperation.
• 2027+ – Consolidation: Incentives for AI research in companies (Deeptech funds); responsible AI startup acceleration program.

Execution is contingent upon available funding, with priority given to European funds.

The official timeline does not clarify several critical dependencies:

• When should compliance preparation begin in order to align with the opportunity window?
• What criteria will determine preferential access to shared infrastructure and sectoral centers?
• What level of organizational maturity is required to capture value from public programs?
• How should organizations map their specific eligibility for the 32 initiatives based on their individual profiles?
• Waiting for the detailed regulation to be published means arriving late for the first allocation of resources.

Waiting for the publication of detailed regulations may result in missing the initial allocation of resources.

Barriers to Implementation: OECD Lessons from the European Context

Reports from the Organisation for Economic Co- operation and Development and the European Union identify structural barriers that prevent initiatives from moving from pilot to scale.

The Digital Europe Programme acknowledges that effective execution in the public sector and among SMEs requires overcoming the lack of advanced digital skills, limited access to high- quality data and infrastructure, and the need for practical guidance through structures such as the European Digital Innovation Hubs.

The OECD report Governing with AI (2024) identifies the following critical challenges in implementing AI in the public sector:

1. Skills gaps: shortage of technical talent and organizational change-management
2. Data quality and access: fragmented, non-interoperable data coupled with governance and quality issues.
3. Limited practical guidance: absence of clear standards, guidelines, and replicable use cases.
4. Risk aversion: reluctance to assume legal responsibility, leading to reputational risk and regulatory
5. Insufficient outcome measurement: difficulty in demonstrating value and justifying investment.

These barriers are explicitly identified within the ANIA: the fragmentation of projects with limited scalability; risk aversion coupled with a low level of strategic technology literacy; dispersed technological infrastructure lacking interoperability and data-sharing standards; and a complex, difficult-to-navigate framework for the public procurement of innovation.

Regulatory Framework and Governance

The EU AI Act establishes requirements for high- risk systems, mandating continuous risk management, data governance, technical documentation, effective human oversight, and technical robustness. The OECD recommends governance cycles that include risk classification, internal controls, testing, monitoring, and auditing.

Responsibility and Ethics: Compliance with the EU AI Act

The ANIA positions trust as a prerequisite for execution, in a context in which 54% of business leaders cite ethical implications as the primary barrier, and only 40% of CEOs believe their organizations have the capability to implement AI in an ethical and secure manner.

Most companies are not yet prepared to comply with the EU AI Act (e.g., technical documentation, robustness testing, continuous monitoring, and risk management).

Planned Compliance Initiatives

• IV.4 – EU AI Act Implementation: definition of competent authorities, coordination models, and the sanctions framework (H1 2026).
• IV.5 – Regulatory Sandboxes: testing AI solutions in controlled environments (H2 2026).
• IV.6 – AI Act Implementation Guide: standards and risk-assessment tools, including Portuguese Standard 4595 and the creation of compliance support infrastructure for high-risk AI (H1 2026).

Compliance costs disproportionately affect SMEs and startups due to the difficulty of sustaining certification processes, audits, and advanced testing.

Practical Implications: Three Critical Dimensions of Preparedness

Execution of the ANIA requires strategic preparation across three interdependent dimensions.

1. Regulatory compliance: three key legal risks.

There is a critical timing gap between internal preparedness and access to public opportunities that is not explicitly reflected in the official timeline.

Key risks include:

• Effective human oversight: requires documented protocols and genuine intervention capability, not merely monitoring;
• Algorithmic bias: decisions affecting individuals (human resources, credit, services) without adequate auditing expose organizations to discrimination-related litigation;
• Innovation-oriented public procurement: new contractual arrangements introduce complexity in the allocation of intellectual property rights and liabilities.

2. Organizational capacity (70% people).

Under the 10–20–70 rule, organizations must first build AI literacy before they can effectively benefit from public training programs.

The scheduled programs (H1 2026) expand supply, but do not address the challenge of internal absorption.

The level of organizational maturity ultimately determines an organization’s real ability to access initiatives under the ANIA.

3. Access to scarce resources: infrastructure, financing, and talent.

Resource allocation will be governed by eligibility criteria that have not yet been fully clarified. Waiting for the publication of detailed regulations means arriving too late for the first allocation window.

Final Notes

Capturing value from the ANIA requires early action in response to three predictable scarcities: access to financing contingent on regulatory compliance, shared computational capacity, and execution-ready talent.

Critical areas for organizational preparation include

• Mapping specific eligibility across the 32 initiatives and identifying relevant access windows;
• Anticipating compliance requirements (EU AI Act, GDPR, Standard 4595) and establishing risk- management and data-governance frameworks;
• Assessing organizational maturity and developing a capability-building plan aligned with the critical timeline;
• Preparing for participation in public tenders, regulatory test environments, and access to Sectoral Centers;
• Securing European financing and managing the complexities of innovation-oriented public procurement.

You can download the full document here.

For more information, please contact:

Nádia da Costa Ribeiro | Partner
nadia.ribeiro@pt.andersen.com

João Sousa Guimarães | Partner
joao.guimaraes@pt.andersen.com