Summary As proposed, the Cloud and AI Development Act (CADA) encourages the procurement of EU-developed AI models by mandating that contracting authorities include specific non-price award criteria in public tenders for innovative cloud and AI services. Under Article 32, buyers must evaluate how well a tenderer integrates technologies developed in the Union, including results from EU-funded research and development programmes. This mechanism aims to strengthen the European digital supply chain and reduce dependence on third-country providers without making EU origin the sole decisive factor for contract awards.

Detail

The Cloud and AI Development Act (CADA), proposed by the European Commission on 3 June 2026 (COM(2026) 502 final), introduces a structured framework to boost the competitiveness of the European cloud and AI ecosystem. A central pillar of this strategy is leveraging public procurement to drive demand for European technologies. Rather than imposing mandatory quotas or excluding non-EU providers outright, CADA uses a "soft power" approach through standardized award criteria that favor Union-based innovation.

The Legal Basis: Article 32

The primary mechanism for encouraging the purchase of EU-developed AI models is found in Article 32 of the CADA proposal, titled "Union added value." This article mandates that contracting authorities include specific non-price criteria when procuring innovative cloud computing services and AI systems.

Article 32(1) states:

"In public procurement procedures for innovative cloud computing services and AI systems, contracting authorities shall include, as part of the quality evaluation of the tender, non-price award criteria that allow them to evaluate the tenderer's contribution to the development of a European cloud and AI ecosystem."

This provision transforms the procurement process from a purely price-driven exercise into a strategic tool for industrial policy. It requires public buyers to look beyond the immediate cost of the service and assess the long-term strategic value of the supplier's offering.

Evaluating EU Integration and R&D

To operationalize this requirement, Article 32(2) sets out general principles for these non-price criteria: they must be linked to the subject matter of the contract, not confer unrestricted freedom of choice on the buyer, be expressly set out in procurement documents, and be ancillary (not decisive) to the award.

The specific factors that buyers must consider are detailed in Article 32(3). This paragraph lists four key areas for evaluation, with Article 32(3)(b) being particularly relevant for AI model developers:

"(b) the tenderer has integrated technologies developed in the Union, including research and development results stemming from Union funded research and development programmes and makes use of tools, such as standards, specification, software, models or other technology developed in the Union;"

This clause explicitly rewards suppliers who build their AI solutions using foundational technologies, software models, or standards that originated within the EU. It creates a direct incentive for cloud providers and AI vendors to integrate outputs from EU-funded initiatives, such as those under Horizon Europe or the Digital Europe Programme, into their commercial offerings. By doing so, vendors can score higher on the "Union added value" metric.

The text specifically highlights "models" as a category of technology that can be integrated. This means that an AI system built upon a foundational model trained or developed within the Union, or one that incorporates R&D results from EU programmes, would satisfy this criterion.

Other Evaluation Criteria

While AI model origin is a key factor, Article 32(3) provides a broader context for "European added value":

  • Article 32(3)(a): Evaluates the tenderer's contribution to strengthening the digital technology supply chain in the Union, including the use of software or hardware designed or manufactured in the Union.
  • Article 32(3)(c): Assesses how the innovation required to deliver the service contributes to strengthening the security of supply and the development of a European cloud and AI ecosystem.
  • Article 32(3)(d): Considers the extent to which the service is delivered using critical computing, storage, and networking hardware components designed and/or manufactured in the Union. If this is not feasible, it allows for hardware from third countries that still contributes to strengthening supply security and the European ecosystem.

Weighting and Proportionality

CADA is careful to ensure that these criteria do not distort competition unfairly or violate international trade commitments. Article 32(2)(d) explicitly states that non-price award criteria must be "ancillary and not decisive in the award of the contract."

Furthermore, the explanatory memorandum and recitals suggest that while these criteria are mandatory, their weight should be proportionate. The proposal indicates that contracting authorities could consider a maximum weighting of 15 out of 120 points for European added value within the overall evaluation methodology. This ensures that technical performance, financial viability, and core service quality remain the primary drivers of the award, while EU development acts as a significant tie-breaker or quality enhancer.

What this means for you

For CTOs, architects, and SMEs operating in the cloud and AI space, Article 32 presents both a compliance requirement for public buyers and a strategic opportunity for vendors.

For AI Vendors and Cloud Providers:

  • Highlight EU R&D Integration: When bidding for public sector contracts involving innovative AI systems, explicitly document how your solution integrates technologies developed in the Union. If your models are trained on datasets from EU common data spaces, or if your architecture uses open-source components from EU-funded projects, make this prominent in your technical proposal.
  • Leverage Union-Funded Research: If your company has participated in Horizon Europe, Digital Europe, or other EU R&D programmes, highlight how the results of that research are embedded in your commercial AI models. Article 32(3)(b) specifically calls out "research and development results stemming from Union funded research and development programmes."
  • Supply Chain Transparency: Be prepared to demonstrate the origin of your core technologies. Vendors who can prove that their AI stack relies on EU-designed hardware, software, or standards will have a competitive advantage in the "quality evaluation" phase.

For Public Sector Procurement Officers:

  • Update Tender Templates: You must now include Article 32 criteria in tenders for innovative cloud and AI systems. Ensure your evaluation matrices have a dedicated section for "Union added value" that specifically asks about the integration of EU-developed technologies.
  • Avoid Over-Weighting: Remember that these criteria are ancillary. Do not allow the EU-origin factor to outweigh technical capability or price to such an extent that it becomes the decisive factor, as this would violate Article 32(2)(d).
  • Focus on Innovation: Note that Article 32 applies specifically to procurement for innovative cloud computing services and AI systems. Standard, off-the-shelf IT procurements may not trigger these specific requirements unless they fall under the scope of "innovative" procurement procedures.

Common misconceptions

Misconception 1: CADA bans non-EU AI models. This is incorrect. CADA does not prohibit the purchase of AI models developed outside the EU. Instead, it creates a preference mechanism. Non-EU models can still win contracts, but they will likely score lower on the mandatory "Union added value" criteria compared to EU-developed alternatives that integrate local R&D.

Misconception 2: The "EU-developed" criterion is the most important factor. Article 32(2)(d) clarifies that non-price award criteria must be "ancillary and not decisive." Technical performance, security, and price remain the core determinants of the award. The EU-origin factor is a quality enhancer, not a veto power.

Misconception 3: This applies to all public IT procurement. Article 32(1) limits this requirement to procurement procedures for "innovative cloud computing services and AI systems." Routine procurement of standard software or hardware that does not fall under this innovative AI/cloud scope may not be subject to these specific Article 32 criteria, though other general public procurement rules still apply.

Misconception 4: Only fully EU-owned companies benefit. The criterion focuses on the technology and its development origin, not just the corporate nationality of the vendor. A non-EU company that integrates significant EU-developed AI models or uses EU-funded R&D results in its offering can still score well on Article 32(3)(b).

Related

This is general information about a draft EU regulation, not legal advice.