Summary Under the proposed Cloud and AI Development Act (CADA), Member States are required to actively use the division of procurement into lots as a strategic tool to improve access for small and medium-sized enterprises (SMEs). Article 33(2) mandates that monitoring and reporting on innovation procurement must be used to identify barriers to SME participation and support the design of simplified, proportionate, and SME-friendly strategies, including division into lots where appropriate. This measure aligns with the existing rules of Directive 2014/24/EU but adds a specific obligation for cloud and AI procurement to widen SME participation, aiming to lower entry barriers and foster a competitive European ecosystem.

Detail

The proposed Cloud and AI Development Act (CADA), COM(2026) 502 final, establishes a framework to strengthen Europe's cloud and AI ecosystem. A critical component of this framework is the reform of public procurement practices to ensure they actively support the competitiveness of European providers, particularly smaller enterprises. While the Act sets ambitious targets for SME participation, Article 33 provides the specific operational mechanisms to achieve them.

The Legal Mandate: Article 33(2) and SME Strategy

Article 33, titled "Monitoring of procurement of innovation in cloud and AI," creates a feedback loop between data collection and strategic action. While Article 33(4) sets a quantitative target for Member States to pursue the objective that "at least 25% of their procurement for cloud computing services and AI systems be awarded to innovative SMEs," Article 33(2) defines how Member States must work towards this goal.

The text of Article 33(2) explicitly requires Member States to take appropriate measures to ensure that monitoring and reporting are actively used to:

  1. Identify barriers to SMEs' participation in procurement procedures;
  2. Improve access of SMEs to procurement markets;
  3. Support the design of simplified, proportionate and SME-friendly procurement strategies, including division into lots, where appropriate;
  4. Promote the participation of SMEs in the innovation procedure foreseen under Directive 2014/24/EU and pre-commercial procurement of cloud computing services and AI systems.

This provision is significant because it moves beyond general encouragement. It legally binds Member States to use the data gathered from their procurement monitoring to actively redesign their strategies. If the data shows that SMEs are excluded, the Member State must adjust its approach, with "division into lots" cited as a primary method to achieve this.

Alignment with Directive 2014/24/EU

CADA does not create a new, standalone procurement regime that overrides existing EU law. Instead, it operates in synergy with the Public Procurement Directives, specifically Directive 2014/24/EU on public procurement.

Directive 2014/24/EU already contains provisions encouraging contracting authorities to divide contracts into lots to facilitate the participation of SMEs. However, under the general directive, this is often a matter of discretion or best practice. Article 33(2) of CADA elevates this practice to a specific requirement within the context of cloud and AI innovation procurement. It mandates that Member States must actively consider and implement lot division as part of their strategy to widen SME participation.

The phrase "where appropriate" in Article 33(2) is crucial. It acknowledges that not every procurement can or should be divided. Contracting authorities must exercise professional judgment to determine if splitting a contract into lots is feasible without compromising the coherence of the project or creating disproportionate administrative burdens. However, the burden of proof shifts: authorities must justify why they are not dividing a contract if the monitoring data suggests SMEs are being excluded due to the scale of the tender.

The Strategic Rationale: Why Lots Matter for Cloud and AI

Cloud and AI projects are often characterized by high complexity, significant capital requirements, and the need for specialized expertise. Without intervention, these characteristics naturally favor large, established "hyperscalers" and incumbents who have the financial depth and broad technical capacity to bid on massive, monolithic contracts.

Article 33(2) addresses this market failure by promoting the division into lots for the following reasons:

  • Lowering Financial and Technical Barriers: By breaking a large cloud infrastructure project into smaller lots (e.g., separating storage, compute, and AI model training), the financial guarantees and technical capacity required for each bid are reduced. This allows SMEs with strong niche capabilities to compete for specific components rather than needing to be a "one-stop-shop" for the entire ecosystem.
  • Enabling Specialization: The cloud and AI value chain is diverse. An SME might excel in data governance or specific AI model fine-tuning but lack the capacity to manage global data center operations. Division into lots allows these specialized players to enter the market, fostering innovation that a single large provider might not prioritize.
  • Reducing Risk for SMEs: Committing to a multi-year, multi-million euro contract carries significant risk for a small enterprise. Smaller lots reduce the exposure, encouraging more SMEs to participate in public tenders.
  • Driving Competition: A market dominated by a few large players can lead to stagnation and higher prices. By structuring tenders to allow multiple SMEs to win different lots, contracting authorities can create a more dynamic and competitive supply chain.

Integration with Broader CADA Objectives

The requirement to divide into lots is not an isolated rule; it is part of a broader strategy outlined in Article 33.

  • Monitoring and Reporting: Member States must report annually on SME participation trends, including the number of contracts awarded to SMEs and their share of total contract value (Article 33(3)). This data is the engine that drives the strategy. If the data shows low SME uptake, the "division into lots" mechanism is triggered as a corrective measure.
  • Pre-Commercial Procurement: Article 33(2) also links lot division to pre-commercial procurement (PCP). PCP is designed to stimulate the development of new solutions before they are commercially available. Dividing PCP projects into lots allows multiple SMEs to compete in the R&D phase, increasing the diversity of solutions developed.
  • SME-Friendly Clauses: The strategy supported by lot division is further reinforced by Article 33(5), which encourages authorities to develop public contract clauses that are favourable for innovative SMEs. This creates a holistic approach: smaller lots make the scope accessible, while favourable clauses make the terms manageable.

What this means for you

For public-sector procurement officers, Member State authorities, and SMEs operating in the cloud and AI sector, Article 33(2) represents a shift from passive compliance to active market shaping.

For Public Procurement Officers and Member States

  • Data-Driven Strategy: You must treat procurement monitoring not just as a reporting exercise but as a diagnostic tool. If your reports show low SME participation, you are legally required to adjust your strategy, with division into lots being a primary lever.
  • Lot Design: When drafting tenders for cloud and AI, you must assess whether the scope can be logically divided. Instead of a single "Cloud and AI Transformation" contract, consider separate lots for "Data Storage," "AI Model Development," and "Cybersecurity."
  • Justification: If you decide not to divide a contract into lots, you should be prepared to document why it is not "appropriate" (e.g., technical interdependence that cannot be separated without loss of efficiency).
  • Alignment with Directive 2014/24/EU: Ensure your lot division strategies comply with the existing rules on lot aggregation and the prohibition of artificial splitting, while simultaneously meeting the CADA mandate to widen SME access.

For SMEs and Cloud Providers

  • Market Access: The proposed CADA creates a more favorable environment for SMEs to bid on public contracts. You should actively monitor national strategies and procurement notices for signs of lot division.
  • Specialization: Focus on developing deep expertise in specific areas of the cloud and AI stack. The lot structure is designed to reward specialization rather than just scale.
  • Engagement: Participate in the preliminary market consultations encouraged by Article 33(5). Use these opportunities to advise contracting authorities on how to structure lots that are realistic for SMEs.

Common misconceptions

"Division into lots always leads to fragmentation and inefficiency." While poorly designed lots can create integration challenges, Article 33(2) explicitly includes the qualifier "where appropriate." This empowers contracting authorities to use their expertise to balance competition with operational coherence. Integration can be managed through framework agreements, lead contractor models, or clear interface specifications. The goal is not fragmentation for its own sake, but the removal of artificial barriers to entry.

"SMEs are not capable of handling complex cloud and AI projects." This is a self-fulfilling prophecy that CADA aims to dismantle. Many SMEs possess specialized, cutting-edge expertise in niche areas (e.g., specific AI algorithms, data privacy tools, or edge computing) that large incumbents may overlook. By dividing contracts into lots, authorities can tap into this specialized knowledge. The assumption that only large hyperscalers can deliver cloud services is outdated and contradicts the EU's goal of fostering a diverse, resilient ecosystem.

"Article 33 replaces the existing Public Procurement Directives." CADA does not replace Directive 2014/24/EU. It complements it. The rules on how to divide contracts, the thresholds, and the procedural requirements remain governed by the Directives. CADA adds a layer of specific obligation and monitoring for cloud and AI procurement, requiring Member States to actively use lot division as a tool to meet the specific strategic goals of the EU's cloud and AI policy. It is a sector-specific enhancement to the general framework.

Related

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