Digital SME Summit recap: Europe does not lack plans, but a market

Digital SME Summit recap: Europe does not lack plans, but a market

November 19, 2026

The Digital SME Summit brings together European digital SMEs with policymakers, researchers and others involved in the European technology market. The meeting in Brussels on 16 November focused on how European providers can gain access to customers, capital and scale. This gave the debate on digital sovereignty a practical focus: who buys European technology, how do alternatives actually reach the market, and what roles do government, regulation and major buyers play? Arnoud van Gemeren attended on behalf of CIO Platform Nederland and wrote this report.

2025-05-01 twee Europese Consultaties

The underlying theme was strikingly consistent. Europe has researchers, start-ups, specialist software companies and public funding. What is missing is a market in which those companies can scale up. Fragmented rules, weak purchasing power, limited scale-up financing and entrenched preferences for large non-European suppliers reinforce one another. The message to policymakers and buyers was therefore pragmatic: talk less about supply and do more to develop demand, access and scale.

The DMA opens up digital markets
Alberto Bacchiega, Director for Digital Platforms at the European Commission, opened with the Digital Markets Act. The DMA is not a conventional, after-the-fact competition case; instead, it imposes rules on the largest digital platforms in advance. Seven gatekeepers and 23 core platform services currently fall under the regime. This is intended to create room for competitors that currently struggle to get past the gates of dominant ecosystems.

He distinguished three levels of intervention. The Commission begins with regulatory dialogue on practical implementation. If that does not produce a workable outcome, it can specify exactly how an obligation must be applied. The most far-reaching step is a non-compliance investigation, potentially resulting in fines. In his words, the ambition is clear: “To open up digital markets and reduce, and where possible reverse, the entrenchment of dominant platforms.”

The examples illustrated where the DMA can have a tangible impact. Android must give alternative AI assistants access on terms comparable to those available to Gemini. Google must share search data with alternative search engines so that they can improve their algorithms. Interoperability with WhatsApp and Messenger has already enabled new services from European SMEs. Business users must also be able, with their customers’ consent, to access data generated by their activities on a platform.

At the same time, Alberto Bacchiega tempered expectations. The legislation has only applied since March 2024. Every specification and enforcement decision is challenged in court by companies with far greater resources than the Commission. Effective enforcement therefore requires prioritisation, evidence and signals from the market. This is precisely where the difficulty lies: smaller companies often depend for revenue and reach on the very platforms they would need to complain about.

This resonates with CIO Platform Nederland. Interoperability, access to data, reasonable switching options and reduced lock-in are prerequisites for genuine choice. The DMA can help open up markets in this respect. Open standards in procurement and a broader compatibility obligation require additional legislation and, above all, consistent procurement policies.

Philippe Aghion on competition and industrial policy
Philippe Aghion, a professor at the Collège de France, broadened the debate. In his view, AI could increase annual productivity growth by around 0.7 percentage points for ten years through task automation alone. Further growth could come from AI accelerating the production and combination of ideas. But that effect will not emerge automatically. “AI can deliver growth and employment, provided the right institutions are in place.”

His warning concerned concentration. From the mid-1990s, the IT revolution brought strong productivity growth to the United States, but much of it accrued to a small group of superstar firms. Their scale initially stimulated innovation, but later began to discourage new entrants. The same pattern threatens to emerge in AI, where cloud, chips and computing power are already highly concentrated.

Philippe Aghion therefore called for a smart combination of competition policy and industrial policy. Open source and data sharing are needed, as is the extension of DMA principles to the AI value chain and cloud. Regulation must contain societal risks without shutting new entrants out of the market. Industrial policy, in turn, must not become a mechanism for protecting incumbents.

His most striking proposal was a European institution modelled on DARPA, linked to high-performance computing infrastructure: a kind of CERN for AI. Rather than subsidising a single champion, several competing teams would work towards a public mission. He sees no reason to wait for unanimity among 27 Member States: a coalition of willing countries could start with AI, defence, biotechnology or the energy transition.

In his view, Europe’s weakness lies primarily in the transition from basic research to application and scale. This requires a deeper single market, stronger public procurement, more venture capital and participation by institutional investors. The labour market is also part of the picture. AI destroys tasks and jobs while also creating new activities. The Danish flexicurity model shows how labour mobility and protection can go hand in hand.

His analysis provided an economic framework for the debate at the Summit. Policy must give innovative companies room to enter the market, grow and challenge established players. Digital SMEs are the diffusion layer for AI: they translate general-purpose technology into applications for sectors and business processes.

Technological sovereignty in practice
The first panel brought together European and national policymakers, entrepreneurs and a Member of the European Parliament to discuss technological sovereignty. Their starting point was that dependencies are no longer solely an economic issue. Cloud, chips, satellite communications and AI also affect state capacity, security and democratic resilience.

The European Commission is working on a package bringing together the Chips Act 2, the Cloud and AI Development Act and satellite policy. Cooperation with trusted partners remains part of this approach, while Europe builds its own capacity and expands its options. Sovereignty thus acquired a practical meaning: the ability to choose, switch providers and maintain vital functions.

The panellists agreed that public procurement could be the fastest lever. Governments spend substantial amounts on technology, yet in practice often choose familiar non-European platforms. A European preference for critical digital services, smaller procurement lots and a venture-client model could make government the first customer for new providers. This generates not only revenue, but also references that enable sales in the private market.

There was also a warning against “sovereignty washing”. A service is not sovereign simply because its data are stored in Europe or because an American supplier puts forward a European partner. Ownership, jurisdiction, control, the software supply chain and continuity rights determine whether a buyer genuinely retains control. A proposed ultimate sovereignty clause would give customers rights to code, processes and licences if a supplier ceased to meet the conditions following an acquisition or other change.

The analysis of the barriers to growth was incisive. Digital SMEs face overlapping obligations, concentrated markets, bundling and exit costs, differences between 27 national regimes and a shortage of growth capital. Europe may have individual suppliers that each control part of the chain, but it often lacks the federated ecosystem that combines them into a usable offering. This also matters for major buyers: optionality does not come from compiling a list of alternatives, but from designing architecture, standards, contracts and procurement so that alternatives can actually be deployed.

Thomas Byrne: European businesses as the benchmark
Thomas Byrne, Ireland’s Minister of State for European Affairs and Defence, brought the Council’s perspective to the discussion. For him, sovereign technology is not an end in itself, but a matter of competitiveness. Businesses must have access to reliable infrastructure, data, connectivity, computing power and secure services, and then be able to scale across the single market.

He linked this to the Irish EU Presidency agenda: the One Europe, One Market Roadmap, CADA, the Digital Networks Act, the Cybersecurity Act and regulatory simplification. Smaller businesses in particular feel the cost of complexity. Europe’s digital agenda will therefore succeed only when innovation extends beyond large technology companies and frontrunners.

Byrne also pointed to the Dublin Declaration on trustworthy, secure and sovereign AI, signed by all Member States. The next EU multiannual budget must support these ambitions. In the Irish compromise proposal, funding for Horizon Europe increases by around 60 per cent and funding for digital programmes by approximately 285 per cent compared with the current budget.

But his test was deliberately concrete: “Success will not be measured by legislation or investment figures alone.” What matters is whether a small manufacturer in Cork, a cybersecurity company in Dublin, a software business in Berlin or a family business in Milan becomes more productive and competitive. In doing so, he shifted the debate from geopolitical language to the day-to-day reality of businesses.

Adoption of European technology
The second panel comprised representatives of a European technology provider, the Commission, a training organisation, the European Parliament, research and digital transformation. They examined why European technology, despite political attention, is still used only to a limited extent by businesses.

The figures present a mixed picture. A European Investment Bank survey put the use of generative AI by European businesses at 37 per cent, compared with 36 per cent in the United States. Other figures show that around 80 per cent of European businesses do not yet use AI and that much cloud usage remains limited to email, files and collaboration. The depth of usage and the origin of the technology therefore also matter for market development.

Proton described migration as a governance challenge with a technical component. Migrating hundreds of mailboxes is feasible once management has made the decision. Resistance lies in habits, perceptions of risk and career incentives: choosing the established American supplier feels safe, whereas a European alternative requires additional explanation. A realistic pathway takes six, twelve or eighteen months and starts with sensitive data and workloads.

Business value must also be the guiding principle for AI skills. Small businesses disengage when programmes take too much time and offer no near-term prospect of revenue, lower costs, reduced risk or time savings. Training works better when participants arrive with a concrete problem and leave with a usable solution. European Digital Innovation Hubs can provide vendor-neutral support, but they too need to make European alternatives more visible.

Public procurement preferences and vouchers resurfaced as instruments for lowering the initial barrier. Examples include support for migration costs, AI vouchers for European solutions and guidance on integration into business processes. The benefits are greater when AI improves core operations, such as the layout of a factory, than when its use remains limited to writing emails.

At the same time, there was realism about buying behaviour. Sovereignty as an abstract concept has little sales appeal. Customers choose on the basis of functionality, reliability, continuity, confidentiality and control over data. European providers can win precisely by making these benefits tangible. This aligns with CIO Platform Nederland’s position: strategic autonomy becomes meaningful only when translated into business risk, freedom of choice and demonstrable value.

Andrius Kubilius: procurement shapes digital defence
European Commissioner Andrius Kubilius applied the same market dynamics to defence. His argument was straightforward: funding for innovative defence technology must go hand in hand with changes in procurement behaviour by governments and armed forces. “The most effective support for an SME is a government contract.”

For him, Ukraine provides the practical lesson. At the start of the full-scale invasion, there was hardly any established military demand for drones. Start-ups first had to prove that the technology worked. Once it became clear that drones could hit more than 80 per cent of ground targets, demand changed. Europe must therefore not only learn to produce drones and missiles, but also build an ecosystem in which innovation is rapidly tested, procured and scaled up.

The current market is concentrated around large prime contractors and procurement systems designed for major platforms. In several large European countries, the ten largest suppliers are said to receive 67 to 90 per cent of the value of military procurement. This leaves specialist digital companies with little access. Moreover, they cannot adapt their products to 27 separate procurement, certification and testing regimes.

He therefore wants defence procurement reform and a more integrated European market. Alongside advanced bespoke systems, “good enough” solutions are needed that are inexpensive, scalable and quickly replaceable. The battlefield evolves too quickly for traditional cycles: drone software in Ukraine can be outdated after just two months because the adversary has developed a countermeasure.

This also changes the concept of readiness. Large stocks of finished digital systems become obsolete quickly. Long-term contracts must pay companies to keep production capacity, components and skilled staff ready so that they can scale up immediately in a crisis. Kubilius described this as keeping the production facility permanently warm.

Cloud also emerged as a defence capability. Weapons systems depend on digital infrastructure for deployment, updates and maintenance. Dependence on foreign providers can therefore become a strategic vulnerability. A pan-European defence cloud and shared digital services could, according to Kubilius, fall within new European projects of common interest. This creates a direct link between discussions about cloud sovereignty and the hard requirements of security and continuity.

Bringing European technology to market
The third panel brought together the Commission, national government, technology companies, research and implementing bodies to discuss market access for European providers. The discussion focused on CADA, public procurement, AI evaluation and the practical process of switching providers.

The proposed cloud sovereignty levels are intended to address risks proportionately. The most stringent requirements apply only to sensitive applications in areas including defence, healthcare, security and critical infrastructure. For a large proportion of tenders, self-declaration may suffice; at higher levels, an audit is required. The benefit for suppliers is that a single assessment should then provide access to the European market, rather than requiring 27 national processes.

But certification alone does not create an industry. That requires customers and contracts. A software company valued at around one billion euros needs roughly 80 to 100 million euros in annual revenue. Capital can finance the initial phase, but procurement builds the business. Through its own sovereign cloud tender, the European Commission demonstrated that European providers could meet quality requirements and offer lower costs than hyperscale alternatives. Such a contract then raises their profile among other public and private buyers.

A second practical theme was AI evaluation. Organisations that can test models on their own tasks are less dependent on the first supplier they choose. They can use a sufficiently capable model today and switch later when a better option becomes available. Evaluation thus becomes a layer in the European technology stack: even if not every foundation model is European, Europe can assess whether systems suit its applications, values and risks.

NOI Techpark’s experience illustrated how challenging implementation can be. The park aims to be digitally sovereign by the end of 2027 and began selecting technology and preparing infrastructure and data management in 2026. Identity management proved to be the logical starting point because other applications depend on it. The real test will come with users: training, familiarisation and visible benefits will determine whether the change takes hold.

The panel thus underlined an important distinction. European components exist, but integrated solutions that are easy to buy are not always available. Federation must mean more than technical interoperability. Suppliers must also cooperate commercially, organise joint service delivery and offer long-term continuity. Buyers, in turn, must accept that reducing dependency is a transition programme, not a one-off product choice.

Demand as industrial policy
The Summit did not reveal any new shortage of plans. Most of the building blocks are already in place: DMA enforcement, CADA, European AI infrastructure, Digital Innovation Hubs, defence funds, innovation programmes and proposals for a deeper single market. The problem is the connection between those instruments and actual buying behaviour.

Three conditions keep recurring. First, public procurement must open up the market by giving innovative companies contracts and references at an early stage. Second, switching and combining solutions must become technically and contractually feasible, with open standards, data portability, interoperability and reasonable exit terms. Third, growing companies need capital and a European market that prevent them from being sold or relocating at their first significant step towards scale.

This is also what makes the discussion relevant to CIOs and other major buyers. European digital autonomy is shaped in part by architectural choices, tenders, contracts and a willingness to test alternatives seriously. Organisations can start by measuring dependencies, prioritising critical functions and designing their technology stack so that choice becomes practicable.

After a day of discussion, one sentence therefore lingered, even though it appeared in several forms: Europe needs not just technology, but customers. Without demand, sovereignty, innovation and competition remain policy objectives. With targeted demand, they can develop into a market.

Close