No one really believed digital sovereignty would be cheap, but a European Centre for International Political Economy report suggests it may not even be affordable. The Real-World Economic Impact of the Cloud and AI Development Act finds that the EU’s landmark bill could cost public and private sector organisations upwards of €115bn between 2027 and 2036 in excess of any savings that might come from ditching non-EU providers.
The report explores three scenarios in which the Cloud and AI Development Act (CADA) is implemented to varying degrees, including a ‘narrow’ scenario in which economic costs to Europe would amount to €35.6bn. A ‘medium’ scenario would cost €115bn, but a ‘broad’ scenario in which sovereign procurement requirements are imposed on 50% of private organisations would saddle the bloc with €233bn in additional costs.
The ECIPE’s proposed solution to such possibilities? Adjust CADA or its implementation so that it doesn’t impose broad restrictions on non-EU technology providers. While the think tank provides a plausible criticism of earlier cost estimates produced by the EU, it’s worth noting that its study was commissioned by CCIA Europe, whose members include hyperscalers Amazon and Google and other US tech firms.
Other studies, meanwhile — including a December 2025 report from McKinsey — suggest that the overall economic impact of European digital sovereignty could be largely positive. Not all independent commentators share this view. Some argue that European organisations will end up taking a middle road, with many using US technology but locally.
Maybe cloud switching costs are the real problem
ECIPE Director Matthias Bauer says the vast bulk of the digital sovereignty’s economic impact will come from migration and system-adjustment costs, which will set back European organisations around €79.1bn under the ‘medium’ scenario.
“Businesses and public authorities may have to move existing applications to alternative cloud environments, replace technologies, modify software and rebuild integrations,” he tells Raconteur. “These processes can be extremely costly, particularly where equivalent services are not readily available.”
Not only is moving to a different provider expensive in itself, but ECIPE argues that European cloud services generally command a significant premium — so much so that higher prices would amount to €34bn in total costs for the EU.
“These additional costs would affect governments, ultimately taxpayers, and businesses in essential industries,” Bauer says. On top of this, ECIPE’s report estimates that EU organisations would lose €1.9bn as a result of delayed digitalisation. Because they’d be required to procure only ‘sovereign’ services, investments and deployments could be delayed if the availability isn’t there, potentially depriving organisations of greater efficiency and productivity.
Delayed productivity may be one thing, but Bauer notes that €80bn of the €115bn total could fall on businesses in essential industries. Under the ‘medium’ scenario, 25% of these businesses are in the private sector, while the ‘broad’ scenario expands to include 50% of them. Article 31 of CADA allows for such a possibility, since it states that procurement requirements may be extended to non-public sector bodies “operating in sectors of high criticality.”
Under the ‘broad’ scenario, this means sovereignty could cost the EU private sector around €116.5bn in total by 2036. This may be a worst-case scenario, but Bauer also explains that ECIPE’s research didn’t consider wider potential effects, such as how sovereignty requirements could impact European export competitiveness, international investment or trading relations. He suspects that such issues could become highly significant.
“If Europe introduces discriminatory restrictions on foreign technology suppliers,” he says, “other governments may respond with comparable measures affecting European businesses abroad. The experience with discriminatory digital services taxes illustrates how international trade tensions and associated costs can ultimately affect domestic businesses and consumers.”
Supporting European growth without restricting outside competition
Based on the potential for such consequences, ECIPE advocates that the EU’s digital sovereignty push should focus on supporting the growth of Europe’s cloud and tech industry rather than simply placing restrictions on non-EU providers. For Bauer, one clear priority should be boosting the development of data centres and computing infrastructure, something which is hampered by a variety of factors.
“Lengthy permitting procedures, regulatory uncertainty and other investment barriers make it unnecessarily difficult to expand European computing capacity,” he says. “Improving these conditions would strengthen Europe’s digital infrastructure without restricting access to technologies that European businesses already depend on.”
ECIPE’s report is at odds with the European Commission’s own estimates of the cost of CADA-derived sovereignty. In June, the Commission published its official impact assessment report, which estimated that public and private entities would face total costs of just over €21bn across the next 10 years. This includes costs of around €3.1bn for private organisations, although this is primarily composed of administrative costs for businesses in the form of risk assessments and corresponding regulatory costs for cloud service providers.
ECIPE might have an argument when it writes that its report “focuses on channels that are not or only partly captured in the European Commission’s impact assessment, including higher cloud prices, migration and software redesign, temporary shortages of qualifying services, delays to cloud and AI adoption, and additional investment requirements.”
On the other hand, Big Tech lobbying group CCIA Europe commissioned the think tank for the study. While the research is methodologically sound, others have drawn quite different conclusions from similar data. In a report published in December, McKinsey concluded that the European economy could boost its GDP by up to €480bn by 2030 in a scenario where high technological sovereignty combines with high AI adoption.
This total is based on estimates that local cloud and AI service providers will capture €63bn of their respective markets, while increased AI adoption would provide a €416bn boost to the EU economy. This latter figure might seem irrelevant, in that adoption can happen under a low-sovereignty scenario as well. But McKinsey’s report suggests little or no sovereignty would limit adoption in sensitive industries. In an ‘externalised growth’ scenario with no strategic control over AI and related technologies, the total boost to GDP would be a somewhat less impressive €375bn.
The act is not law… yet
A spokesperson for the European Commission told Raconteur that the proposed Cloud and AI Development Act has still not been passed into law. The European Council and Parliament still have the opportunity to help finalise the act’s details. Even so, they defended CADA in its current form, arguing that it will result in greater competitiveness, resilience and reduced strategic dependencies in the long term.
“(CADA) will enable more efficient public procurement and faster infrastructure deployment,” they said. “The proposal is supported by a detailed impact assessment, including of economic costs for public and private operators, based on an external study and input from stakeholders.”
As for whether sovereignty will ultimately have wider benefits, such as delivering a greater degree of autonomy and independence, some commentators believe that this probably wouldn’t outweigh or override the bottom line for businesses, even if it did.
“Delivering AI compute is going to be all about cost of tokens,” argues independent tech researcher Richard Windsor, founder of Radio Free Mobile. “This [and] the AI Act will hamper and cripple businesses based in the EU and make them much less competitive.”
Given that implementing CADA in full could incur considerable costs, Windsor suggests that many European firms will opt for a solution that gives them greater control over non-EU products and services. “Nvidia and others have a solution for this, which is to build your own AI compute and run your models locally,” he says. “If the European providers are more expensive, then no one is going to use them and instead buy products from overseas and stand them up in their own premises.”
No one really believed digital sovereignty would be cheap, but a European Centre for International Political Economy report suggests it may not even be affordable. The Real-World Economic Impact of the Cloud and AI Development Act finds that the EU’s landmark bill could cost public and private sector organisations upwards of €115bn between 2027 and 2036 in excess of any savings that might come from ditching non-EU providers.
The report explores three scenarios in which the Cloud and AI Development Act (CADA) is implemented to varying degrees, including a ‘narrow’ scenario in which economic costs to Europe would amount to €35.6bn. A ‘medium’ scenario would cost €115bn, but a ‘broad’ scenario in which sovereign procurement requirements are imposed on 50% of private organisations would saddle the bloc with €233bn in additional costs.
The ECIPE’s proposed solution to such possibilities? Adjust CADA or its implementation so that it doesn’t impose broad restrictions on non-EU technology providers. While the think tank provides a plausible criticism of earlier cost estimates produced by the EU, it’s worth noting that its study was commissioned by CCIA Europe, whose members include hyperscalers Amazon and Google and other US tech firms.
