Digital sovereignty is becoming a question of operational power, hidden dependencies and whether essential services can continue when the systems beneath them fail.
On 20 October 2025, a technical failure inside Amazon Web Services’ Northern Virginia cloud region disrupted digital services across multiple countries. In the United Kingdom, the government subsequently confirmed that the Home Office, the Driver and Vehicle Licensing Agency (DVLA), the Department for Work and Pensions (DWP) and HM Revenue & Customs (HMRC) had experienced impacts.
The failure began inside infrastructure that most citizens would never recognise, much less know how to locate. Yet its consequences reached institutions responsible for services millions of people depend upon.
One detail in Amazon’s technical account of the incident is especially revealing. Engineers corrected the original DNS problem, which affected how systems located an important database service, but recovery did not end there. Other components remained impaired, including systems involved in launching new computing resources and managing network traffic.
In other words, fixing the initial problem did not immediately restore everything that depended on it.
For technical teams, this illustrates the complexity of interconnected cloud infrastructure. For institutions and the people relying on them, it raises a different question.
What happens when an organisation remains responsible for delivering an essential service but lacks the power to restore the infrastructure that makes it possible?
That question takes us beyond a discussion about cloud reliability. It leads directly into digital sovereignty, democratic accountability and the invisible concentration of power beneath modern services.
Why can one cloud failure affect so many different organisations?
Modern digital services often appear independent because they belong to different organisations, use different applications and operate under separate contracts.
A hospital has its clinical systems. A logistics company manages deliveries. A university operates its learning platform. A public administration provides access to official records and services.
From the outside, these appear to be distinct technological environments. Underneath, however, they may rely on shared infrastructure: cloud computing facilities, identity-management services, database systems, communications networks or software components operated by companies several layers removed from the visible service.
This creates an important distinction between supplier diversity and infrastructure diversity.
An organisation might use three different software suppliers and reasonably believe it has avoided dependence on any single company. Yet those suppliers could all rely on the same upstream technology provider.
Different contracts do not necessarily create different points of failure.
The problem can extend beyond the organisation’s direct commercial relationships. A public institution may not have selected or contracted with the operator of a particular infrastructure component, even though its service depends on that component functioning.
This is what makes concentrated digital infrastructure difficult to govern. A decision made by one organisation can contribute to a collective dependency that becomes visible only when something fails.
And when many organisations share the same underlying systems, an interruption can affect entire networks of activity rather than remaining confined to one customer.
What is the accountability gap in digital infrastructure?
The accountability gap is the separation between three positions: the people experiencing the consequences of an interruption, the institutions responsible for the affected services, and the actors who control the technical capability needed to restore them.
These positions sometimes overlap. Increasingly complex technological arrangements, however, can place them in different organisations and jurisdictions.
Imagine that an important public service becomes unavailable.
Citizens naturally contact the institution responsible. They expect information, assistance and an explanation. Public officials may be questioned, administrators may need to account for their decisions, and regulators may investigate whether obligations have been met.
Yet the underlying problem could originate inside an infrastructure component that the institution cannot repair itself.
Its own technology team may understand the failure. Its leadership may have activated every available continuity procedure. Its suppliers may be working responsibly to resolve the disruption.
The institution could nevertheless remain dependent on the actions of another operator.
This is not necessarily evidence of negligence. Specialised technology providers can deliver reliability, security and capabilities that individual organisations would struggle to maintain independently.
The difficulty is that institutional responsibility and operational authority are not always located in the same place.
And they operate on different timescales.
A technical failure can interrupt a service within seconds. Political accountability, legal remedies, contractual disputes and changes in infrastructure policy may take months or years.
Those processes matter enormously for preventing future failures. They may offer little immediate assistance to the person who needs an essential service restored today.
Why digital sovereignty goes deeper than European technology providers
European debates about digital sovereignty often focus on data location, strategic autonomy, industrial capacity and dependence on technology companies headquartered outside Europe.
These questions matter. But the nationality of a supplier cannot, by itself, establish whether an institution possesses meaningful operational autonomy.
A European-owned platform may depend on software or infrastructure controlled elsewhere. A system hosted inside Europe may still rely on proprietary management tools that cannot be replaced quickly. Conversely, internationally developed open-source technologies can sometimes provide organisations with greater freedom to maintain or adapt important components.
Neither geographical location nor an open-source licence guarantees resilience.
The more demanding question is whether an institution retains enough knowledge, authority and workable alternatives to continue fulfilling consequential responsibilities when its normal technological arrangements fail.
In REDefine’s earlier analysis, Europe’s Open Source Moment: Why Digital Sovereignty Cannot Be Built Only with Slogans, we examined why genuine sovereignty requires much more than buying technology from the right country.
This new investigation takes the argument further.
It asks what happens when the power to interrupt, modify or restore a technological capability is concentrated somewhere beyond the reach of the institutions depending on it.
The issue is not simply who supplies the technology. It is where consequential power resides within the infrastructure itself.
That difference matters for governments, businesses and the public alike.
Can organisations be resilient individually but vulnerable collectively?
Yes. An organisation can maintain strong internal safeguards while remaining exposed to failures affecting infrastructure shared by many independent institutions.
Consider two companies with separate applications, different software vendors and independently documented continuity plans.
Both may appear well prepared for an interruption affecting their own systems.
But if their recovery arrangements depend on the same upstream identity provider, cloud management service or communications network, a shared failure could undermine both plans simultaneously.
The important question is whether the alternative arrangements are genuinely independent of the failure they are intended to survive.
A backup is of limited use if restoring it requires an unavailable system. An alternative supplier may offer little immediate protection if it depends on the same upstream infrastructure. And a recovery procedure that works during a controlled test may face different constraints when thousands of organisations need assistance simultaneously.
This is why organisational resilience and systemic resilience are not interchangeable.
Individual organisations can identify and manage many of their own dependencies. They may not have sufficient visibility to assess how those dependencies accumulate across an entire sector or economy.
That wider problem requires cooperation between infrastructure operators, institutions, regulators and policymakers.
It also requires something easily overlooked in digital transformation: the capacity to understand the system as a whole.
As we explored in What Is System Literacy? Why Digital Literacy Is No Longer Enough, an institution may have many people who understand their individual responsibilities while nobody possesses a complete view of the systems connecting them.
When technological dependencies are invisible, it becomes difficult to make informed decisions about resilience, governance and public accountability.
What would meaningful digital sovereignty look like in practice?
Meaningful digital sovereignty does not require every country or organisation to build and operate all its own technology.
Such an approach would be unrealistic, prohibitively expensive and potentially less secure than well-managed international cooperation.
Instead, digital sovereignty should include the practical ability to understand critical dependencies, preserve essential functions and exercise meaningful choices when technological arrangements change.
For institutions responsible for consequential services, that means knowing which activities could be interrupted, which infrastructure makes them possible, who possesses the authority to restore them and what alternatives have actually been demonstrated to work.
For policymakers, it means recognising that risks created by concentrated infrastructure cannot always be managed through individual procurement contracts or institutional cybersecurity policies.
And for citizens, it means that the organisations responsible for delivering essential services should remain capable of offering more than an explanation when something beyond their immediate control goes wrong.
This is ultimately a question of public responsibility.
A hospital should not need to own every computer system supporting its operations. A government should not need to manufacture every technological component used by its public administration.
But the organisations responsible for essential activities must understand what they depend upon and how they will protect those activities when normal arrangements fail.
The difference is between relying on technology and becoming incapable of acting without it.
Beneath the Floorboards: Someone Else Holds the Kill Switch
The October 2025 AWS disruption was an accidental technical failure. Other incidents have revealed different forms of power embedded within digital systems, including the ability to restrict functionality or remotely remove access to something people believed they controlled.
These are not equivalent events. Their causes, responsibilities and consequences differ substantially.
But together they reveal an increasingly important feature of modern technological life: the systems we see are often only the visible endpoints of much larger infrastructures, governed by arrangements that remain almost entirely outside public view.
In our new Code Beneath the Floorboards article, Someone Else Holds the Kill Switch, REDefine examines three documented incidents and traces what they reveal about digital sovereignty, concentrated technological power and the democratic accountability gap.
The professional continuation goes further into the architecture of shared dependencies, systemic risk and the evidence organisations need to establish whether they can continue operating when critical technological capabilities become unavailable.
It also introduces REDefine’s Denial-Power Review, supported by The Kill Switch Audit, a structured professional assessment workbook.
Because when essential infrastructure stops working, the people affected do not begin by asking who owns the server or which supplier operates the network.
They turn to the institutions responsible for protecting them.
And those institutions must retain enough practical capacity to respond.
