AI raises huge questions about sovereignty but independence is not an option – Professor Steve Watt
The UK relies on technology produced by a small number of mostly American companies
From cloud computing and semiconductors to operating systems, data and digital identity – the phone in your pocket, the systems which keep your money safe in the bank – so much of the infrastructure on which the UK and Europe depend is controlled elsewhere.
Digital sovereignty is about retaining meaningful control over the new technologies on which our society and economy depend. That matters because they increasingly underpin critical national infrastructure and essential services – governments, universities, healthcare, financial services and almost every part of the modern economy.
Yet, despite our growing dependence on technology, we exercise surprisingly little control over much of it. For the UK and Europe, a significant proportion of the digital infrastructure on which we rely is provided by a relatively small number of predominantly American technology companies.
Not an argument against US tech
Microsoft, Amazon and Google play a disproportionately large role in cloud computing. Microsoft, Apple and Google dominate operating systems. Nvidia has become central to the computing infrastructure behind artificial intelligence, while OpenAI, Google, Meta and Anthropic are shaping the emerging market for advanced AI models.
These companies have achieved their positions through unprecedented levels of investment and innovation. The argument for digital sovereignty should, therefore, not become an argument against American technology. The issue is more fundamental. How comfortable are we with the level of dependency we have created?
An organisation may believe that moving its systems into the cloud has simply changed where its computers are located. In reality, it may also have transferred important elements of operational control to another organisation, potentially within another legal jurisdiction.
Data sovereignty adds another dimension. It’s no longer sufficient simply to ask where information is stored. We increasingly need to understand where it is processed, which organisations can access it, what legal regimes apply, how easily it can be moved elsewhere and what happens if geopolitical or commercial circumstances change. Sovereignty, therefore, is not simply about geography. It’s about control, jurisdiction, and the ability to continue operating when circumstances change.
READ MORE: Why I won't let Scotland become so afraid of AI that it loses £23 billion boost to GDP – Alyn Smith
AI platform dependency
Artificial intelligence makes these questions more urgent. When an organisation adopts an AI platform, it may become dependent not only on the software, but the underlying model and services required to operate it.
Over time, applications and business processes will be built around certain models. Employees will become familiar with particular AI ecosystems. Organisational data will become connected to them. This could evolve into outright dependency.
This matters particularly as AI develops from systems that generate information into autonomous agents capable of taking actions. An AI assistant that helps write a report presents one level of dependency. An AI agent capable of accessing databases and initiating transactions creates another.
At that point, sovereignty becomes closely connected to security, identity and accountability. It also raises a more fundamental question of who, or what, has the authority to act on an organisation's behalf?
There’s also the question of physical infrastructure. The digital world sometimes appears intangible, but it depends upon very tangible things – data centres, telecommunications networks, undersea cables and, increasingly, advanced semiconductors.
Technological nationalism not the answer
The EU has demonstrated considerable regulatory power in the digital world but this is not the same as possessing the industrial and technological capacity to build it. The scale of that challenge is already the subject of debates about data centres, with the UK government's AI Growth Zones promising faster planning and grid access while questions grow about energy, water and local consultation.
Europe's leadership in regulating privacy and artificial intelligence needs to be matched by technological, industrial and human capability. Regulation is important, but setting the rules does not equip us to compete within them.
In my view, the answer is not technological nationalism. It would be unrealistic and counterproductive for the UK or Europe to attempt to recreate every layer of the global technology industry. Digital supply chains are international, collaboration with US and other technology companies remains essential.
Instead, we should think about sovereignty in terms of strategic autonomy, choice and resilience. Can organisations move their data between providers? Can critical services continue if a particular supplier becomes unavailable? Do we retain the skills required to understand and operate the technology ourselves? Are there credible European and open-source alternatives for strategically important technologies?
READ MORE: Why that desperate call from your daughter may have been a cybercriminal using AI – Ronald Hepburn
Before committing to a strategically important platform, organisations should be able to explain how they would move their data, maintain critical services and replace the underlying AI model if necessary. They should also identify the skills they need to retain and the credible alternatives available.
These issues should increasingly influence technology strategy and public procurement. The UK's Be Ready campaign asks households to prepare for three days of disruption to essential services such as power, water and the internet. At an institutional and national level, being prepared means considering resilience across digital infrastructure and technology supply chains, alongside the physical infrastructure for which we already plan.
Universities have an important role. They educate the engineers, computer scientists and entrepreneurs who will build future technologies, while also undertaking the research from which new industries emerge. Maintaining domestic technological capability and expertise may ultimately be just as important to digital sovereignty as building data centres or semiconductor factories.
The objective should not be independence from the global technology ecosystem. In an interconnected world, complete technological independence is neither realistic, nor desirable. The objective should be strategic autonomy, retaining sufficient capability to ensure that dependence on any single provider, platform or jurisdiction does not become an unacceptable risk.
For many years, we treated these kinds of decisions as technology or procurement choices. Artificial intelligence shows they’re turning into something much bigger. As AI becomes embedded in the infrastructure of society, we need to decide not only what technology can do for us, but how much control we’re prepared to give away to use it.
Steve Watt is professor of practice in computer science at the University of St Andrews
Comments
Want to join the conversation? Please or to comment on this article.