The recent abolition of the Department for Science, Innovation and Technology (DSIT) has raised questions about the future of the UK’s ambitions in artificial intelligence (AI). While Kanishka Narayan’s new role as Minister for AI ensures that AI has a voice in the Cabinet, the integration of DSIT’s responsibilities into other government departments has sparked uncertainty about how the UK will coordinate the infrastructure, funding, and public-sector adoption needed to advance its AI goals. Recent parliamentary reviews of AI contracts in the public sector have also highlighted growing concerns about how well institutions can manage and control AI systems once they are in use. The push for "sovereign AI" is gaining momentum, emphasizing the need for the UK to not only access powerful AI models but also maintain control over how they are used in critical services. Sovereign AI is often mistakenly equated with simply keeping data within the country, but that’s only part of the picture. Even if data is stored domestically, the systems that process it may rely on foreign companies for their operation. True sovereignty means that public institutions should be able to understand, govern, and continue running these systems even if the original provider changes. This includes the ability to adapt models, move workloads, or replace components without rebuilding entire services from scratch. There is a common misconception that sovereign AI must be slower or less advanced than using the largest commercial AI platforms. However, this ignores the fact that many public-sector tasks do not require the most cutting-edge models. Instead, a tailored AI system that fits the specific needs of an organization, with strong governance and adaptability, can be more effective. Open-source and portable AI technologies can support this by allowing institutions to run models on different systems, test their performance, and upgrade components as needed. While openness alone doesn’t guarantee control, it increases flexibility and reduces dependency on a single provider. The UK already has strong foundations for AI development, including top universities, experienced researchers, and growing AI companies. The next step is to ensure these resources are effectively used in areas of the public sector where control over AI is most critical. The government doesn’t need to favor one domestic company or exclude international players. Instead, it can create a competitive market where public bodies retain control over the systems they use. This could involve setting procurement rules that require AI systems to be portable, independently testable, and compatible with infrastructure controlled by the user. This approach would allow public institutions to make informed choices now, rather than facing unforeseen challenges later. By treating the restructuring of DSIT as an opportunity to clarify goals rather than reduce ambitions, the UK can ensure its AI strategy remains adaptable and resilient in the face of changing technologies and suppliers.