The topic of "digital sovereignty" drives us. Questions such as "Where is our data stored?", "Who has access to our data?", and "Who can gain access to the data against our will?" are on the table. Aspects such as autonomy, independence, and data security are all relevant here. Viewed through these dimensions, digital sovereignty emerges as a spectrum rather than a list of bullet points that can be checked off.
04.11.2026
Christian Sprecher
Sovereign AI
There are very clear criteria for operating an AI-based solution (which these days is usually a large language model).
The LLM used must be as open as possible.
It must be possible to define the operation geographically according to one's own requirements
Access to the LLM must be provided via interfaces that are as standardized as possible
Are you interested in AI sovereignty? Contact us for an evaluation of your options.
Christian Sprecher CISO, AI Expert
On-premises or in the cloud?
Modern, open LLMs can run on consumer devices—and our hearts skip a beat whenever we see yet another article on LinkedIn featuring a Mac Studio M5 Ultra cluster.
In our opinion, however, this comes with real drawbacks.
Hardware prices are skyrocketing.
Operating an LLM involves a significant amount of effort.
Reliability comes at a high cost.
Scalability isn’t guaranteed.
And finally: Is running it in an office really secure?
The solution: Swiss providers such as Phoeniqs or Infomaniak offer “Model as a Service.” Similar providers are also available in Europe.
„It's worth reading the exact terms and conditions under which the services are offered. “
Our Conclusion
- AI sovereignty is possible.
- You define the level of autonomy.
- Think about how your data is processed.