T-Systems Switzerland’s Swiss AI Factory is intended to give organizations an Artificial Intelligence (AI) platform where they can choose the models, storage location, and processing location. The service is aimed particularly at companies handling sensitive data and operators of critical infrastructure. At the same time, Swiss cantons are investing in talent, coordination roles, and research, because a platform alone does not create a functioning AI ecosystem.
Why sovereign infrastructure matters
Digital sovereignty is an organization’s ability to retain control over its data, systems, and technology dependencies. It does not necessarily mean developing every component internally or operating everything exclusively in Switzerland. The practical requirements are transparent choices, controlled data flows, and the ability to switch providers or models.
The announced Swiss AI Factory combines computing power, AI models, and tools for production use, according to T-Systems. The platform uses graphics processing units (GPUs), specialized chips for computationally demanding work, in the company’s Swiss data centers. It also includes GPU-as-a-Service, meaning computing capacity rented as needed, and open-source language models whose technical foundations can generally be inspected and adapted.
Customers are supposed to be able to select which models they use and where their data is processed and stored. For particularly demanding workloads or the training of custom models, T-Systems also offers access to a German cloud with more than 10,000 GPUs. Computing is therefore not confined entirely to Switzerland: organizations can combine Swiss data processing with European capacity.
This choice is a practical benefit, but it is not independent proof of complete sovereignty. Statements about control, security, and platform capabilities come from the provider, and the report does not mention independent testing. Prices, contract terms, and supported languages are also absent, making the platform’s financial appeal to Swiss organizations difficult to assess.
The wider discussion also shows that sovereignty involves more than the location of a data center. A feature on digital self-determination highlights choice, technology dependencies, and the protection of critical infrastructure. It also warns that uncritical use of AI can put human capabilities at risk. Local infrastructure does not automatically solve that problem.
How platforms and governance work together
Governance covers the rules, responsibilities, and controls surrounding the use of a technology. At the Swiss AI Factory, an AI gateway, a central access layer between applications and models, is intended to perform part of that role. It is supposed to enforce anonymization, access controls, and audit logs; audit logs are records that allow access and changes to be reviewed later.
This can have a direct effect on everyday work. A company handling sensitive information could require certain workloads to remain in Swiss data centers and allow only approved models to access them. It could then add German capacity for a computing-intensive task. Whether that separation meets a particular regulatory or contractual requirement is something each organization must determine for itself.
Government bodies are also creating new responsibilities. The canton of Uri plans to establish an AI coordination office in 2027 within the Landammannamt, its central executive office. The cantonal parliament approved the proposal by 52 votes to 8. The plans also include a leadership role for digitalization and information technology strategy, plus additional positions for application management, systems, and projects.
Uri will increase its personnel budget by a total of 11.9 million Swiss francs for the coming four-year period. That amount includes additional roles in the police, migration, and justice departments as well as IT positions; the cost of the AI coordination office is not listed separately. The report therefore indicates a direction, but not a precise price for AI governance.
For residents, a coordination office is initially less visible than a new digital service. Its potential benefit sits behind the scenes: projects can be assessed using consistent rules, responsibilities can be clarified, and IT security can be connected to the digital strategy. Whether this leads to better government services will depend on the office’s mandate, expertise, and execution.
Why talent and research make the difference
Powerful infrastructure has limited value if organizations cannot operate it securely or fit it into useful workflows. A salary study by Michael Page describes broadly stable Swiss salaries, but notable increases in IT and especially for specialists in AI and machine learning. Machine learning refers to methods through which systems identify patterns in example data.
According to Michael Page, the study draws on more than 500 job profiles, about 20,000 recruitment interviews, and a database containing over 500,000 candidates. Companies are reportedly hiring more selectively, reviewing applications for longer, and controlling staffing plans more strictly. They want more than technical knowledge, including communication, stakeholder management—the coordination of affected groups—local experience, adaptability, and teamwork.
This challenges the idea that AI competence mainly consists of knowing how to operate a tool. A person introducing a system in government or business must explain requirements, coordinate access rights, and place its output within existing decisions. According to the study, the ability to work with AI is itself among the most sought-after skills.
The canton of Zug is taking a longer-term approach by funding two professorships at ETH Zurich. It plans to provide up to 55 million Swiss francs over 25 years. One professorship will develop AI tools that search software for vulnerabilities, while the other will study how to protect AI systems and their associated software and hardware from attacks.
The connection to daily life comes through the National Test Institute for Cybersecurity in Zug. It tests digital products and systems, including those used in hospitals and electric-vehicle charging infrastructure. Researchers are expected to develop new testing methods and apply them to real products there. Zug also plans a five-year partnership with ETH’s Zurich Information Security and Privacy Center from April 2027, with funding of up to 3.65 million Swiss francs according to the announcement.
These examples reveal two sides of the same talent challenge. Companies currently need people who can translate AI projects into operational processes, while universities are developing expertise for future security problems. The investments do not guarantee quick relief in the labor market, but they do connect research, public funding, and practical testing.
Pros and Cons of sovereign AI
Pros:
- Control – According to the provider, organizations can choose models, storage locations, and processing locations to match their requirements.
- Security – Access controls, anonymization, and traceable logs can be managed through a central layer.
- Scalability – Swiss data centers can be supplemented with European computing capacity when demand is high.
- Local expertise – Coordination offices, professorships, and testing partnerships build knowledge across government, research, and practical operations.
Cons:
- Dependency – A platform marketed as sovereign still creates reliance on a provider, its interfaces, and its contract terms.
- Unclear costs – No prices are given for the Swiss AI Factory, while Uri does not list the cost of its AI office separately.
- Talent pressure – Rising salaries and more selective recruitment can make AI projects harder for smaller organizations.
- Missing verification – Key performance and security promises are provider claims and are not independently verified in the reports.
What this means for you
If you are introducing AI to a company or government office for the first time, do not begin with the largest available platform. Choose a clearly limited process and document what data it contains, who may access it, and where processing should occur. You can then assess whether a local model, a Swiss data center, or additional European capacity provides a demonstrable benefit.
A useful first application could be a security review of existing software, as proposed in the Zug research initiative. You could also test whether anonymization and access controls work reliably in a system that handles sensitive information. Both examples provide a defined object for evaluation and are more concrete than a general ambition to “use more AI.”
If you already have experience, examine how data, roles, and business processes interact. A sponsored article from PwC and Microsoft argues that business value emerges only when technology, data, processes, and employee adoption fit together. That claim comes from interested parties, but it is consistent with the other reports: more licenses or computing power cannot replace clear priorities and usage rules.
Advanced teams can assign different models and storage locations to different data classes, manage access rights centrally, and review logs regularly. They should also document when German computing capacity may be used and which tasks must remain in Switzerland. Operations, security, and business specialists need to work together, because technology-only projects tend to miss the organizational half of the problem.
Switzerland’s emerging AI landscape is therefore not defined by one flagship project, but by a network of infrastructure, government, labor-market, and research initiatives. The Swiss AI Factory expands the available choices, while Uri is creating responsibilities and Zug is investing in long-term security expertise. The main unresolved issues are the actual costs, independent verification of security claims, and whether enough qualified people will be available to turn these initiatives into dependable applications.
Sources
- T-Systems Schweiz lanciert Plattform für souveräne KI – Medium unknown, 2026-11-12
- Schweizer IT-Löhne steigen – besonders jene der KI-Profis – Medium unknown, 2026-11-12
- Kanton Uri erhält eine Koordinationsstelle für KI – Medium unknown, 2026-11-12
- Die neue Netzwoche und das Special “Digitale Souveränität” sind da – Medium unknown, 2026-11-12
- Kanton Zug finanziert zwei ETH-Professuren für KI und Cybersicherheit – Medium unknown, 2026-11-12
- KI schafft Potenzial. Transformation schafft Geschäftswert. – Medium unknown, 2026-11-12
Image: panumas nikhomkhai via Pexels