Data & AI as Commons
Switzerland is an early mover with MIDATA and data-space initiatives. Yet bridging the gap from the commons concept to practice at scale remains the central challenge.
Where does Switzerland stand, and where are the fault lines?
Training data and the AI capabilities derived from it sit today predominantly with a small number of foreign platform companies. Switzerland has its own public computing infrastructure in the CSCS and the Alps supercomputer. With the open model Apertus, Switzerland demonstrated in 2025 that high-performance AI can also be developed transparently and under public ownership. Data cooperatives such as MIDATA also demonstrate alternative ownership models.
Value creation from data occurs only through aggregation. The individuals who generate it have neither control nor any share in the returns. Without binding rules on data ownership, data levies, and open infrastructure, Switzerland risks becoming a mere data supplier to foreign platform companies rather than benefiting from its own raw material.
Pathways for Switzerland
Following the MIDATA model, cooperatives manage data as trustees on behalf of their members and release it in a controlled way for research and AI training. For Switzerland, this connects to a strong cooperative tradition and builds trust. A share of the value created from their data stays with those who provide it. The approach scales slowly, however, and requires critical mass.
Trade-off: self-determination and trust versus speed of scale and data volume.
Feasibility: A promising ownership model with a real Swiss precedent that merits public support.
The federal government would expand the public computing infrastructure around CSCS and Alps, together with open models such as Apertus, into a universally accessible Universal Basic Services provision. For Switzerland, this safeguards technological sovereignty and prevents one-sided dependence on foreign platform companies. Research institutions, SMEs, and public authorities gain access to trustworthy AI. The approach requires substantial and sustained public investment.
Trade-off: independence and access versus fiscal burden and competition with private providers.
Feasibility: Strategically the strongest option. Apertus and Alps show that Switzerland already has the capabilities.
Protected data spaces with uniform rules on access and use would be established for health, mobility, energy, and finance. For Switzerland, this enables sector-specific value creation within clear trust and data-protection frameworks. Data remains under the control of the respective community. The approach is coordination-intensive and requires agreement among many stakeholders.
Trade-off: precision by sector versus uniformity and cross-sector usability of data.
Feasibility: A pragmatic middle path that aligns well with European data-space initiatives.
The data levy option imposes a charge on the commercial use of large data sets, with the proceeds flowing into public infrastructure and social security. For Switzerland, this creates a new revenue source in an economy where labour is shrinking as a tax base. It links value creation from data to the public interest. Defining the assessment basis and enforcing the levy internationally are, however, demanding tasks.
Trade-off: fiscal participation in data value versus locational attractiveness and the risk of capital flight.
Feasibility: Conceptually important for long-term financing, but viable in practice only once international coordination is in place.
Trigger-based phases
Trigger: Growing dependence on the computing infrastructure of foreign platform companies is being recognised as a strategic risk.
Expanding CSCS, Alps, and open models such as Apertus into accessible Universal Basic Services. Data cooperatives receive public support. Switzerland positions itself as a trusted hub for data.
Trigger: sectors are calling for binding rules on the shared use of their data.
Sectoral data spaces for health, mobility, and energy are being built on common standards. Rights over data and terms of use are being clarified in law. Co-operative models are reaching their first critical mass.
Trigger: Data-driven AI value creation visibly outpaces the earnings generated by conventional paid employment.
Mechanisms for data levies or profit-sharing are being examined and coordinated internationally. Returns flow into infrastructure and social protection. Data and AI are institutionally anchored as a commons.
Real-world pilot programmes
Status: Active; non-profit cooperative running medical data projects.
Members manage their own health data and release it for research in a controlled manner. The model demonstrates how data control can remain with citizens. It provides the blueprint for broader cooperative data use.
Status: Apertus published in September 2025; first cantons, including Ticino, are deploying it in 2026.
EPFL, ETH Zurich and CSCS have developed a fully open, multilingual language model on public infrastructure. It demonstrates technological sovereignty in practice. This foundation makes it possible to build publicly provided AI as Universal Basic Services.
Status: Under development; several sectoral data spaces in the EU.
Gaia-X organises federated data spaces with shared rules for access and sovereignty. The model demonstrates how data sharing can succeed without centralised control. Switzerland should assess its compatibility with European standards.
What we are watching
This measures Switzerland's technological independence. If the share remains below the threshold, dependence on foreign platform companies and their pricing and access conditions will deepen. A shift upward signals gains in sovereignty. For policymakers and investors alike, this means lower concentration risk and a stronger negotiating position.
What is being assessed is the viability of citizen-owned data models. Once the threshold is crossed, a credible alternative to the market dominance of platform groups emerges. A tipping point signals that trust and volume are converging. For investors, this creates new opportunities centred on fiduciary data stewardship and data monetisation.
This pillar measures the ability to make data usable across an industry. Below the threshold, value creation remains fragmented and inefficient. A tipping point means that shared data becomes a competitive location advantage. For policymakers and investors, it signals mature conditions for data-driven business models.
The measure tracks whether trust and neutrality pay off economically. If the threshold is crossed, Switzerland establishes itself as the preferred location for data. A shift signals a viable locational advantage in the economy ahead. For investors it means growth in data centres, fiduciary services, and trust-based data services.