¶ Switzerland and AI
As of: January 2026. The Swiss AI landscape in the international context.
¶ Political decisions: as of 6 September 2026
New political decisions checked on 6 September 2026; earlier background retains its stated reference date.
23.06.2026 — Business 26.3221 calls for a digital sovereignty stimulus programme. Its updated procedural status is assignment to the responsible committee. This does not establish an adopted or funded federal programme. Original source.
29.06.2026 — The EU Council gave final approval to amendments to AI rules. Its release specifies delayed high-risk application dates: 2 December 2027 for stand-alone systems and 2 August 2028 for product-integrated systems. This does not postpone every AI obligation; the published legal text remains authoritative. Original source.
01.09.2026 — For business 24.3209 on sovereign digital infrastructure in the AI age, examination by the National Council committee is complete. This procedural step does not mean legislation is complete or infrastructure already operational. Original source.
04.09.2026 — The Administrative Delegation approved an approximately one-year PIA pilot from the autumn session, costing up to CHF 150,000 from the existing IT budget. Swisscom operates it in Switzerland using open-weight models, without training on inputs. Current internet research is not yet supported. Original source.
06.09.2026 — Verified status: Switzerland is preparing implementation of the Council of Europe AI convention. The Federal Council requested a consultation draft by the end of 2026; an intention to ratify is not completed ratification. Original source.
Further reading: KI / IA / AI · Digital sovereignty
¶ Rank 3 of 186 -- and Still No Reason for Complacency
The International Monetary Fund (IMF) maintains an AI Preparedness Index that measures how well countries are prepared for the AI revolution. Switzerland ranks 3rd out of 186 countries -- behind Singapore and Denmark, ahead of the USA [1].
The ranking assesses digital infrastructure, human capital, innovation capacity and regulatory frameworks. Switzerland scores highly in all categories -- yet the index measures readiness, not implementation.
¶ The Four Pillars of Swiss AI Research
¶ ETH Zurich -- AI Center
ETH Zurich regularly ranks first in European rankings for AI research and AI patents. In October 2020, the ETH AI Center opened in Zurich-Oerlikon -- a consortium of over 100 professorships and more than 1,500 researchers from ten fields of research [2].
The AI Center is one of the largest AI hubs worldwide, connecting fundamental research with applied AI in medicine, robotics, climate science and materials research.
¶ EPFL Lausanne
The Ecole Polytechnique Federale de Lausanne (EPFL) operates the Centre for Intelligent Systems and ranks among the world's leading universities in machine learning and computational neuroscience. EPFL complements ETH as the second pillar of public AI research [3].
¶ IDSIA Lugano -- Where LSTM Was Born
The Istituto Dalle Molle di Studi sull'Intelligenza Artificiale (IDSIA) in Lugano, founded in 1988, produced one of the most influential AI architectures of the 20th century with the invention of Long Short-Term Memory (LSTM) networks [4].
Juergen Schmidhuber and Sepp Hochreiter published LSTM in 1997. By the mid-2010s, LSTMs were running on over three billion devices worldwide -- in Apple's Siri, in Google Translate, in Amazon's Alexa [4]. Swiss fundamental research that changed the world, almost without the world noticing.
¶ Swiss AI Initiative
The Swiss AI Initiative unites over 70 professors from Swiss universities and uses the supercomputer Alps -- one of the most powerful research computers in the world -- to develop industry-specific AI models [2].
¶ Google Zurich -- 5,000 and Counting
Google opened its first office outside the USA in Zurich in 2004 -- with two employees. By 2025, the site had grown to around 5,000 employees from 85 nations -- the company's largest development centre outside the United States [5].
The "Zooglers" work on Google Search, YouTube, Google Cloud and Google Translate. Since 2016, the site has hosted its own machine learning research group focusing on natural language recognition and processing [5].
Google did not choose Zurich by accident. The density of AI talent per square kilometre in Switzerland is higher than almost anywhere else in the world.
¶ AI Ecosystem at a Glance
| Institution | Location | Focus | Since |
|---|---|---|---|
| ETH AI Center | Zurich | Interdisciplinary AI research | 2020 |
| EPFL CIS | Lausanne | Intelligent Systems, ML | -- |
| IDSIA | Lugano | Deep Learning, LSTM, Robotics | 1988 |
| Google Zurich | Zurich | Search, NLP, Cloud AI | 2004 |
| IDIAP | Martigny | Speech recognition, ML | 1991 |
| CSCS (Alps) | Lugano | Supercomputing for AI | 1991 |
| Disney Research | Zurich | Computer Vision, Graphics | 2008 |
For a country of 8.9 million inhabitants, this density of AI research institutions is remarkable.
¶ Strengths and Weaknesses
Strengths:
- World-class universities (ETH and EPFL among the global top 10)
- High density of AI talent and international researchers
- Political stability and legal certainty
- Strong private sector with AI adoption (pharma, finance, industry)
Weaknesses:
- No own large language model (dependence on US and Chinese providers)
- Energy bottleneck: 60% hydropower, but 2017 popular vote against new nuclear power plants [6]
- Lack of a national AI strategy with binding investment targets
- Small domestic market limits scaling
¶ The Energy Question
Whoever wants to be an AI hub needs electricity -- lots of it. The International Energy Agency (IEA) estimates global electricity consumption by data centres in 2024 at 415 terawatt-hours. By 2030, this figure is expected to more than double to over 945 TWh [7].
Switzerland consumes a total of around 58 TWh per year [8]. A country that wants to be an AI hub but lacks the energy to power it will become a spectator to a revolution happening elsewhere.
¶ Bibliography
[1] IMF: AI Preparedness Index. International Monetary Fund, 2024.
[2] ETH Zurich: ETH AI Center -- About Us. ai.ethz.ch, 2024.
[3] EPFL: Centre for Intelligent Systems. epfl.ch, 2024.
[4] Schmidhuber, Juergen / Hochreiter, Sepp: Long Short-Term Memory. In: Neural Computation 9(8), 1997.
[5] Google Switzerland: About Google Zurich. about.google/intl/de_ch, 2024.
[6] Federal Department of the Environment, Transport, Energy and Communications (DETEC): Energy Strategy 2050 -- Popular vote 21 May 2017.
[7] International Energy Agency (IEA): Electricity Mid-Year Update 2025. iea.org, 2025.
[8] Federal Statistical Office (FSO): Swiss Overall Energy Statistics 2024.