Writing Zürich
Zürich has the labs. Now it needs the users
Google, OpenAI, Anthropic, Meta and Microsoft all do frontier AI research in Zürich. Swiss companies have been slower to put it to work: only about a third of small firms have built it into their processes.

In 2004 Google opened a small office in Zürich. By 2022 it had more than 5,000 employees in the city, at the Hürlimann-Areal on Brandschenkestrasse and at the Europaallee by the main station. Zürich is its most important research centre outside the United States.
For 20 years that was a Google story. In the past two it became an AI story. In December 2024 OpenAI announced a Zürich office led by Lucas Beyer, Alexander Kolesnikov and Xiaohua Zhai, three researchers it had hired from Google DeepMind. Six months later Mark Zuckerberg hired all three for Meta's superintelligence effort. Anthropic opened a research office in early 2025, led by Neil Houlsby, another DeepMind veteran. In November 2025 Microsoft AI, run by Mustafa Suleyman, announced a Zürich lab of its own, founded by three more researchers from DeepMind.
The pattern is plain. The most valuable AI labs in the world keep hiring each other's people in Zürich, often from a few streets apart. ETH Zürich and EPFL train them, and the labs follow. In September 2025 the two institutes, with the national supercomputing centre, also released Apertus, a fully open Swiss language model, followed by version 1.5 in July 2026.
The other Switzerland
Walk a kilometre from Brandschenkestrasse into the rest of the economy and the picture changes. AXA's annual survey of 336 Swiss small and medium-sized firms, run by Sotomo and published in September 2026, found that 35% deliberately build AI into their processes, 39% are still trying it out and 26% do not use it at all. After a big jump between 2024 and 2025, adoption has levelled off.
Our impression is that Zürich exports frontier research and imports most of its software. The people who build the models sit within walking distance of the fiduciaries, property managers, bakeries and machine builders who could profit from them, and the two rarely meet.
Closing the kilometre
That kilometre is where OptAI® works. We are on Brandschenkestrasse too, a short walk from Google. We take what the labs publish and put it into the systems of Swiss companies, into their ERP, their inbox, their documents and their phones, in their languages and under Swiss law. Some of it we build as products of our own, such as Lotti for social media and Huus& for property professionals.
The labs will keep coming, and they should. A country that trains the people who build frontier AI should also be among the first to use it well. Today only about a third of its small firms have built AI into their work. That is a problem for Swiss productivity, and an opening for anyone willing to do the unglamorous work in between.
Sources
- swissinfo.ch: Google expands presence in Zurich
- Netzwoche: Google erweitert Europaallee-Campus, 27 June 2022
- heise: OpenAI opens office in Switzerland
- TechCrunch: Meta's recruiting blitz claims three OpenAI researchers, 25 June 2025
- The Decoder: Meta hires leading OpenAI researchers
- Greater Zurich Area: Anthropic appoints Neil Houlsby to lead its Zurich office
- Switzerland Global Enterprise: Microsoft to open AI lab in Zurich, 4 November 2025
- EPFL AI Center: Apertus, a fully open, transparent, multilingual language model, 2 September 2025
- EPFL AI Center: Apertus 1.5, 24 July 2026
- AXA: KMU-Arbeitsmarktstudie, Künstliche Intelligenz, 15 September 2026
More writing
LabThe smallest model that does the jobNVIDIA's researchers argue most work inside AI agents is better done by small models. For Swiss companies that is good news on cost, speed and privacy.Author: Rea Stark
IntegrationThe plumbing is the productRivals adopted Anthropic's MCP. OpenAI and 19 investors put over $4bn into a firm that places engineers with customers. AI's value is in the connections.Author: Rea Stark
World modelsLanguage was the easy partYann LeCun's AMI Labs raised $1.03bn for world models. Fei-Fei Li and Google DeepMind work on them too. The next AI has to understand a kitchen.Author: Rea Stark