ETH Zurich
Last reviewed
Sources
30 citations
Review status
Source-backed
Revision
v2 · 2,137 words
Improve this article
Add missing citations, update stale details, or suggest a clearer explanation.
Suggest edit
CtrlK
Last reviewed
Sources
30 citations
Review status
Source-backed
Revision
v2 · 2,137 words
Add missing citations, update stale details, or suggest a clearer explanation.
ETH Zurich (the Swiss Federal Institute of Technology in Zurich) is a public research university founded by the Swiss Confederation in 1854, with first lectures held on October 16, 1855, and one of continental Europe's most important centers for artificial intelligence and robotics.[1] In the QS World University Rankings 2026 it placed 7th in the world and 1st in continental Europe, the only continental European institution in the global top ten.[2] Its AI and robotics activities span the ETH AI Center (opened October 2020), the legged-robotics work of Marco Hutter's Robotic Systems Lab and its ANYmal quadruped, the Swiss AI Initiative run jointly with EPFL that produced the open Apertus large language model in September 2025, and a prolific pipeline of robotics and AI spin-off companies including ANYbotics.
ETH Zurich enrolled 26,942 students as of 2025, and 22 Nobel laureates have been affiliated with the school as students, faculty, or researchers, including Albert Einstein.[3] Long regarded as one of continental Europe's leading technical universities, it has built that reputation into a position as a hub for AI and robotics research. The university also operates, on behalf of the Swiss Confederation, the Swiss National Supercomputing Centre (CSCS) in Lugano, home of the Alps supercomputer that trains its open foundation models.[4]
The ETH AI Center, opened on October 20, 2020, is the university's interdisciplinary hub for AI research, education, and entrepreneurship, with an emphasis on trustworthy AI. It was launched with machine learning professor Andreas Krause as chair and entrepreneur Alexander Ilic, who previously led Magic Leap's Swiss subsidiary, as its first executive director.[5] As of May 2026 the center brought together 139 professors (36 core faculty and 103 associated faculty) spanning all 16 ETH departments, alongside competitive doctoral and postdoctoral fellowship programs.[6]
ETH Zurich is best known in robotics for the Robotic Systems Lab (RSL) led by Marco Hutter, which developed the ANYmal quadruped robot. The lab pioneered reinforcement learning approaches that train legged locomotion controllers in simulation and transfer them to hardware; in a January 2022 demonstration on Mount Etzel, an ANYmal that fused visual perception with proprioception climbed 120 vertical meters in 31 minutes, about 4 minutes faster than the estimated time for human hikers, with no falls or missteps.[7] In September 2021, the CERBERUS consortium, which included ETH Zurich alongside the University of Nevada Reno, NTNU, UC Berkeley, the University of Oxford, Flyability, and Sierra Nevada Corporation, won the DARPA Subterranean Challenge final and its 2 million dollar first prize using four ANYmal C robots as primary platforms, correctly localizing 23 of 40 hidden artifacts.[8] Hutter also directs the ETH Center for Robotics, and when the RAI Institute (formerly the Boston Dynamics AI Institute) opened its European office in Zurich in early 2024, it chose Hutter to lead it.[9]
Beyond RSL, the Autonomous Systems Lab built up by Roland Siegwart has long worked on autonomous mobile robots and drones, and Raffaello D'Andrea, professor of dynamic systems and control, is known for swarming quadrotor research and for founding drone and warehouse-robotics companies.[10] In machine learning, Krause's group works on probabilistic methods and safe exploration, while Martin Vechev's Secure, Reliable, and Intelligent Systems Lab spans AI for code and AI safety; Vechev also founded the Bulgarian institute INSAIT, created in partnership with ETH Zurich and EPFL.[11] The Institute of Neuroinformatics, run jointly with the University of Zurich since the end of 1995, is a birthplace of neuromorphic engineering and of the dynamic vision sensor behind modern event cameras.[12]
In December 2023, ETH Zurich and EPFL launched the Swiss AI Initiative, a national effort to position Switzerland as a hub for transparent and trustworthy AI and to develop open foundation models. The initiative was seeded with more than 10 million GPU hours on the new Alps supercomputer, computing time its organizers compared to a single GPU running at full load for over 1,100 years.[13] It draws on more than 800 researchers, including around 70 AI-focused professors, from over ten Swiss academic institutions, and its organizers describe it as the largest open-science effort for AI foundation models worldwide.[14]
The hardware backbone is Alps, inaugurated on September 14, 2024 at CSCS in Lugano. Built on the HPE Cray EX platform with 10,752 NVIDIA GH200 Grace Hopper superchips, Alps delivered around 270 petaflops and ranked sixth on the June 2024 TOP500 list.[4] In October 2024 the two universities deepened the collaboration by founding the Swiss National AI Institute (SNAI), a joint venture of the ETH AI Center and the EPFL AI Center co-directed by Alexander Ilic and EPFL's Scarlet Schwiderski-Grosche, backed by CHF 20 million from the ETH Board for 2025 to 2028.[15]
The initiative's flagship result is Apertus (Latin for "open"), released on September 2, 2025 by EPFL, ETH Zurich, and CSCS. The model ships in 8-billion and 70-billion parameter versions, trained on more than 15 trillion tokens covering over 1,000 languages, with roughly 40 percent of the data non-English, including underrepresented languages such as Swiss German and Romansh.[16] Imanol Schlag, the research scientist at the ETH AI Center who co-led the training, said: "Apertus is built for the public good. It stands among the few fully open LLMs at this scale and is the first of its kind to embody multilingualism, transparency, and compliance as foundational design principles."[16]
Unlike most open-weight releases, the entire pipeline is public: architecture, weights, training data recipes, and intermediate checkpoints are documented and released under the Apache 2.0 license.[16][17] The training effort was co-led by Schlag with EPFL professors Antoine Bosselut and Martin Jaggi, and used more than 4,000 of Alps' GH200 chips.[17][18] Apertus launched with distribution through Swisscom, Hugging Face, and the Public AI network, and was later made available on Amazon SageMaker.[16][19] Its developers present it as designed for compliance with the transparency obligations of the EU AI Act, though commentary at release noted that the first versions trailed frontier proprietary models in raw capability, positioning Apertus more as public, sovereign AI infrastructure than as a consumer chatbot.[20] Early public-sector adoption followed: in 2026 the canton of Ticino put an Apertus-powered in-house translation service into use.[21]
ETH Zurich runs one of Europe's most productive university spin-off programs, with 583 companies founded since 1973; a record 43 spin-offs were created in 2023 alone, 12 of them with clear links to AI.[22] Several of its AI and robotics spin-offs have become internationally significant.
| Company | Founded | Focus | Status (June 2026) |
|---|---|---|---|
| ANYbotics | 2016 | ANYmal quadruped robots for industrial inspection | Independent; over 150 million dollars raised, 200+ robots shipped |
| Verity | 2014 | Autonomous indoor inventory drones | Independent; deployed across IKEA warehouses |
| RIVR | 2023 | Wheeled-legged delivery robots | Acquisition by Amazon announced March 20, 2026 |
| Sevensense | 2018 | Visual SLAM navigation for mobile robots | Acquired by ABB in January 2024 |
| LatticeFlow AI | 2020 | AI robustness and compliance evaluation | Independent; co-created the COMPL-AI framework |
| DeepCode | 2016 | AI-based code analysis | Acquired by Snyk in September 2020 |
ANYbotics was founded in 2016 by RSL researchers to commercialize ANYmal and sells autonomous inspection robots, including the explosion-proof ANYmal X, to oil and gas, chemical, power, and mining customers; by September 2025 its total funding exceeded 150 million dollars after Climate Investment joined its investor base.[23][24] Verity, founded in 2014 by D'Andrea two years after Amazon bought his previous company Kiva Systems for 775 million dollars, operates self-flying inventory drones, most visibly inside IKEA stores.[10] RIVR, spun out of Hutter's lab in April 2023 as Swiss-Mile by roboticist Marko Bjelonic, raised a 22 million dollar seed round co-led by Jeff Bezos' Bezos Expeditions and HongShan before Amazon announced its acquisition on March 20, 2026 to apply its wheeled-legged robots to last-mile parcel delivery.[25][26] Sevensense, a 2018 spin-off providing AI-based 3D vision navigation for mobile robots, was acquired by ABB in January 2024.[27] In software, DeepCode, founded in 2016 out of Vechev's lab, was acquired by developer-security firm Snyk in September 2020,[28] and Scandit, a computer-vision barcode and data-capture spin-off of the computer science department, passed a 1 billion dollar valuation in a 150 million dollar Series D led by Warburg Pincus in 2022.[29] LatticeFlow AI, co-founded in 2020 by Vechev, Petar Tsankov, Pavol Bielik, and Krause, released COMPL-AI together with ETH Zurich and INSAIT on October 16, 2024, the first evaluation framework mapping EU AI Act requirements onto technical benchmarks for large language models.[11]
ETH Zurich's scientific lineage includes Albert Einstein, who earned his diploma there in 1900 and returned as a professor from 1912 to 1914; John von Neumann, who completed a chemical engineering degree at ETH in 1926; and Niklaus Wirth, the creator of the Pascal programming language and ETH's only Turing Award winner (1984), who taught there from 1968 to 1999.[3]
In contemporary AI and robotics, prominent faculty include Marco Hutter (legged robotics, RAI Institute Zurich), Roland Siegwart (autonomous systems), Raffaello D'Andrea (dynamic systems, Verity), Andreas Krause (machine learning, ETH AI Center chair), and Martin Vechev (safe and reliable AI, INSAIT founder). Alumni and former faculty populate the wider AI industry: Ce Zhang, an ETH computer science professor until the early 2020s, co-founded Together AI and serves as its chief technology officer;[30] Marko Bjelonic carried RSL's wheeled-legged research into RIVR and then Amazon;[25] and Imanol Schlag co-leads the Swiss AI Initiative's national language-model effort behind Apertus.[17]
ETH Zurich is consistently rated the strongest technical university in continental Europe and a global top-ten institution. In the QS World University Rankings 2026 it placed 7th in the world, holding that position for the third consecutive year and remaining the only continental European university in the global top ten. QS noted that "ETH Zurich has come in seventh place for the third time in a row. This means that it has now been among the world's ten best universities for over 11 years."[2] Its Department of Computer Science (D-INFK) is regularly ranked among the leading computer science programs in Europe and the world, and the university places among the global top ten in many engineering and technology subjects.[2]