EU-Wide for the First Time TU Munich Develops Its Own AI Chip Using 7nm Technology

From Sebastian Gerstl | Translated by AI 2 min Reading Time

Related Vendors

The first AI chip in the EU to use modern 7-nanometer technology has been created at the Technical University of Munich. The neuromorphic processor was manufactured according to the standard of the Taiwanese market leader TSMC. Based on this design, ESMC plans to manufacture its own AI processors in Dresden from 2028.

The Chair of AI Chip Design at the Technical University of Munich (TUM) has created the EU's first AI chip using modern 7-nanometer technology. In the picture: Chair holder Prof. Hussam Amrouch.(Image: Andreas Heddergott / TU Munich)
The Chair of AI Chip Design at the Technical University of Munich (TUM) has created the EU's first AI chip using modern 7-nanometer technology. In the picture: Chair holder Prof. Hussam Amrouch.
(Image: Andreas Heddergott / TU Munich)

The Covid pandemic, the war in Ukraine and geopolitical changes have made supply chains uncertain. "That's why it will be important in future to take all steps in Germany and Europe, from the training of AI chip specialists to the technological development and production of AI chips," says Amrouch. Just three months ago, he opened the AI chip research and training center MACHT-AI at TUM - funded by the Bavarian Ministries of Science and Economic Affairs. In future, he envisages that students in Germany will learn AI chip design and develop AI chips and algorithms themselves in companies. It will also be possible to produce them in Germany.

AI Chip Processes Data Locally

Unlike manufacturers of cloud-based AI chips such as Nvidia, the Munich professor is focusing on a technology that processes data directly on the device and does not send it to the cloud. "This is a fundamental solution for protecting the privacy of our citizens," says Amrouch.

The AI chips based on the RISC-V open source standard can be customized for a specific task, for example to record and evaluate heartbeat or brain signals in a health application or to use a language model. This specialization is more efficient than a jack-of-all-trades AI chip. "You can buy a Ferrari, but you won't necessarily be faster in the city. An e-bike is more efficient here," says Amrouch. His AI chips will also be used for control electronics in quantum computing in a few years' time.

Data Remains Private and Secure

By processing the data locally, the technology is cyber-secure and therefore also protected against misuse by third parties. "Whoever designs and builds the chip knows what is inside and can guarantee what it will do," says Amrouch, who believes it is important that companies can guarantee their customers the privacy of their data.

For economic sectors such as the automotive industry, trust is an important "currency" that is currently not in European hands when it comes to powerful chips. And the defense industry also demands absolute security when it comes to the use of chips for drones, for example. Possible built-in Trojans are an incalculable risk here, especially if chips are developed and manufactured outside Germany and the EU.

"Geopolitical risks have increased dramatically in recent years," emphasizes TUM President Thomas Hofmann. "It is therefore important that cutting-edge technology is developed and produced in Germany and Europe. With his AI chips, Prof. Hussam Amrouch impressively demonstrates how dependence on Asia and the USA can be successfully reduced and how Europe's technological self-determination can be strengthened. And, incidentally, with a very sustainable concept." The development in Munich was also expressly praised and welcomed by Bavarian state politicians. (sg)

Subscribe to the newsletter now

Don't Miss out on Our Best Content

By clicking on „Subscribe to Newsletter“ I agree to the processing and use of my data according to the consent form (please expand for details) and accept the Terms of Use. For more information, please see our Privacy Policy. The consent declaration relates, among other things, to the sending of editorial newsletters by email and to data matching for marketing purposes with selected advertising partners (e.g., LinkedIn, Google, Meta)

Unfold for details of your consent