Neustart München bekommt eine KI-Fabrik. Gut so?
Nov 6, 2025
Join Eva Wolfangel, a ZEIT digital editor and AI expert, as she dives into the implications of Munich's billion-euro AI data center with Nvidia and Telekom. She raises eyebrows over privacy concerns after her conversation with a medical AI assistant. The discussion tackles Germany's tech sovereignty and whether such a center can truly rival US and Chinese infrastructures. Eva questions the ambitious timeline and the political stakes, exploring Europe's position in the global AI race while calling for necessary reforms in cybersecurity legislation.
AI Snips
Chapters
Transcript
Episode notes
AI Receptionist Phone Call
- Eva Wolfangel described calling her orthopedist and reaching a Dr. Lib AI that recorded calls and repeatedly asked for personal data without booking a slot.
- She tested prompt attacks but found the chatbot strictly constrained and ultimately unhelpful for scheduling.
Scale Means Different Things
- The planned Munich data center will double Germany's AI compute capacity while remaining small compared to US deployments with hundreds of thousands of chips.
- For Germany the project is large and symbolic but for Nvidia it is relatively modest.
Why GPUs Are Essential
- Modern AI training requires GPUs or specialized accelerators because models need massive parallel math operations that CPUs can't handle efficiently.
- Nvidia Blackwell chips are examples of those expensive, specialized components.