According to TechCrunch, Nvidia has signed a €1 billion partnership with Deutsche Telekom to build an “AI factory” in Munich that will increase Germany’s AI computing power by 50%. The project involves more than 1,000 Nvidia DGX B200 systems and RTX Pro Servers with up to 10,000 Blackwell GPUs. Called the “Industrial AI Cloud,” it’s scheduled to begin operations in early 2026 and will focus on AI inferencing while complying with German data sovereignty laws. Early partners include Perplexity, which will use the facility for in-country AI services, and Agile Robots, whose bots will install server racks. Deutsche Telekom provides the physical infrastructure while SAP contributes its Business Technology platform.
The technical guts of this AI factory
So what exactly are they building here? We’re talking about a massive deployment of Nvidia‘s latest Blackwell architecture GPUs – 10,000 of them. That’s serious computing firepower. The DGX B200 systems they’re using represent Nvidia’s current flagship AI infrastructure, basically supercomputers in a box designed specifically for training and running large AI models.
Here’s the thing about inferencing versus training: training AI models requires enormous computational resources, but running those trained models (inferencing) at scale can be even more demanding long-term. This facility seems focused on the inferencing side, which makes sense for serving German companies that want to use AI without building their own infrastructure.
The European sovereignty angle
This partnership lands right in the middle of Europe’s ongoing debate about technological sovereignty. European companies and governments have been increasingly vocal about reducing dependence on US and Chinese tech infrastructure. Deutsche Telekom’s CEO Tim Höttges specifically mentioned strengthening “European strengths” in their announcement.
But here’s the irony: they’re partnering with Nvidia, an American company, to achieve this sovereignty. It’s a classic case of “the enemy of my enemy is my friend” – European companies want independence from US cloud giants, but they still need Nvidia’s chips because frankly, nobody else can deliver this scale of AI performance right now.
The funding gap is real
Let’s talk numbers for a second. The EU committed €200 billion for “AI gigafactories” across the continent. Sounds impressive until you realize that single companies like Microsoft, Google, and Nvidia are individually spending hundreds of billions on AI infrastructure. This €1 billion project, while significant, is basically a drop in the bucket compared to what’s happening stateside.
Deutsche Telekom was careful to note this is separate from the EU’s initiative, which makes you wonder: are European efforts too fragmented? Instead of coordinated continental strategy, we’re seeing individual countries and companies making their own deals. That approach hasn’t exactly worked well for Europe in previous tech waves.
What this actually means for AI in Europe
For German companies, this could be a game-changer. Having local access to this level of AI infrastructure while keeping data within German borders addresses two major concerns: performance and privacy. Companies that were hesitant to use AI services hosted in US data centers might finally have a viable alternative.
But the bigger question is whether this model can scale across Europe. If every country needs its own €1 billion AI factory to feel sovereign, that’s going to get expensive fast. Still, it’s a start – and given how far behind Europe has fallen in the AI race, any movement is better than none.
