Back to News
IndustryAI Understanding briefing

Nvidia explores converting German nuclear sites into AI data centers

Nvidia and PreussenElektra have entered a preliminary partnership to assess the feasibility of converting decommissioned nuclear power plants in Germany into high-performance AI data centers.

4 min readRead the original reporting
Source-provided image accompanying Nvidia explores converting German nuclear sites into AI data centers
Attributed reportingSource recorded
Publisher
tomshardware.com
Source type
Reporting by a news outlet — not a first-party document.

What we could not confirm independently: This claim is attributed to the named outlet. We did not verify it against a first-party document. (tomshardware.com)

ContextUnderstand this in 60 seconds

Key terms

Inference
The runtime phase where a trained model generates predictions or outputs.
Compute
The processing resources required to train and run models, often measured in FLOPS or GPU hours.

What happened

Nvidia and German energy firm PreussenElektra (PEL) announced a preliminary partnership on October 21 to evaluate the conversion of decommissioned nuclear power plant sites into AI data centers.

According to Tom's Hardware, Nvidia and PreussenElektra GmbH (PEL) entered a preliminary partnership on October 21. The core objective of this collaboration is to assess whether PEL's decommissioned nuclear power plant sites in Germany can be successfully converted into high-performance AI data centers.

The partnership is described as an exploratory phase, focusing on the technical and logistical feasibility of repurposing these specific energy sites. No construction has begun, and no final commitment to build has been made, as the firms are currently in the assessment stage.

Source details: tomshardware.com

Why it matters

This collaboration addresses the critical infrastructure bottleneck of power availability for AI training. By leveraging existing nuclear sites, the partnership aims to reduce the cost and timeline associated with building new data centers from scratch, potentially accelerating AI deployment in Europe.

AI data centers require immense amounts of power, and securing reliable, high-capacity energy sources is a major hurdle for the industry. Nuclear sites offer the advantage of existing high-voltage grid connections and land infrastructure, which can significantly lower the barrier to entry compared to greenfield projects.

By targeting decommissioned nuclear plants, Nvidia and PEL aim to reduce both the capital expenditure and the development timeline. This approach could provide a faster route to deploying the massive resources needed for large-scale AI model training and in the European market.

Interactive Mechanism

Interactive Mechanism: How It Actually Works

Explore the underlying technology behind this development interactively.

Model Parameter Size:8B Parameters
VRAM Required5.5 GBGPU memory footprint
Target HardwareMacBook / Single GPUDeployment tier
Privacy100% Air-GappedLocal device capability
Core takeaway: Small, quantized models (3B–8B) now run directly inside smartphones and laptops with complete data privacy, while mammoth 400B+ models remain the domain of datacenter clusters.
Interactive Concept Check+10 Points
Future of AI Quiz

What should a useful AI forecast state?

What to watch next

Watch for feasibility study results, regulatory approvals for site conversion, and whether this pilot leads to a broader infrastructure strategy for Nvidia in Europe.

The next step will likely involve detailed engineering studies to determine if the existing infrastructure at PEL's sites can support the cooling and power demands of modern AI hardware.

Regulatory hurdles in Germany regarding the repurposing of nuclear sites will be a key factor. Stakeholders should monitor for any official feasibility reports or announcements regarding the specific sites selected for this pilot program.

Related guides & quizzes

Found this useful?