Hetzner is working on LLM Inference

News has emerged that Hetzner, a major European cloud service provider, is preparing a GPU-based LLM Inference service.

Currently, Hetzner has been providing infrastructure centered on traditional CPU/GPU virtual servers and Dedicated Servers, but it appears to be moving to expand into AI-dedicated infrastructure in order to respond to demand for AI workloads.

The key points are as follows:

• GPU Infrastructure Expansion: Beyond existing general-purpose computing, the company plans to build a GPU-accelerated environment optimized for LLM inference.
• Cost Efficiency: Based on higher cost-effectiveness compared to existing cloud giants (AWS, GCP, etc.), it is expected to target AI developers and enterprises.
• Infrastructure Shift: Beyond simple server rental, the goal is to provide an infrastructure layer specialized for Model Serving.