Hetzner

A German hosting provider famous for the best raw price/performance in the market — cloud VPS and dedicated servers at a fraction of hyperscaler prices. You trade managed convenience for control and savings: you administer the box (OS, Docker, networking), but you get serious compute for very little money. Ideal for self-hosted, long-running, compute-heavy workloads that would be expensive or impossible on serverless.

Description

  • Cloud VPS — affordable virtual servers with fast NVMe and generous bandwidth.
  • Dedicated servers — bare-metal at standout prices (incl. GPU/AX lines).
  • Volumes & networks — block storage and private networking.
  • EU data residency — German/Finnish datacentres (GDPR-friendly).
  • Full control — root access; run any container/stack you like.

Download or use

Reasoning for

Hetzner runs Qamera AI’s self-hosted backend: a Docker stack hosting the worker + queue (RabbitMQ) that drives AI image/video generation. This is the deliberate counterpart to Vercel in Qamera’s split architecture — the Next.js app lives on Vercel for its deploy/preview DX, while the long-running, GPU/CPU-heavy generation orchestration lives on Hetzner where there are no serverless timeouts and compute is cheap enough to run continuously. For a credit-based product where margins depend on per-generation cost, Hetzner’s price/performance directly protects unit economics. Where a project wants a managed PaaS instead of administering servers, DigitalOcean is the friendlier (pricier) alternative.

Alternatives considered

  • Vercel — paired with Hetzner, not replaced by it: Vercel hosts the web tier, Hetzner the workers.
  • DigitalOcean — managed App Platform + Spaces; easier ops, higher cost — chosen for Tech To The Rescue but not for Qamera’s cost-sensitive compute.
  • AWS EC2 / GCP — vastly more services, far higher cost and complexity for the same raw compute.
  • Fly.io / Railway — nice DX for containers; more expensive at the sustained compute Qamera’s workers need.

Resources


Template: tool