EU-based inference / LLM dev teams: where are you hosting right now?
We’re trying to tighten up our infra setup and a lot of the “default” LLM stacks still end up routing through the US in some part of the pipeline (even when the main compute is EU).
Right now we’re looking at a mix depending on latency / compliance / cost:
- Telnyx (EU-friendly infra / comms layer)
- Scaleway (EU cloud option we’ve seen used for hosting parts of pipelines)
- Hetzner (cheap + solid for certain workloads)
- AWS / GCP / Azure (still using them, but trying to be strict about region + data flow)
- plus a few LLM APIs like OpenAI / Anthropic / Mistral depending on use case, though routing + data residency gets tricky once we add tools/RAG/agents
Curious how others are handling this in practice..