---
canonical: https://enum.co/industry/ai
locale: en
---

# AI. Your data stays yours.

Teams run AI applications and agents on enum in Frankfurt: services and retrieval pipelines on managed Kubernetes, documents and embeddings in S3-compatible object storage, with no US cloud underneath. GPU capacity is on the roadmap for 2027.

Run AI applications, agents and retrieval pipelines in Frankfurt, so documents, embeddings and logs stay in Germany. Pair it with an EU model provider, or open models on enum, to keep inference in Europe too.

## What changes. From data risk to control.

- Sensitive documents end up in US cloud services → Documents, embeddings and logs stay in Frankfurt
- Agents and pipelines are glued together across vendors → Services, workers and vector stores run side by side on Kubernetes
- Customers ask where their data goes during inference → A German provider with no US parent: one clear answer
- Usage-based bills are hard to predict → Published euro prices per resource, no fees for API requests

## How AI teams run on enum. From documents to agents.

1. **Store the data**: Documents, datasets and embeddings in S3-compatible object storage in Frankfurt.
2. **Run the services**: APIs, workers, vector databases and agents as workloads on Kubernetes.
3. **Connect your models**: Call the model provider you choose, or run open models on CPU today. GPU capacity is on our roadmap for 2027.

## What you run it on.

- [enum Kubernetes Engine](https://enum.co/kubernetes-engine): Services, workers and agents on upstream Kubernetes.
- [Object Storage](https://enum.co/object-storage): Documents, datasets and embeddings in Frankfurt.
- [Networking](https://enum.co/networking): Private networking between services, load balancers at the edge.

Day 1: BasePeak has run sovereign AI agents for public administration on enum since its first commit.

## Questions. About AI on enum.

### Does enum offer GPUs?

Not yet. GPU capacity is on our roadmap for 2027. Today teams run AI applications, agents, retrieval pipelines and smaller open models on enum, and call the model provider of their choice.

### Where does our data live?

In Frankfurt, Germany, on infrastructure enum operates itself. enum GmbH is a German company with no US parent.

### Can we run vector databases on enum?

Yes. Vector databases run as workloads on Kubernetes, with block storage for their data and object storage for backups.

### Who uses enum for AI?

BasePeak runs sovereign AI agents for public administration on enum, since its first commit.