Nebius AI Cloud
Amsterdam AI cloud with managed MLflow for experiment tracking, on-demand GPUs and regions in Finland, France and the UK
Quick Overview
| Company | Nebius B.V. |
|---|---|
| Category | MLOps Platforms |
| Headquarters | Amsterdam, Netherlands |
| Founded | Not stated |
| EU Presence | Yes - Netherlands (EU) |
| Data Location | You choose a region when creating a cluster: Finland, France, UK, Israel or US (Kansas City) |
| Open Source | No |
| GDPR Compliant | Yes |
| Self-Hosting | No self-hosting stated (managed service) |
| Main Features | Managed MLflow, on-demand and preemptible GPUs, serverless AI, managed inference, Slurm and orchestration integrations, shared filesystem |
| Pricing | GPU prices per hour on its price list, for example HGX H100 from $4.50 on demand; MLflow cost see website |
| Best For | Teams that want managed MLflow and GPUs from the same European provider |
| Replaces | Weights & Biases hosted tracking, SageMaker, Azure ML |
Detailed Review
Nebius is an infrastructure company first. Its AI cloud sells GPU capacity, and Managed Service for MLflow gives you a hosted tracking server where you log runs, version models, compare metrics and deploy the best one.
The Terms of Use name Nebius B.V. as the provider, and the privacy policy gives its address as Burgerweeshuispad 101, Amsterdam. The same policy lists subsidiaries in Israel and the United States.
Managed MLflow
MLflow is open source, and Nebius runs it as a highly available service so you do not operate the tracking server and database. The documentation covers creating, monitoring and maintaining clusters.
Because it is standard MLflow, existing logging code works, which makes migration from other trackers straightforward.
Regions you pick
Managed MLflow is available in Finland, Israel, Kansas City in the US, the United Kingdom and France. For an EU-only data policy you select a Finnish or French region when you create the cluster.
You select the region for each cluster at creation.
GPU price list
The public price list shows on-demand GPU hours, for example HGX H100 at $4.50, H200 at $5.40 and B200 at $8.50, plus preemptible rates from $0.79 for some models. Storage and CPU prices are listed too.
Prices are shown in US dollars.
Limitations to Consider
Nebius is a compute cloud with MLflow attached, not a full MLOps product, so pipelines, feature stores and data versioning come from elsewhere such as ZenML or Hopsworks. Its privacy policy lists a US subsidiary, Nebius Inc., and MLflow runs in a US region too, so you have to select an EU region deliberately. Prices are in dollars, and the MLflow service price was not found on the pages read. It is a managed service.
Who Should Use Nebius AI Cloud
Teams that already rent GPUs and want hosted MLflow from a Dutch contracting entity in a chosen European region.
Pros and Cons
Pros
- Managed MLflow with standard logging code
- Region choice including Finland, France and the UK
- Public GPU price list
- Dutch contracting entity (Nebius B.V.)
- GPUs and tracking from one provider
Cons
- Compute cloud first, MLOps second
- US subsidiary listed in the privacy policy
- MLflow service price not found
- Prices in US dollars
Alternatives to Nebius AI Cloud
Other European mlops platforms worth comparing before you decide:
ZenML
Germany · Munich open source pipeline orchestration with artifact and model tracking
Valohai
Finland · Turku platform for versioned ML runs, pipelines and GPU scheduling
Hopsworks
Sweden · Stockholm feature store with model registry and serving, free tier
DagsHub
Israel · Experiment tracking, data versioning and annotation on one platform
See all European mlops platforms or compare them directly against Weights & Biases hosted tracking on our alternatives page.
Frequently Asked Questions
The documentation lists Finland, Israel, Kansas City in the US, the United Kingdom and France.
The Terms of Use name Nebius B.V. as the provider; the privacy policy also lists Nebius Israel Ltd and Nebius Inc. as subsidiaries.
No. It is an AI cloud with managed MLflow and built-in tooling, so many teams add a pipeline or feature tool.