For users interested in running AI/ML workloads within the European Weather Cloud, we have published a number of Items that help setting up commonly required software stacks. You may deploy those onto an existing instance, and combine them as needed.

The stacks currently part of this collection are:

  • ML Basic: provides a Conda environment with the basic AI/ML tools in python such as Torch, Tensorflow, Keras, Scikit-learn, and others.
  • AI Models: sets up a Conda environment with the AI-models package, which allows you to run popular data-driven weather forecasting models such as Panguweather or Graphcast.
  • Anemoi: installs a Conda environment featuring all the Anemoi components. It includes the basic packages such as datasets, training, graphs, models or inference.
  • AIFS Single MSE: installs the ECMWF AIFS Single MSE Data-Driven Forecasting system and supporting dependencies.
  • AIFS ENS CRPS: installs the ECMWF AIFS ENS CRPS Data-Driven Forecasting system and supporting dependencies.

Disk space requirements

AI/ML software stacks usually require a significant amount of disk space.  We recommend at least 15 GB of free disk for each of Item you may want to deploy.

How to provision an AI stack

If provisioning after October of 2025, we recommend deploying based on the community hub, to benefit form upcoming features or fixes.

Deployment based on the EWC Community Hub

When deploying AI stack Items in particular, we suggest the use of the EWCCLI, as it assists you in provisioning a new VM and deploying the Item with reasonable defaults (see Deploying via ewccli).

Checkout the corresponding Community Hub Items:

Deployment via Morpheus UI (deprecated)

This method of deployment is deprecated, and kept in documentation only as reference for  ECMWF tenancies that relied on it prior to October of 2025.

1. Go to Library → Workflows and verify that the relevant Morpheus Workflows are available (filter by label "ecmwf-ai-stacks" ):

2. Go to Provisioning →  Instances, and select an existing VM to deploy onto. If you do not have a VM yet, you can provision one via Morpheus UI by following these steps.
3. Click Actions →  Run Workflow and type the name of the desired one. Some stacks offer the possibility of customizing certain parameters, such as the path of the Conda installation. You may tune them if needed, for example if you are installing to a secondary volume with more space. You can also configure the Ansible command options (for example, passing a -v for a more verbose output).
4. Click Execute
5. You may follow the progress of the workflow execution under History.