ESA PROSPECT: the ground platform for a drill on the Moon's south pole

PROSPECT will drill up to a metre into the lunar surface near the south pole, collect samples and heat them in a miniature laboratory to look for water ice and other volatiles. It will also test whether oxygen can be extracted from lunar soil, a resource future crews will need. Launch is planned for 2027.

Once it lands, the payload has about ten Earth days of sunlight to do its science. It is not designed to survive the lunar night, and nothing can be repaired. Commands and results pass between the European team at ESTEC in Noordwijk and the Intuitive Machines control centre in Houston, with limited visibility into the commercial lander itself. Every hour of that window counts, so the ground segment has to stay up throughout and be qualified long before launch.

ESA needed a complete PROSPECT Operations Centre (POC) at ESTEC: a control room, plus an IT backend to host the Ground Segment Software, monitoring and secure links to partners. The backend had to run on ESA's own cloud (ESA Cloud IaaS), use only open-source software, support GPU workloads and reach 99.9% availability during operations. It also had to host further missions later and keep working from portable units if the central facility was out of reach.

Our role in the consortium. Telespazio Benelux leads the project as prime contractor. Astro42 leads two work packages: backend compute, storage and network, and the multi-mission platform. We also contribute to the Ground Segment Software deployment, the secure external interfaces, operations preparation and project documentation. Every partner is credited at the end of this page.

Highly available Kubernetes on ESA's cloud. We are deploying a cluster with at least three control plane nodes, GPU worker nodes exposed through the NVIDIA device plugin with quotas, and persistent storage on NFS-CSI and MinIO. Keycloak handles identity over OIDC and LDAP, with Entra ID federation.

Segmented and hardened. The network is split into VLANs on private RFC1918 addressing, with governed firewall rules and an isolated administration network. Hosts follow CIS Level 1 hardening. Prometheus, Grafana and OpenSearch provide metrics, central logging and alerting, so the operations team can see the state of the platform throughout the surface window.

Everything as code. Terraform, Ansible and Helm provision every layer, so the platform can be rebuilt from scratch and audited line by line. The CI/CD pipeline pulls container images by digest and scans them with Trivy.

One platform for many missions. Each mission gets its own namespace and resource quota, while identity, monitoring, storage and CI/CD stay shared and centrally governed. ESA can onboard the next mission without touching PROSPECT operations.

A control centre in a case. Two POC-in-a-Box units run the core services on lightweight K3s clusters, with offline Keycloak credential caching, NTP hold-over and a local registry mirror. They keep operations going if the central facility is unavailable and give the mission a self-contained presence at the US site.

This is an active engagement. The contract is in execution and the project is working through its Preliminary and Critical Design Reviews. Next, the control room and backend platform are installed at ESTEC, ready to receive the mission software and reference hardware. From there the team works toward the Ground Segment Readiness Qualification Review, then supports operations through the IM-4 launch and the surface mission until contract close-out.

We designed the platform as a multi-mission foundation from day one, so ESA can host future missions on the same infrastructure with little onboarding effort. Open-source components and infrastructure as code keep every layer reproducible and auditable. The POC-in-a-Box units give the mission a fallback that runs without the central facility.