December 18, 2025 11:30 CEST
Live on Microsoft Teams.
On 18 December at 11:30 CEST, the ESA Φ-Lab Collaborative Innovation Network will host a new Φ-talk. Details are below.
Meet the speakers
Nicolas Longepe, Earth Observation Data Scientist, ESA
Gianluca Furano, On-Board Computer Engineer, ESA
Talk abstract
This seminar explores the paradigm shift in Earth Observation (EO) and space infrastructure, driven by the integration of Artificial Intelligence (AI) and Geospatial Foundation Model directly into satellites. We will examine one pioneering initiative developed in the framework of the InCubed programme at ESA Phi-lab that is transforming satellites into intelligent, real-time data hubs: the commercial AI-eXpress (AIX) constellation.
The AI-eXpress constellation, led by Planetek Italia and partners, marks the completion of its first generation with the launch of AIX-1 Plus. AIX introduces a "satellite as a service" concept by implementing a hybrid edge/cloud architecture that leverages AI and blockchain for onboard processing. This capability delivers near-zero data latency and transforms raw EO data into immediate, actionable insights, directly challenging traditional ground-based processing models.
More about AI-eXpress
Key-features
- Onboard AI: AI-based analytics and autonomous decision-making performed directly on board via Geospatial Foundation Models.
- 24x7 TM&TC: Demonstration of IOP connectivity through OneWeb telecom constellation, that will provide low speed 24/7 command and control connection via any IP connected computer
- Service Delivery: On-demand services covering data acquisition, processing, downlink, and distribution.
- Modular Applications: Support for domain-adaptive AI inference and application deployment via an orbital "Space App Store".
- Operational Benefits: Reduced data-handling effort and the ability to delete non-relevant data to free memory and conserve bandwidth.
Towards Onboard Autonomy, Versatility and Performance by Geospatial Foundation Models
Delivering actionable intelligence with minimal latency is critical for domains such as disaster response and maritime monitoring. Shifting AI inference from ground-based systems to Earth Observation (EO) satellites is a key enabler for achieving this goal. Geospatial Foundation Models (GFMs), with their strong generalization and few-shot transfer capabilities across diverse tasks, unlock the potential for streamlined deployment and high-performance analytics directly in orbit. Terramind model, developed by IBM Research Europe and ESA Φ-lab has been deployed on AI-eXpress. This achievement bridges research and operational reality, enabling adaptive AI services onboard satellites and paving the way for a new era of intelligent, flexible EO missions, where Cognitive Cloud Computing in Space (3CS) becomes the backbone of a truly semantic, interconnected constellation; and ultimately a federated set of constellations working as a unified cognitive network.
This seminar will cover the technological and strategic objectives of this mission and discuss the opportunities they create for new space companies, researchers, and traditional institutions to deploy and validate AI models directly in orbit.