Loft Orbital Flies First Vision-Language Model on YAM-9, Cutting Raw Image Downlinks
Updated
Updated · Trend Hunter · Jun 26
Loft Orbital Flies First Vision-Language Model on YAM-9, Cutting Raw Image Downlinks
2 articles · Updated · Trend Hunter · Jun 26
Summary
YAM-9 identified Earth targets directly in orbit from natural-language prompts, in what Loft Orbital said is the first reported vision-language model deployment aboard an Earth observation satellite.
Google DeepMind’s Gemma 3 ran with NASA JPL’s NAVI-Orbital software on constrained hardware including an Nvidia Jetson Orin AGX, showing large AI models can operate within satellite power and memory limits.
The onboard system classified urban-natural boundary zones and spotted infrastructure around railway hubs, reducing the need to send large volumes of raw imagery to the ground for analysis.
Loft built YAM-9 as a pathfinder for its orbital infrastructure-as-a-service model, where third parties can place computing and sensing workloads in space.
For Earth observation users, that approach points to faster alerts, lower bandwidth demand and more autonomous satellite constellations capable of continuous, real-time monitoring.
Orbital AI costs four times more than on Earth. When will this 'space cloud' become commercially viable beyond government contracts?
When satellites decide what data to keep, what crucial information might be permanently lost due to an AI's mistake in orbit?
As AI moves data centers into orbit, how will this reshape global power dynamics and control over real-time information access?
Loft Orbital’s YAM-9: First Vision-Language Model in Space Ushers in Autonomous Satellite Intelligence (2026)
Overview
In April 2026, Loft Orbital made history by launching the YAM-9 satellite, the first Earth observation spacecraft to run a vision-language model (VLM) in orbit. This achievement was made possible through collaboration with NASA’s Jet Propulsion Laboratory and Google DeepMind, integrating advanced technologies like the Gemma 3 VLM, NAVI-Orbital software, and Nvidia Jetson Orin AGX processor. The mission marks a major shift in space AI, allowing satellites to perform tasks that once needed human analysts, such as monitoring areas and providing analyzed insights. This breakthrough opens the door to more autonomous and intelligent space operations.