Marlan Space and Loft Orbital Announced AI Satellite Fleet
A $1 billion program will deploy 50 AI-enabled satellites by 2031 to process imagery directly in orbit.
Updated on Sept. 28, 2026 in Artificial Intelligence
Live Poll
Do you trust AI-equipped satellite networks to process sensitive information securely in space?
Marlan Space and Loft Orbital have launched a $1 billion initiative to build a 50-satellite constellation equipped with onboard artificial intelligence. The project, which uses models from Mistral AI, is currently in the research and production stage with ten satellites under construction.
Why it matters
The constellation aims to eliminate the latency bottleneck associated with downloading raw imagery for ground processing by enabling real-time interpretation onboard. This approach shifts data analysis from terrestrial servers to the edge of space, potentially delivering actionable alerts in seconds.
The constellation will support 50 satellites, with 10 units currently in production in Abu Dhabi. The system utilizes an orbital marketplace where users select specific large language models to process data from onboard radar and optical sensors.
The players
Marlan Space
An aerospace developer co-leading the consortium to integrate edge AI into satellite hardware.
Loft Orbital
A provider of modular satellite infrastructure that manages the marketplace for space-based processing.
Mistral AI
A French artificial intelligence company that supplies the large language models for onboard data processing.
Orbitworks
An aerospace manufacturing firm based in Abu Dhabi responsible for constructing the satellite hardware.
The details
The satellites operate by performing onboard data interpretation, which removes the need to transmit large volumes of raw imagery to ground stations for traditional processing. This is enabled by an integrated app store that allows operators to upload and run specific large language models—algorithms trained on massive datasets to recognize patterns—to analyze sensor inputs in real time. Orbitworks is managing the hardware manufacturing process for the fleet in Abu Dhabi.
Timeline
2026-2031: Deployment period for the 50-satellite constellation.
The Tech Race
The project follows a broader industry trend of moving compute resources from terrestrial data centers to low-earth orbit. This effort represents a significant escalation in space-based edge computing, moving beyond passive observation toward active, autonomous data interpretation.
Industrial and governmental users will eventually access this network via a centralized app store to select AI models for specialized satellite imagery analysis. While the hardware is currently under production, the timing for user-facing availability remains dependent on the five-year launch window.
The takeaway
The transition to orbital AI signals a shift where satellites function as autonomous processing nodes rather than simple cameras. Industry observers should watch for the first orbital testing phase of the initial ten units as a benchmark for processing accuracy compared to traditional ground-based analysis.
Further reading
For broader trends in orbital machine learning, see our coverage of Artificial Intelligence.
Source note: This article includes information reported by TechRadar.
Live Poll
Do you trust AI-equipped satellite networks to process sensitive information securely in space?
