Orbital AI

Orbital AI allows for the deployment of intelligence where it makes the most sense, with onboard compute detecting, classifying, and acting on events without waiting for ground processing. It offers rapid, scaled cloud deployment of new models and utilizes common formats and APIs to seamlessly target compute between Earth and Space. The AICore component brings high-performance compute to the spacecraft, combining CPUs, GPUs, and VPUs to run AI and ML workloads directly onboard across a broad range of mission types. This enables missions to detect and classify objects of interest, fuse data across multiple sensors or payloads, compress and prioritize data before downlink, and autonomously cue other sensors or systems in response to in-orbit events, without relying entirely on a ground-in-the-loop processing cycle. AICore is built for missions where latency is not an option, particularly for defense customers, unlocking mission architectures where speed of decision and operational autonomy are paramount and continuous ground connectivity cannot be assumed. It is also an open platform for on-orbit compute and application development, available to customers with a developer toolkit for building and deploying on-orbit applications running on the same PayloadCore compute environment that manages instruments.