Aitech to integrate NVIDIA IGX Thor platform into S-A2300 space AI supercomputer for edge computing missions

CHATSWORTH, Calif. - Aitech in Chatsworth, Calif., a rugged embedded computing company focused on artificial intelligence (AI)-ready systems for aerospace and defense applications, announced it will integrate the NVIDIA IGX Thor platform into its S-A2300 COTS AI Supercomputer and future system designs to expand onboard AI processing for space and other extreme environments.

Aitech said the integration is intended to accelerate deployment of mission-ready edge AI computing systems designed for low Earth orbit and other harsh operational domains. The company stated the move builds on its long-standing work in ruggedized high-performance computing, including early development of space-qualified GPU-based processing systems.

The integration will incorporate the NVIDIA IGX Thor into Aitech’s S-A2300 architecture, which is based on the NVIDIA Jetson AGX Orin Industrial system-on-module. According to Aitech, the IGX Thor platform will expand compute capacity and high-speed networking capabilities for real-time inference workloads in space applications.

Aitech said it now has access to IGX Thor evaluation boards and modules, enabling hardware-in-the-loop testing, system integration, and accelerated development cycles for future mission systems.

The company also highlighted its existing S-A2300 COTS AI Supercomputer, designed for low Earth orbit missions and optimized for radiation tolerance and extreme environmental conditions. Key capabilities include high-density AI processing for onboard applications such as autonomous navigation, real-time image processing, climate monitoring, and space debris tracking.

The S-A2300 architecture currently supports up to 2,048 CUDA cores and 64 Tensor cores, delivering up to 248 TOPS of AI compute performance. It also includes dual configurable NVIDIA Deep Learning Accelerator engines intended to support standardized inference workloads across spaceborne systems.

With the addition of IGX Thor, Aitech said future configurations could significantly expand AI compute performance, up to 5,581 FP4 TFLOPS, along with 400 GbE connectivity enabled by NVIDIA ConnectX-7 SmartNIC technology. The platform also includes a Functional Safety Island within the Thor system-on-chip to support safety-critical operations through the IGX software stack.

Aitech said the combined architecture is intended to reduce development timelines by enabling earlier validation of AI workloads and system integration using evaluation hardware before deployment.

For more information, please visit https://aitechsystems.com/.