Asus announces PE3000N mini PC powered by Nvidia Jetson IGX Thor T5000 with 2,070 FP4 TFLOPS AI processing power

Nvidia introduced the Jetson IGX Thor platform back in August, but availability was staggered, just as it happened with the DGX Spark platform. Now that the dev kit as well as the separate module are finally available for purchase through the official site, Nvidia’s OEM partners also prepare to launch their own devices powered by the Jetson IGX Thor platform. One of the first partners to announce such devices is Asus IoT with the PE3000N rugged mini PC engineered to meet the advanced requirements of next-generation robotics and intelligent automation.

The Jetson IGX Thor T5000 module integrates a Blackwell GPU together with a 14-core ARM processor and 128 GB LPDDR5X RAM to deliver 2,070 FP4 TFLOPS of AI processing power. Thanks to its compact form-factor, the PE3000N mini PC enables developers and integrators to achieve new levels of autonomy, sensor fusion, and AI-driven control for industrial, commercial, and smart infrastructure deployments. Since it focuses on industrial use cases, this device features a rugged chassis compliant with MIL-STD-810H standards and can operate at temperatures ranging from -20°C up to 60°C.

Asus offers great configurability and versatility for the PE3000N, as it can be equipped with up to 4x 25 GbE NICs and support for 16 GMSL cameras, enabling high-bandwidth sensor fusion and advanced machine vision. On the I/O side, we find the usual desktop ports including 4X USB-A, 2x USB-C, and HDMI video out and audio jacks. Many industrial use cases are also supported via an optional second stack adding interfaces such as PoE, GMSL, CAN, and QSFP28. Other supported standards include PTP/PPS for precise sensor synchronization, onboard TPM 2.0 for robust security, and LTE/5G/GNSS modules for wide area connectivity.

Thanks to the advanced AI-processing capabilities of the Jetson IGX Thor T5000 module, the PE3000N provides support for efficient execution of generative AI models, such as visual language models (VLMs) and large language models (LLMs), as well as real-time video analytics and autonomous control tasks. Nvidia’s Jetson IGX Thor is fully compatible with physical AI applications like Nvidia Isaac for robotics, Nvidia Holoscan for real-time sensor processing and Nvidia Blueprint for video analytics AI agents.

No pricing info as of yet, but seeing that Nvidia’s own dev kit costs ~3,000 EUR, one could expect a similar price, possibly even higher with all the additional I/O options. Availability is scheduled for Q1 2026.