UGREEN MasterAgent MA100 — NVIDIA Jetson Thor T5000 Local AI Hub

Specifications

Overview

The UGREEN MasterAgent MA100 is the clearest sign yet that NVIDIA's Jetson Thor silicon has crossed from industrial robotics into consumer hardware. It carries a Jetson Thor T5000 module — a 14-core Arm Neoverse V3AE CPU paired with a Blackwell GPU and 96 Tensor Cores delivering up to 2,070 FP4 TFLOPS — inside a cubic hub designed to sit on a shelf next to a router.

UGREEN's justification for that much compute in a smart-home device is agentic AI: running larger models locally, understanding richer context and intent, and coordinating multiple devices and tasks at once without shipping audio, video or sensor data to a cloud service. The MA100 acts as a Matter controller with a built-in Thread border router, and combines those roles with NAS-style storage.

The line scales down through the HomeAgent HA100 (26 TOPS, 16 GB DDR5) and HomeAgent HA100 Pro (205 TOPS) on RK3588 and CIX-P1 silicon respectively. Because the top tier is Armv9 Jetson Thor, models and applications validated on the MA100 share a toolchain with industrial Jetson Thor edge systems — and UGREEN exposes developer packages and skills so integrators can extend it rather than treat it as a fixed appliance.

Key Benefits

2,070 FP4 TFLOPS and 128 GB of unified memory in a non-rack device is enough to serve multi-billion-parameter models locally. Matter plus Thread border router means the same box that runs inference also controls the building. Shared Armv9 Jetson Thor stack lets code move between consumer hubs and industrial edge systems without rewriting.

Applications

Private, on-premises AI assistants and agents, local video and sensor analytics across cameras, smart-building and home automation control, edge model development and prototyping, and small-office inference where data cannot leave the premises.

Related Products