We take cryptocurrency — so the coins you hold can buy real hardware.Pay in crypto, spend it on real hardware. Free insured shipping worldwide over $399. Plain, unbranded boxes, sent from Germany.Free insured shipping over $399. Your rate is locked for 30 minutes once checkout opens.Rate locked 30 minutes. Every item sealed, serial-checked and covered by a 24-month warranty.Sealed, 24-month warranty.

Saved items Sign in
Black M.2 2280 AI module with gold edge connector and 8GB LPDDR4 memory

ASUS

ASUS 90YT00B1-M0AA00 UGen300 M2 AI Module, Hailo-10H, 8GB LPDDR4

SKU CH-9XRW-NCSBP MPN 90YT00B1-M0AA00

M.2 Key-M 2280 AI accelerator with Hailo-10H chipset delivering 40 TOPS at 2.5 watts and 8GB LPDDR4 memory

  • TypeAI Module
  • ColorBlack
  • SpecificationsChipset: Hailo-10H AI Performance: 40 TOPS @ INT4 | 20 TOPS

Option UGen300 M2 AI Module

  • Up to 40 TOPS INT4 inference at only 2.5 W typical power
  • 8 GB LPDDR4 onboard memory enables fully offline AI processing
  • M.2 Key-M 2280 PCIe 3.0 x4 form factor fits compact edge systems
  • Runs on Windows, Linux and Android across x86 and ARM architectures
  • Access to 150 plus pre-trained models for vision, LLM and VLM workloads

ASUS UGen300 M2 AI Module

The ASUS UGen300 M2 AI Module is an edge accelerator in M.2 2280 Key-M form factor. It targets developers and integrators who need local inference on x86 or ARM platforms running Windows, Linux or Android.

40 TOPS INT4 and 8 GB LPDDR4

The Hailo-10H chipset delivers 40 TOPS at INT4 and 20 TOPS at INT8 within a 2.5 W typical envelope. Eight gigabytes of LPDDR4 at 4266 MT/s feed the engine, so larger vision or language models fit on-device until memory or compute ceiling is reached.

Model zoo and framework support

Access to 150-plus pre-trained models covers LLM, VLM, Whisper and vision networks. Native TensorFlow, TensorFlow Lite, Keras, PyTorch and ONNX compatibility lets existing pipelines deploy without rewrite.

Highlights

  • Up to 40 TOPS INT4 inference at only 2.5 W typical power
  • 8 GB LPDDR4 onboard memory enables fully offline AI processing
  • M.2 Key-M 2280 PCIe 3.0 x4 form factor fits compact edge systems
  • Runs on Windows, Linux and Android across x86 and ARM architectures
  • Access to 150 plus pre-trained models for vision, LLM and VLM workloads

Specifications

BrandASUS
ModelUGEN300-8GLD4-M2
TypeAI Module
ColorBlack
SpecificationsChipset: Hailo-10H
AI Performance: 40 TOPS @ INT4 | 20 TOPS @ INT8
Interface: PCIe3.0 x 4 lanes, M2 2280 Key-M
Memory: LPDDR 8GB
Memory Speed: 4266 MT/s
Power Consumption: 2.5 Watts (Typical)
Supported OS: Windows, Linux, Android
Supported Host Architectures: x86, ARM
Supported AI Framework: Keras, TensorFlow, TensorFlow Lite, PyTorch, ONNX
Supported AI Models: Vision Models (> 150+),
GenAI Models (LLM, VLM, Whisper)
Package Contents: 1 x Quick Start Guide
1 x Type-C to Type-C cable, white, 20 cm
Operating Temperature: 0°C ~ +40°C
Operating Humidity, RH: 20% ~ 85%, non-condensing
Sto
Shipping weight0.09 kg
Package size140 × 89 × 38 mm

Other versions of this model

1 more in the same range

Same family, different memory, cooler or clock — compare before you commit.

Questions about this item

Will the M.2 Key-M 2280 slot and PCIe 3.0 x4 interface work with my older motherboard that only has PCIe 2.0 x4?

The module uses a PCIe 3.0 x4 interface on an M.2 Key-M 2280 connector. It will physically fit and negotiate down to PCIe 2.0 speeds on older boards, but AI throughput will be limited by the reduced lane bandwidth.

How can I verify the Hailo-10H is delivering the rated 40 TOPS at INT4 after installation?

Run a benchmark from the Hailo model zoo — such as a quantised LLM or vision network — and compare the reported inference latency against the published 40 TOPS INT4 figure; the supplied software stack includes profiling tools for this check.

Does the box include a heatsink or mounting screw for the M.2 2280 slot, or must I buy them separately?

The module ships with 8 GB LPDDR4 memory soldered on board; a heatsink and M.2 mounting screw are not included and must be sourced separately to suit your chassis and motherboard standoff height.

New to this? Which coin should you pay with — what to check, what it costs to pay, and what is never asked for.

Also in this aisle

People compared these

Same shelf, same checkout — eight coins and a 30-minute rate lock.