All News | Boards | Chips | Devices | Software | LinuxDevices.com Archive | About | Contact | Subscribe
Follow LinuxGizmos:
Twitter Facebook Pinterest RSS feed
*   get email updates   *

Forlinx FAI-ARA240-M Packs Ara240 NPU into M.2 2280 Module

Apr 13, 2026 — by Giorgio Mendoza 6,286 views

Forlinx Embedded has officially launched the FAI-ARA240-M, an M.2-based AI accelerator built around the NXP Ara240 processor. The module provides a discrete NPU for offloading inference workloads from embedded host systems.

The Ara240 processor was first seen during the launch of the OK-MX9596-C. Forlinx provides additional details with this module, which delivers up to 40 TOPS of AI performance and supports a range of model types, including convolutional neural networks, transformer-based models, and multimodal architectures.

— ADVERTISEMENT —



OK-MX9596-C SBC
(click image to enlarge)

The accelerator is available with 8GB or 16GB of LPDDR4 memory for handling larger models and high-throughput workloads.

The module adopts a standard M.2 2280 form factor with an M-Key interface, allowing integration into existing platforms through PCIe without requiring changes to the base hardware.


FAI-ARA240-M dimensions
(click image to enlarge)

It supports PCIe Gen4 x4 and USB 3.2 Gen1 interfaces for data transfer between the host processor and the accelerator.

In this configuration, the FAI-ARA240-M operates as a co-processor, where the host system manages control and application logic while the accelerator handles AI inference tasks.


FAI-ARA240-M cooling system
(click image to enlarge)

The module is compatible with host platforms such as those based on NXP i.MX8M Plus and i.MX95 processors.

Software support includes compatibility with TensorFlow, PyTorch, and ONNX, along with development tools for model deployment, quantization, and optimization. The platform supports multiple data types, including INT4, INT8, and mixed-precision formats.


FAI-ARA240-M
(click image to enlarge)

The design includes secure boot and root-of-trust features, along with a thermal solution intended to maintain stable operation under sustained workloads.

Forlinx states that the module is specified for continuous operation and has undergone environmental validation for industrial use.

Specifications listed for the FAI-ARA240-M include:

  • Processor:
    • NXP Ara-240 Edge AI processor
  • AI Performance:
    • Up to 40 TOPS
  • Memory:
    • 8GB / 16GB LPDDR4
  • Form Factor:
    • M.2 2280 (M-Key)
  • Interface:
    • PCIe Gen4 x4
    • USB 3.2 Gen1
  • Mechanical:
    • 80 x 22mm

Further Information

Forlinx indicates the FAI-ARA240-M is available for order, although pricing has not been disclosed.

(advertise here)


PLEASE COMMENT BELOW

2 responses to “Forlinx FAI-ARA240-M Packs Ara240 NPU into M.2 2280 Module”

  1. Derek says:

    How can this be implemented into a gaming PC for use with AI gaming and upscaleing? There has to be some kind of benefits to using this module for gaming. This could be a great addition for lower end gaming pcs that could use some extra power to upscale. Just a thought but I’m sure many others have the same idea while reading this.

  2. Kelvin W says:

    I do wonder if could be used with a Thunderbolt3/USB4 to NVMe adapter.. running 2 of these with a Strix Halo would be insane!

Please comment here...