BrainBoard1500 with AKD1500 Neuromorphic AI on an Arduino-Compatible Development Board

Related Vendors

Taiwan Machine Tool & Accessory Builders' Association (TMBA)

SIMTOS (Seoul International Machine Tool Show)

Brainchip and Neuromorphyx bring the neuromorphic AI accelerator AKD1500 to a compact development board. The BrainBoard1500 is designed to enable developers to test Spiking Neural Networks directly with sensor and MCU hardware—including power consumption measurement.

The companies Brainchip and Neuromorphyx introduced the BrainBoard1500, a developer and evaluation board for neuromorphic edge AI, on August 18, 2026. At its core is Brainchip's Neural Network Accelerator AKD1500. Neuromorphyx develops, manufactures, and distributes the board.

Unlike larger evaluation platforms with M.2 modules or PCIe cards, the BrainBoard1500 is directly aimed at embedded applications. It uses the form factor of the Arduino Nicla product family and is compatible with its connectors. An open-source library and corresponding drivers are intended to facilitate integration into existing development environments.

Sensor data can be transferred to the AKD1500 via SPI or QSPI and processed there using Spiking Neural Networks. Additionally, I²C is available for system control. Models can be stored in a flash memory on the board.

Directly Measure the Power Consumption of AI Inference

Interesting for the evaluation of edge AI applications is an integrated dual-channel current monitor. This allows the power consumption of the AKD1500 to be recorded during various inference workloads. Developers can also examine different sleep and wake configurations.

This is precisely where the concept differs from host-centered evaluation boards: neuromorphic processing can be studied under the energy, latency, and space conditions that will later apply to a compact embedded system. The architecture is also not limited to Arduino hardware. The board is designed to connect with additional MCU and sensor platforms via the SPI/QSPI interface. Brainchip and Neuromorphyx specifically mention Raspberry Pi, Espressif, OpenMV, Seeed Studio/XIAO, and STMicroelectronics, as well as custom MCU- or FPGA-based hosts.

Neuromorphic Processing for Always-On Systems

The AKD1500 is based on Brainchip's Akida architecture. Instead of continuously processing neural networks following the classic principle, the architecture is designed for event-based processing. This specifically targets edge applications where power consumption and latency are limited.

Brainchip recently made the AKD1500 available as an M.2 module. This module is intended for integration into existing industrial and commercial systems and, due to its low power consumption, is designed to enable fanless designs. In contrast, the BrainBoard1500 takes a more embedded approach: the accelerator can be directly combined with microcontrollers and sensor hardware. Neuromorphyx also already uses the AKD1500 in its own neuromorphic vision systems. There, the company combines event-based sensors, FPGA pre-processing, and neuromorphic inference for applications in robotics and industrial automation, among others.