Back in November 2025, we wrote about the Brainchip AKD1500 PCIe/SPI Edge AI co-processor designed for low-power systems and delivering up to 800 GOPS at just 300 mW.
In recent months, the company has launched several hardware platforms based on the ADK1500. At the end of July, Brainchip introduced the AKD1500 M.2 B+M Key development card for $129, shortly followed by the $99 BrainBoard 1500 module in August, and a few days ago, the company started selling the AKD1500 PCIe development card for $149.
Brainchip AKD1500 M.2 B+M Key development card
- AI accelerator – AKD1500 neuromorphic co-processor
- Akida Neuron Fabric clocked at 5 to 400 MHz
- 1MB on-chip Local memory
- 800 GOPS at less than 300 mW
- Adaptive on-chip learning, no cloud required
- Package – 7×7 mm MFCTFBGA169 package, 0.5 mm pitch
- Process – 22 nm FD-SOI CMOS digital logic process
- Host Interface – B+M Key edge connector for PCIe or SPI connection to the host
- Misc – Fanless design
- Power Consumption – 250 mW typical at 400 MHz
- Dimensions – M.2 2230 module
- Temperature Range – 0 – 70°C
The M.2 module works with the Raspberry Pi 5 using an M.2 HAT, as well as other compatible hosts.
BrainBoard 1500 module
The BrainBoard 1500 module was developed in collaboration with Neuromorphux
- AI accelerator – AKD1500 neuromorphic co-processor
- Storage – 4 MB flash for Akida-compatible model deployment (8MB upon request)
- Host Interface – SPI / QSPI up to 60 MHz, I2C for configuration via through- and castellated holes
- Power profiling – Integrated two-channel power monitor for real-time inference measurement
- Misc
- Red/green power LEDs
- Blue talk-through LED
- I2C pull-ups and address via 0603 jumpers
- Fanless design
- Power Supply – 3.0 to 5V DC
- Power Consumption – AKD1500 sleep down to mW, AKD1500 OFF down to 37 uW
- Dimensions – 22.86 x 22.86 mm (Arduino Nicla-compatible footprint)
The BrainBoard 1500 works best with the Arduino Nicla Vision and Arduino Nicla Sense ME boards. The documentation also mentions a BB15 Nicla convenience board, an OpenMV AE3 adapter board, and a Seeed Studio Seeeduino adapter board, but I could not find details about the last three.
AKD1500 PCIe development card
- AI accelerator – AKD1500 neuromorphic co-processor
- Host Interface – PCIe 2.0 x1 edge connector
- USB – Micro USB port (undocumented, but likely for debugging)
- Misc – Fanless design
- Power Consumption – 250 mW typical at 400 MHz
- Dimensions/Form Factor – Standard PCIe card
- Temperature Range – 0 – 70°C
Software support and availability
The AKD1500 M.2 and PCIe development card follows the same development path as the earlier AKD1000 M.2/PCIe development cards. They work on Linux hosts (Debian/Ubuntu), and after installing the drivers, the creation, training, and testing of neural networks is done through the MetaTF development environment. It supports TensorFlow and Keras for neural network development and training, and includes a model zoo of pre-trained models, as well as tools to convert CNN models to SNN models.
The BrainBoard 1500 is a little different since its development workflow follows that of the Arduino Nicla microcontroller boards. Neuromorphyx provides an open-Nicla Sense, and a model-flashing workflow on GitHub
You can access documentation and purchase the AKD1500 PCIe and M.2 development cards, the BrainBoard 1500 module, or a pack of 5 AKD1500 chips ($199.99) on the Brainchip shop.
Jean-Luc started CNX Software in 2010 as a part-time endeavor, before quitting his job as a software engineering manager, and starting to write daily news, and reviews full time later in 2011.
Support CNX Software! Donateods on Amazon or Aliexpress. We also use affiliate links in articles to earn commissions if you make a purchase after clicking on those links
