BrainChip launched the AKD1500 Raspberry Pi 5 Development Kit on 29 October 2026. It puts the company's Akida AKD1500 chip on an M.2 card attached to a Raspberry Pi 5, so developers can try event-based, always-on AI on a board they probably already know. The kit is on sale now in BrainChip's online shop for $1,295, excluding tax.

What is in the box

The kit contains:

  • a Raspberry Pi 5 with 8 GB of RAM
  • an AKD1500 M.2 card (B+M key), already fitted
  • a 32 GB microSD card with Akida demos and BrainChip's MetaTF tools preinstalled
  • a USB-C power supply

Setup happens in a browser. After power-on, the kit creates its own Wi-Fi network called akida-devkit and serves a web page at http://10.42.0.1. You can also plug it into a router over Ethernet and open the address your DHCP server assigns. The shop page lists visual object classification, anomaly detection and deploying models from Edge Impulse as the intended uses.

Why an event-based chip

BrainChip's pitch is about devices that have to watch, listen or measure all the time: sensors, monitors, wearables and, in the company's own list, smart home products. A conventional processor analyses every frame and every sample, even when nothing has changed. That drains batteries and adds heat. Akida is a neuromorphic design that processes only the events that matter, and everything runs locally, so no audio or video has to leave the device.

The other selling point is on-chip learning. According to BrainChip, an application can learn a new keyword, gesture or object in the field without retraining in the cloud, and the data used for it stays on the device. For a privacy-minded smart home that is the interesting part: a local wake word or a new object class without sending samples anywhere.

From prototype to product

BrainChip says the AKD1500 on the kit is the same chip companies design into production hardware, over PCIe or SPI. The kit joins the AKD1500 PCIe card and AKD1500 M.2 board. Anyone who already has a Raspberry Pi 5 with an M.2 HAT can buy only the M.2 card and use the same software. MetaTF tools and models are free through the BrainChip Developer Hub, and the company says support for the kit will include Docker images.

Worth knowing

At $1,295 this is not a hobby accessory. The Raspberry Pi 5 inside costs a fraction of that, so most of the price is the AKD1500 card and BrainChip's software stack. The kit makes most sense for teams evaluating low-power vision or audio detection for a product. BrainChip has not published power figures for the kit itself or examples tied to Home Assistant or other home platforms.