The neural processor Al IP developed by BrainChip is an event-based technology that is inherently lower power when compared to conventional neural network accelerators. The IP supports incremental learning and high-speed inference in a wide variety of use cases, such as convolutional neural networks with high throughput and unsurpassed performance in low power budgets. The AKD1000-powered boards can be plugged into the M.2 slot — around the size of a stick of gum, with a power budget of about 1 W — to unlock capabilities for a wide array of edge AI applications where space and power is limited and speed is critical, including industrial, factory service centers, network access devices and more.

“BrainChip’s AKD1000 chips and boards are available today for industry evaluation, development, proof of concept and demonstration platforms with the IP available to license for integration into SoCs. Releasing the AKD1000 on the M.2 form factor continues our commitment to aid developers in creating AI systems with our Akida IP,” said Sean Hehir, BrainChip CEO. “By providing neuromorphic developers access to Akida via M.2, we can expand their options for proof-of-concept designs with streaming sensors at the edge.”

The AKD1000 product is available in both B+M Key and E Key configurations of the M.2 2260 form factor. It can be purchased integrated into stand-alone Raspberry PI or Edge AI box enclosures, or for integration into custom designed products.