The AKD1500 is BrainChip's Akida™ neuromorphic edge AI co-processor — event-based, purely digital, with built-in on-chip learning. It pairs with any host CPU or MCU via PCIe or SPI, enabling ultra-low-power AI inference for IoT, Industrial, Automotive, Healthcare, and Smart Home applications.
What's in the box
- AKD1500 Edge AI Co-Processor IC (7×7mm MFCTFBGA169 package, 0.5mm pitch)
Specifications
- Architecture: Akida neuromorphic co-processor
- Process: 22nm FD-SOI CMOS digital logic
- Package: 7×7mm MFCTFBGA169, 0.5mm pitch
- Peak performance: 800 effective GOPS at <1mW/GOPS
- On-chip memory: 1MB SRAM
- Host interface: PCIe Gen2 endpoint; SPI S/D/Q/O peripheral
- Memory expansion: SPI D/Q/O interface
- Clock frequency: 5–400 MHz
- Power dissipation: 250mW typical at 400MHz
- On-chip learning: Yes (no cloud required)
- Software: MetaTF (TensorFlow/Keras, PyTorch APIs)
Designed for
- Edge AI Vision Systems: ADAS/AV, Robotics, Drones, Surveillance
- Industrial IoT & Predictive Maintenance: Ultra-low-power sensor fusion
- Smart Home, Appliances & Voice Control: Always-on keyword spotting
- On-Device Personalized Learning: Secure on-chip adaptation, no cloud
