BrainChip Holdings Ltd. (ASX: BRN, OTCQX: BRCHF) said Monday it has begun shipping production quantities of its AKD1500 neuromorphic processor to multiple customers in defense and wearable applications.
The fully digital, event-based chip is now commercially available in both packaged silicon and bare-die formats while it completes comprehensive industrial and military-grade qualification. It functions as a standalone processor or hardware co-processor for edge AI workloads that demand sub-watt power budgets and cloud-independent operation.
AKD1500 Production Specifications
Performance: Up to 800 effective GOPS with near-terabyte operations-per-second scale efficiency
Power Consumption: <300 mW in PCIe mode; <200 mW in serial mode
Architecture: Neuromorphic, event-driven processing with on-chip learning
Interfaces: Dual — PCIe for high-speed co-processing alongside x86, ARM or RISC-V hosts; SPI/I2C serial for direct integration with low-power microcontrollers
Form Factors: Standard packaged silicon for high-reliability boards or bare die for space-constrained and custom multi-chip modules** Process Node**: GlobalFoundries 22nm FDX
Key Differentiator: Processes data only when events occur, eliminating the always-on power draw typical of conventional AI accelerators
“ The commercial availability of the AKD1500 production chip marks a profound milestone in our commercialization roadmap,” BrainChip CEO Sean Hehir said in a statement. “By delivering true, sub-watt neuromorphic compute in both packaged and die formats, we are providing industrial and defense partners with the rugged hardware flexibility they need to deploy intelligence anywhere — completely free from cloud dependence.”
The AKD1500 targets industrial IoT, automotive, aerospace, defense, and wearable markets where continuous inference must occur locally with minimal power, heat, and latency. Its dual-interface design allows system designers to upgrade existing multi-processor platforms or add advanced pattern recognition and local learning to battery-constrained embedded devices without full redesigns.
Production shipments follow earlier design wins, including an initial order framework with Parsons subsidiary Blue Ridge Envisioneering for defense platforms and selection by Neuromorphyx for its Vision NeuroNode edge-AI devices. The move to volume production validates BrainChip’s Akida neuromorphic platform after years of development and positions the company to capture demand in sovereign, always-on edge AI deployments where traditional GPU or NPU solutions remain power- or thermally constrained.
For high-intent readers tracking edge AI infrastructure, the AKD1500 represents one of the first commercially shipping neuromorphic co-processors purpose-built for real-world, disconnected operation in high-reliability verticals.