Fabless chips will enable local AI processing on devices, marking a first step toward broader automation goals.
Indian fabless semiconductor startups are increasingly targeting the video surveillance market with edge AI silicon, aiming to process artificial intelligence locally on devices and reduce reliance on cloud computing, industry sources said.
As reported by The Economic Times, Startups such as Netrasemi, BigEndian Semiconductors and Sensemi Technologies are developing AI/ML capable system-on-chips (SoCs) specifically for cameras and network video recorders (NVRs), enabling real-time analytics and inference directly on devices. This approach allows for efficient data processing, minimises latency, reduces backhaul costs, and improves reliability during operations.
Edge AI silicon refers to chips that run AI functions locally on devices. Firms are designing chips for the front layers of the surveillance value chain, from camera subsystems to application-specific chips and ultra-low-power sensor intelligence. By performing inference on the camera or sensor subsystem, only meaningful events are transmitted, making the system more efficient and cost-effective.
Sensemi, for instance, began with ultra-low-power sensing for health and wearable devices, later applying the technology to vision and surveillance. BigEndian is building application-specific integrated circuits (ASICs) for IP cameras, while Netrasemi focuses on designing AI/ML-enabled SoCs for edge devices.
The move towards edge AI chips is seen as a first step towards a broader vision of automated, intelligent surveillance networks. Industry executives note that while cameras and NVRs remain hardware-dependent, integrating AI on silicon enhances efficiency, reduces vulnerability to cyber and geopolitical risks, and paves the way for wider adoption of smart, autonomous monitoring systems in India.
