There is a quiet, exhausting reality at the absolute edge of the modern physical world. We are aggressively deploying millions of smart cameras, industrial sensors, and autonomous drones, expecting them to make split-second, intelligent decisions. But the architecture we rely on is fundamentally flawed. We force these devices to stream massive amounts of raw video and sensor data back to a centralized cloud just to run a basic AI model, creating agonizing latency, crippling bandwidth costs, and severe security vulnerabilities. If a robotic arm on an assembly line or an autonomous vehicle has to wait for a server round-trip to decide its next move, the system fails.
It takes builders with profound semiconductor engineering rigor and the ambition to fundamentally rewrite how physical devices compute to solve this at the root. That is exactly what Jyothis Indirabhai, Deepa Geetha, and Sreejith Varma are doing with Netrasemi.
Founded in 2020, the Kerala-based deep-tech team is building a deeply intelligent, application-specific semiconductor platform that moves advanced AI processing completely out of the cloud and directly onto the edge device.
Netrasemi designs ultra-power-efficient Edge AI System-on-Chips (SoCs) featuring proprietary deep-neural AI acceleration cores (NPUs). Instead of relying on power-hungry, off-the-shelf GPUs, their domain-specific architecture (DSA) and modular chiplet designs allow heavy AI workloads — like real-time facial analytics, sensor fusion, and predictive maintenance — to run directly on the motherboard. To bridge the gap between silicon and deployment, they provide the Netra Smart Edge Studio, a comprehensive software suite that allows hardware teams to actually deploy complex AI models onto the chip effortlessly.
The true moat here is uncompromising silicon ownership paired with real-time autonomy. Netrasemi doesn’t just assemble generic parts; they own the foundational Intellectual Property (IP) inside the silicon. By completely abstracting away the cloud dependency, they eliminate latency and reduce massive power consumption. It shifts the paradigm from retroactive, centralized AI processing to deterministic, instant intelligence right at the moment of data capture.
The market? Deep-tech manufacturers building the next generation of smart surveillance networks, industrial robotics, drones, and autonomous vehicles who desperately need high-performance AI computing without breaking their thermal or power budgets.
Seeing homegrown founders head-down, quietly architecting the heavy-duty physical silicon that will power the autonomous infrastructure of the future, is profoundly inspiring. They are giving edge devices their autonomy and their speed back.
Let’s celebrate the builders.
✍️ Jay Ingle
#EdgeAI #DeepTech #ProductNation