The world's first commercial neuromorphic microcontroller — now in the hands of Indian engineers. Not in a research lab. In a classroom. On a workbench. In a startup. This is what the partnership exists to do.
Innatera is a Dutch semiconductor company spun out of Delft University of Technology in 2018. They build neuromorphic processors — chips that mimic how the brain processes sensory input. Instead of moving streams of data through conventional digital logic, their chips process only meaningful signals, reacting to changes the way neurons do. The result: a processor that makes decisions at the edge — in a wearable, a sensor, an industrial device — in real time, consuming almost no power.
Their flagship product, Pulsar, launched May 2025, is the world's first commercial neuromorphic microcontroller. Hybrid analog-digital architecture. RISC-V CPU. CNN and FFT accelerators. Talamo SDK that maps directly from PyTorch and TensorFlow. The chip that makes neuromorphic computing accessible to engineers who are not neuroscientists.
Innatera also has a significant presence in Bengaluru — making India not just a market for this technology, but a genuine part of where it is being built.
Neuromorphic computing has existed in research labs for decades. Intel built Loihi. IBM built NorthPole. But none of it reached the engineer who just graduated from an engineering college in Pune or Coimbatore. The gap between research-grade neuromorphic hardware and the working engineer has never been bridged — not in the US, not in Europe, and certainly not in India.
VLSI EXPERT chose to bridge that gap. Not because it was the easy next step, but because it is the right next step. The India Semiconductor Mission is creating demand for engineers who understand next-generation chip architectures. The industry needs people who can design for spiking neural networks, event-driven computing, and ultra-low-power edge AI — not five years from now, but today. Every engineer we train on Pulsar is one less gap in India's semiconductor workforce.
Three distinct programs — each designed for a different kind of engagement with neuromorphic technology. All built around the Innatera Pulsar EVK and Talamo SDK.
Conventional AI chips process data continuously — every clock cycle, whether anything meaningful is happening or not. This is wasteful. The brain doesn't work that way. It only processes information when there is a meaningful change — a sound, a movement, a signal. Spiking neural networks (SNNs) mirror this: event-driven, sparse, efficient.
Pulsar puts this into silicon. The result is a chip that can run always-on AI at the sensor level — in a wearable, an industrial monitor, a smart home device — on microwatts of power, with sub-millisecond response times. No cloud. No latency. No privacy concern.
The Talamo SDK removes the last remaining barrier: it lets developers map workloads directly from PyTorch and TensorFlow onto Pulsar. No PhD in neuroscience required.
Deep dive — Products page →Neuromorphic computing needs to be embedded in India's academic fabric — not just taught in training centres. These collaborations are how that happens.