Embedded Insiders Spiking Intelligence: How Neuromorphic Computing Enables Brain-Inspired AI
Jan 29, 2026
Sumeet Kumar, CEO of Innatera Nanosystems, builds energy-efficient neuromorphic chips like Pulsar for edge sensors. Paul Golden, Marketing Director at the Wireless Power Consortium, leads Qi/Ki wireless power standards and certification. They discuss spiking neural networks, hardware-software co-design, temporal sensing at the edge, and recent advances in wireless charging standards and applications.
AI Snips
Chapters
Transcript
Episode notes
Temporal Encoding Powers Efficiency
- Spiking neural networks encode information in sparse voltage spikes where timing and distribution carry meaning.
- This temporal encoding lets SNNs be smaller, faster, and more energy-efficient than conventional networks.
Zebrafish Brain Demonstrates Sparse Spikes
- Sumeet showed a zebrafish brain video to illustrate clustered, sparse neural activity during behavior.
- The example highlighted how repeated stimuli evoke roughly the same neural clusters over time.
Process Data At The Sensor
- Use spiking accelerators at the sensor to tokenize and filter raw data before sending it upstream.
- Combine spiking stages for time-series tasks with non-spiking models for spatial or large-scale inference.