Get Christian Pehle on Dwarkesh Podcast

AI is running headlong into an energy and cost bottleneck, and Christian Pehle is one of the few researchers who can show, with working code and proofs, how spiking neural networks can break that wall today, from exact-gradient training with EventProp to the production-ready Norse library. His EventProp work established exact gradients for spiking networks, and Norse brings those ideas to PyTorch so practitioners can actually use them now.

He also helped build Heidelberg’s BrainScaleS‑2 neuromorphic platform, which recently added a real-time spike interface enabling thousandfold-accelerated control for high-speed robotics, a timely path to efficient agents and edge AI. Put that in front of Dwarkesh’s audience of builders who just heard deep brain-meets-AI conversations, and you get a rare, concrete roadmap for scaling useful intelligence without blowing out power budgets.

Pehle has not been on the Dwarkesh Podcast before, so this would be a first that lands at exactly the right moment for the industry to pay attention. Support this nomination today.