Verkko Robotics unveils VOLTAIC, a low-energy spiking neural inference model

• •London-based Verkko Robotics released VOLTAIC, an AI system that claims to enable continuous learning while reducing energy consumption.
• •The model utilizes spiking neural network architecture to perform inference with significantly lower compute requirements than standard multimodal models.
• •Independent verification of these efficiency claims is currently missing, and it is unclear how the system scales when compared to high-parameter LLMs.

Verkko Robotics has launched VOLTAIC, an AI system designed for continuous learning and low-power inference. Unlike current energy-intensive multimodal models that often suffer from catastrophic forgetting, this architecture uses spiking neural networks to retain information while minimizing power usage, according to EU-Startups. However, the practical constraints of the model remain unproven outside of controlled laboratory settings. Whether VOLTAIC can bridge the gap between niche robotics applications and general-purpose enterprise computing will depend on its ability to maintain accuracy at scale.