Verkko Robotics Unveils VOLTAIC AI Engine
European deep-tech lab Verkko Robotics has launched VOLTAIC, an AI engine designed for continuous learning. The system aims to significantly reduce the cost and energy demands of current frontier AI models.
Verkko Robotics, a European deep-tech AI research lab, has introduced VOLTAIC, a new neural inference engine. The system is engineered for continuous learning, resistance to catastrophic forgetting, and substantially lower operational costs and energy consumption compared to existing large AI models.
The VOLTAIC engine utilizes a sparse network architecture, meaning only a fraction of neurons are active at any given time. This design drastically reduces power demands, making it suitable for deployment on private, localized hardware. The company claims that its internal benchmarks show VOLTAIC costs approximately 1/50th per query compared to comparable queries on leading large language models.
In independent benchmarks for continual learning, VOLTAIC reportedly achieved a catastrophic forgetting rate of approximately 1.2% and 77% accuracy on Split ImageNet-1K. According to Verkko Robotics, these results position the engine at the forefront of its class, achieved with a fraction of the energy cost of conventional transformer-based systems.
The launch comes at a time of increasing European focus on technological sovereignty. By enabling processing on local hardware and reducing reliance on large, external cloud infrastructure, VOLTAIC aims to address concerns regarding data jurisdiction and dependence on foreign AI providers. The company plans a full release for consumers and enterprises in Q4 2026.