Leuven-based Edgise is going to develop custom hardware for AI computing at the edge. According to the company, the latest member of the Raccoons group of technology start-ups, this approach has several advantages over doing it in the cloud, including speed, latency and privacy benefits. Depending on project requirements, Edgise plans to use commercially available chips or program its own FPGAs.
AI computing in the cloud requires large quantities of data to be transferred across the network. By moving the intelligence to the edge, Edgise aims to shorten response times, allowing for faster applications, and reduce privacy risks, among other things. “We’re developing hardware that enables running the AI directly on the devices in the field and making decisions locally,” explains co-founder Nick Destrycker.
Edgise is the seventh venture of the Raccoons group. Other family members include Brainjar (advanced AI and machine learning applications), Craftworkz (innovation advice and software prototyping), Theledger (blockchain technology) and QNTM (quantum computing).