
Mentium Technologies provides ultra-fast, reliable AI inference at the edge for mission-critical use cases. It builds ultra-low-power AI co-processors that deliver cloud-quality inference using AIM, a hybrid in-memory and digital computation architecture. The AIM approach eliminates the need for external memory while achieving the company's reported 100x speed and 50x efficiency improvements versus current solutions. Mentium's co-processors are intended for integration into edge electronics across sectors such as smart security, robotics, and aerospace. The product is hardware-focused and targets companies that require high accuracy, low-latency, and low-power AI inference.

Mentium Technologies provides ultra-fast, reliable AI inference at the edge for mission-critical use cases. It builds ultra-low-power AI co-processors that deliver cloud-quality inference using AIM, a hybrid in-memory and digital computation architecture. The AIM approach eliminates the need for external memory while achieving the company's reported 100x speed and 50x efficiency improvements versus current solutions. Mentium's co-processors are intended for integration into edge electronics across sectors such as smart security, robotics, and aerospace. The product is hardware-focused and targets companies that require high accuracy, low-latency, and low-power AI inference.
Founded: 2017
Headquarters: Santa Barbara / Goleta, California
Product: Ultra-low-power, radiation-tolerant AI co-processors for edge/mission-critical use
Notable partners/customers: NASA (collaborations and testing support)
Latest reported funding: Seed round reported Aug 1, 2023
Edge AI inference for mission-critical and space applications (radiation-tolerant, SWaP-constrained systems).
2017
DeepTech
“Alumni Ventures; Unpopular Ventures; NASA; National Science Foundation; Santa Barbara Angel Alliance; Silicon Catalyst; Terra VC; Unpopular Ventures”