Mentium develops co-processors that utilize hybrid in-memory and digital computation to deliver cloud-quality AI inference at ultra-low power for mission-critical applications on the ground and in space. Their technology addresses the need for reliable and efficient AI processing in environments where performance and power consumption are critical, achieving 100 times the speed and 50 times the efficiency of current solutions without requiring external memory.
Funding
Funding not disclosed
Founders
Product
Problem
Many mission-critical applications, especially in edge computing scenarios, require reliable and efficient AI inference but are constrained by power consumption and the need for external memory. Existing AI processing solutions often lack the speed and efficiency required for these demanding environments.
Solution
Mentium develops co-processors that leverage hybrid in-memory and digital computation to deliver cloud-quality AI inference at ultra-low power. Their AIM (AI in Memory) technology achieves significant improvements in both speed and efficiency compared to current state-of-the-art solutions, without compressing DNN models or requiring external memory. This bolt-on solution enhances existing systems, providing dependable AI at the edge for complex and critical applications.
Target Audience
The primary target audience includes organizations requiring AI solutions for mission-critical applications in environments with limited power and space, such as aerospace, defense, and industrial automation.
Features
- Hybrid in-memory and digital computation architecture
- 100x the speed and 50x efficiency of current solutions
- No external memory required
- Compatible with existing systems as a bolt-on solution
- Designed for reliable AI performance in mission-critical applications
- Suitable for smart security cameras, space edge computing, and robotics