
Mentium
Mentium develops the technology to bring ultra-fast AI to low power electronics, on ground, air, and space.
Date | Investors | Amount | Round |
---|---|---|---|
- | investor investor investor | €0.0 | round |
investor | €0.0 | round | |
investor | €0.0 | round | |
N/A | Seed | ||
Total Funding | 000k |
Related Content
Mentium Technologies Inc., a semiconductor company based in Goleta, California, specializes in developing high-efficiency AI chips for edge computing. Founded in February 2017, the company's mission is to enable complex AI computations in power-constrained environments. The founding team includes CEO Mirko Prezioso, CTO Farnood Merrikh Bayat, Dimitri B. Strukov, and Konstantin Likharev. Prezioso, who holds a Ph.D. from Parma University, has a background in condensed matter physics and experience with memristive systems from his time at the University of California, Santa Barbara. This academic and research foundation, particularly work on in-memory computing under a DARPA-funded project, directly informs the company's technological approach.
The core of Mentium's business is the development and sale of AI co-processors designed to perform sophisticated video and data analytics in low-power settings. The company targets original equipment manufacturers (OEMs) and system integrators in mission-critical sectors such as space, robotics, security, and aerospace. Mentium's business model revolves around selling its proprietary AI-Memory (AIM) chips, which act as bolt-on solutions for existing systems, enhancing them with cloud-quality AI inference capabilities without requiring a complete system overhaul. This approach helps clients avoid the high latency, cost, and security issues associated with cloud-based AI processing.
Mentium's flagship product is its AIM co-processor, a hybrid analog-digital chip that enables large deep neural network (DNN) models to run directly on edge devices. This technology uses a unique architecture based on in-memory computing with non-volatile memory, which overcomes the data transfer bottleneck that typically slows down computation. The key benefit is the ability to deliver high-precision AI inference at ultra-low power consumption (under 0.5w), offering significant improvements in speed and efficiency compared to other edge AI semiconductors. This allows for rapid, real-time decision-making in applications where immediate response is crucial. A significant milestone for the company includes multiple contracts with NASA under the Small Business Innovation Research (SBIR) program to develop and test radiation-hardened neuromorphic chips for space applications, including lunar exploration. In May 2025, Mentium successfully completed heavy ions radiation testing at the Berkeley Accelerator Space Effects (BASE) Facility, a critical step for its space-focused applications.
Keywords: edge AI, AI co-processor, neuromorphic computing, in-memory computing, low-power electronics, semiconductor, AI chip, video analytics, space technology, radiation-hardened electronics, robotics AI, security cameras, AI accelerator, deep neural networks, system on a chip (SoC), NASA SBIR, analog computation, AI inference, mission-critical AI, situational awareness