MemryX’s MX3 M.2 AI Accelerator Module has been awarded the 2025 Edge AI and Vision Product of the Year Award in the Edge AI Computers and Boards category.
The MemryX MX3 M.2 AI Accelerator delivers AI model-based computer vision processing with ultra-low power consumption averaging under 3W for multiple camera applications.
The MX3 is based on an advanced on-chip memory architecture that reduces data movement, boosting efficiency and reducing power and cost. 16-bit inference processing delivers high accuracy without the need for retraining or hand-tuning. Model compilation for MX3 is straightforward – the MemryX software stack eases deployment, eliminating the need for deep hardware expertise. Thousands of computer vision models have been directly compiled with no intervention, shortening development cycles and speeding up time-to-market. The M.2 form factor enables quick integration into existing platforms with minimal thermal concerns.
Developers can import models directly from popular frameworks like TensorFlow or PyTorch, and the MemryX compiler automates optimizations such as quantization and layer fusion. These tools can even run on resource-constrained devices like the Raspberry Pi, enabling cost-effective development and testing. This streamlined workflow eliminates the need for deep hardware expertise, significantly reducing development time and complexity. The MX3 hardware offers an innovative approach to scaling. It allows MX3 devices to be daisy-chained to add capacity for large models while also allowing fewer devices to be deployed to reduce cost and power when performance requirements are lower.
“On behalf of everyone at MemryX, we’re honored to be selected as a winner of the 2025 Edge AI and Vision Product of the Year Award. At MemryX, we’ve dedicated ourselves to delivering a production-ready Edge AI solution for computer vision that combines exceptional power efficiency and performance, while enabling rapid and broad deployment—from hobbyists and entrepreneurs to multinational corporations,” said Keith Kressin, CEO of MemryX
MemryX empowers edge AI applications with high-performance, energy-efficient inference solutions designed for simplicity. Our AI inference processors, featuring a unique at-memory compute architecture, deliver exceptional performance and minimize power consumption and space, making them perfect for edge devices. Our comprehensive software platform simplifies deploying complex AI models, abstracting away hardware complexities. Using MemryX, developers can easily integrate AI into applications like vision, robotics, and industrial automation, where low latency, power efficiency, and ease of use are critical.
The award trophy will be presented at the 2025 Embedded Vision Summit May 20-22 in Santa Clara, CA. The awards program is organized by the Edge AI and Vision Alliance. The Edge AI and Vision Product of the Year Awards celebrate the innovation and achievement of the industry’s leading companies that are enabling and developing products incorporating edge AI and computer vision technologies.
