GEMESYS

Next Generation Edge AI-Hardware for Training and Inference

Contact for Pricing

Description

GEMESYS is pioneering next-generation AI hardware with its unique analog chip architecture inspired by the human brain. This innovative approach aims to overcome current bottlenecks in information technology by offering significantly increased data and energy efficiency. The company focuses on enabling both AI training and inference capabilities directly at the edge.

The core offering is a fully analog integrated circuit designed to revolutionize AI hardware. By leveraging principles of the human brain's information processing and utilizing memristor technology, GEMESYS anticipates a major leap in AI training efficiency. Their hardware supports neural networks with high sparsity and low depth, optimizing inference operations and enabling smart, autonomous, and adaptable capabilities for edge devices and sensors with low power consumption and a small footprint.

Key Features

Edge AI Training & Inference Capability: Performs both AI model training and execution on edge devices.
Fully Analog Chip Design: Utilizes an analog paradigm, moving away from traditional digital architectures.
Human Brain-Inspired Architecture: Mimics the information processing mechanisms of the human brain.
High Training Efficiency: Expected efficiency increase by a factor of 10,000 due to the analog approach.
Energy Efficiency: Designed for low power consumption suitable for mobile and edge devices.
Data Efficiency: Optimized for efficient data handling at the edge.
Small Chip Size: Enables integration into compact devices.
Memristor-Enabled: Leverages memristor technology for its analog computation.
Optimized for Sparse Networks: Supports high sparsity and low-depth neural networks for efficient inference.