Giada is preparing to introduce a new AI accelerator module designed to enhance edge computing performance while maintaining extremely low power consumption. Built with advanced AI processing capabilities and a compact industrial design, the upcoming module aims to provide developers and system integrators with a flexible solution for deploying AI inference directly at the edge.

25 TOPS of AI Computing Power

At the heart of the module is the DeepX M1M NPU, delivering up to 25 TOPS of AI computing power. The accelerator supports INT8 and FP16 mixed precision, enabling efficient execution of modern neural networks used in computer vision, object detection, and other AI inference tasks. This powerful yet efficient processing capability allows edge devices to handle complex workloads locally without relying heavily on cloud computing resources.

3.6W Ultra-low Power

One of the most notable features of the module is its ultra-low power consumption. With a typical power usage of only 3.6W and idle power below 0.5W, the accelerator is ideal for energy-sensitive edge environments. It also supports dynamic voltage and frequency scaling (DVFS), allowing the system to automatically adjust performance based on workload demands to further optimize power efficiency.

M.2 Standard Form Factor

The AI accelerator adopts a standard M.2 M+B Key 2242 form factor (22mm × 42mm), making it easy to integrate into existing embedded systems. Through a PCIe Gen3 x2 interface, the module can deliver high-speed data transfer while maintaining compatibility with a wide range of industrial motherboards, edge PCs, and embedded platforms.

Proven Software Ecosystem

To support AI development and deployment, the module comes with a proven software ecosystem. It supports popular operating systems including Windows, Ubuntu, and Debian, allowing developers to integrate it into different computing environments. The accelerator also works with major AI frameworks such as TensorFlow and PyTorch, and provides Python and C++ APIs through the DEEPX Runtime, simplifying model deployment and application development.

Designed for industrial-grade environments, the module supports an operating temperature range from –25°C to 65°C, ensuring stable performance in demanding conditions.

With its powerful AI performance, compact design, and efficient power profile, Giada's upcoming AI accelerator module is set to become a valuable tool for accelerating edge AI applications across industries, including intelligent vision systems, smart retail, industrial automation, and next-generation embedded AI solutions.

© 2026 Giada Technology. All rights reserved.