Deepx announces new AI accelerators with its Genesis core
The South Korean company Deepx has unveiled its new range of artificial intelligence accelerators, presented at the CES 2026 event. The models DX-H1 V-NPU, DX-H1 Quattro, and DX-M1 M.2 aim to compete with traditional GPU-based solutions, proposing a more efficient specialized architecture. 🚀
The DX-H1 V-NPU: specialist in video streams
This device is designed to handle video operations. It can decode, encode, and transcode multiple streams simultaneously. According to Deepx, integrating this module allows reducing hardware costs by 80% and energy consumption by 85%, compared to GPU-based configurations, without sacrificing the number of supported channels.
Key advantages of the DX-H1 V-NPU:- Focused on processing video for security and streaming applications.
- Achieves extreme energy savings compared to conventional GPUs.
- Maintains high channel density to handle many transmissions at once.
The key to saving energy is not turning off the equipment, but having an efficient chip work for you.
Product family and common architecture
In addition to the video model, the series includes the DX-H1 Quattro and the DX-M1 in M.2 format. The latter is designed for easy integration into various systems. All share the Genesis NPU neural processing core, which the company has specifically optimized to execute AI inference tasks quickly and with low power consumption.
Features of the Genesis NPU architecture:- It is the common heart of the three new accelerators.
- Optimized for executing AI inference efficiently.
- Aims to offer high performance in a compact form factor that saves space and electricity.
A bet on specialized efficiency
With these launches, Deepx positions its Genesis NPU technology as a powerful and efficient alternative to general-purpose GPUs for specific AI workloads. The focus on reducing costs and consumption makes them attractive for implementing artificial intelligence solutions at scale, where operational efficiency is crucial. 💡