AI semiconductor company DeepX is collaborating with Ultralytics, the developer of the object recognition AI model 'YOLO', to integrate its NPU into a global AI development platform. This collaboration enables developers to simplify model conversion and deployment tailored to DeepX chips within the Ultralytics environment.

DeepX recently announced that it has established a strategic technology and market entry partnership with Ultralytics. Ultralytics is a company that operates the YOLO ecosystem, an object recognition model used in industrial cameras, robots, autonomous driving, and smart cities.

The key feature of this collaboration is that the DeepX NPU integrates natively with the Ultralytics platform. Developers can convert and deploy AI models to the DeepX chip environment using the 'format=deepx' command in Ultralytics Python packages.

Previously, applying AI models to actual devices required separate processes of model conversion, optimization, and verification. Both companies expect that this integration will reduce the complexity of the hardware implementation process and shorten the time it takes for developers to move from model development to the actual product application stage.

DeepX also plans to provide a hardware environment optimized for the latest YOLO model family and the next-generation model, YOLO26. Along with the DX-M1 currently in mass production, the next-generation DX-M2 based on the 2nm process is also being designed with support for these models in mind.

This collaboration is expected to serve as an opportunity for DeepX to expand its reach to developers within the global AI software ecosystem. Ultralytics' hardware partner group also includes global semiconductor companies such as Intel and STMicroelectronics.

DeepX explains that its NPU is designed to run object recognition models in low-power environments. Accordingly, it plans to expand its potential applications, focusing on fields requiring AI computation in real-world environments such as robotics, industrial automation, and smart cities.

Kim Nok-won, CEO of DeepX, stated that this collaboration marks a turning point where K-AI semiconductors become a standard choice for developers worldwide, and expressed his intention to establish them as the standard for physical AI semiconductors in various devices such as robots, smart cities, and home appliances.