5.2 AIConnect's Investment in AI Edge Computing Model Training and Development

AIConnect is responsible for the training and development tasks of edge computing models in the AI laboratory built with a $100 million investment from JUBI. Edge large models are large-scale artificial intelligence models that operate in edge computing environments. Compared to traditional cloud computing platforms, edge large models are closer to the data source, allowing for faster response to user needs and improved data processing efficiency. Moreover, edge large models can be deployed in various ways, such as on edge computing devices or distributed computing networks, providing enterprises with more flexible and efficient AI services.

In practical applications, the value that edge large models bring to enterprises is evident. First, edge large models can analyze and process large amounts of data in real-time, providing enterprises with timely and accurate feedback, helping them make faster and wiser decisions. Secondly, edge large models can be integrated with various IoT devices, achieving intelligent interconnectivity of devices and further expanding the business scope of enterprises. Lastly, edge large models can also provide customized services such as facial recognition and speech recognition to enterprises, enhancing their level of intelligence.

AIConnect has explored product forms oriented towards the era of large models and AIGC, and is gradually implementing them. It is reported that AIConnect, based on the AIC edge intelligence platform, has built an edge GPU computing power platform and an open-source large model training and deployment platform for scenarios such as large model inference and training. It also provides solutions for vertical fields to adapt to market demands and technological development, empowering various industries with large models.

Among them, the AIC edge GPU computing power platform, based on the widely distributed node resources of AIConnect, provides lightweight computing resources, supports GPU virtualization, and can meet the needs of lightweight AI task scenarios, such as AI inference, deep learning, and graphical visualization.

The AIConnect open-source large model training and deployment platform, based on open-source pre-trained large models, provides an end-to-end large model service platform that includes model fine-tuning training, performance evaluation, deployment monitoring, and lightweight inference, which can reduce the cost of large model application implementation and help customers build their own exclusive large models.

In addition, AIConnect provides vertical field solutions, including ready-to-use enterprise knowledge base solutions based on private large models, which can serve both internal and external customers of enterprises, as well as image generation solutions for the e-commerce field, such as AI models, to help vertical fields reduce costs and increase efficiency.