Minerva is designed to accelerate AI inference. Inference is the process of using a trained AI model to make predictions, identify patterns or generate insights from new data. Minerva, developed in collaboration with World Wide Technology is equipped with 64 NVIDIA B200 GPUs (graphics processing units).

This system will support the ALCF Inference Service, giving researchers cloud-like access to a wide and diverse set of AI models. The service will support hundreds of models and help users automatically select the best model for their needs. By providing secure and easy access to inference models, the service aims to accelerate research in areas such as drug discovery, fusion energy science, material design and nuclear reactor development