NVIDIA recently introduced ‘Apollo’, an open-source physics-based AI model, projected to impact a range of advanced scientific and industrial applications. Initially revealed at the SC25 international conference, Apollo is part of a series of four models released within a month, including Nemotron for agentic AI and Clara for biomedical AI.
The Apollo model is designed to integrate real-time functionalities within various simulation software, supporting fields like defect detection, computational lithography, and fluid-structure interaction. It offers pre-trained checkpoints and reference workflows for learning and inference, allowing developers to customize the model based on their specific needs. The impending release of Apollo will be available through NVIDIA’s website and other platforms.
However, concerns arise regarding the dependency on NVIDIA’s ecosystem as developers may gravitate towards their software and hardware solutions. Experts suggest that the overwhelming presence of NVIDIA’s technology in supercomputing, with over 80 systems announced this year alone, represents a concerning trend towards vendor reliance rather than choice, potentially compromising autonomy in scientific progress.
👉 Pročitaj original: CIO Magazine
