Builds infrastructure and software for large-scale AI inference. Its platform combines dedicated accelerator clusters with managed and reservable model-serving capacity, aiming to improve useful tokens per megawatt across hardware architectures.
See something inaccurate or outdated?
