Etched

Etched builds frontier inference clusters designed to make AI inference dramatically faster, cheaper, and more abundant. Through deep co-design across the entire stack, Etched's rack-scale systems deliver best-in-class throughput and latency, powered by two novel computing technologies: Low Voltage Inference (LVI) and Cluster Scale Memory (CSM).