Etched

Etched builds specialized ASIC chips for transformer model inference, enabling AI companies to deploy large language models with lower latency and higher efficiency than GPUs.