Cerebras

Cerebras builds a new class of computer systems, including wafer-scale chips, to accelerate training for the world's largest AI and HPC models.

Cerebras Systems is pioneering a new approach to artificial intelligence compute, moving beyond the limitations of traditional GPU clusters. The company designs and manufactures specialized hardware, centered around its Wafer-Scale Engine (WSE), the largest chip ever built. This chip is integrated into a full computer system, like the CS-3, which is engineered to train massive AI models with trillions of parameters as if they were running on a single, powerful device. This drastically simplifies the distributed computing challenges associated with training large-scale models. Cerebras is built for large enterprises, research institutions, and governments that need to train foundational AI models, particularly large language models (LLMs) and complex high-performance computing (HPC) simulations. The company offers its technology through two primary models: direct purchase of its on-premise CS-3 systems and cloud-based access to its Condor Galaxy AI supercomputer. By providing extreme performance and simplified scaling, Cerebras aims to reduce the time and complexity required to make breakthroughs in AI.