OpenAI Signs Partnership with Cerebras

OpenAI has partnered with Cerebras to add 750MW of ultra low-latency AI computing resources to its platform.

Cerebras builds purpose-built AI systems that integrate massive compute, memory, and bandwidth into a single giant chip, eliminating the bottlenecks of traditional hardware. This accelerates long outputs from AI models and dramatically increases inference speed.

The core of this collaboration is improving AI response speed to real-time levels. It minimizes the latency that occurs when users ask complex questions, generate code, create images, or run AI agents, enabling more natural interactions.

This low-latency capacity will be integrated into the inference stack in multiple phases through 2028, gradually expanding to cover a variety of workloads.