According to an announcement from OpenAI and Broadcom, the companies have unveiled a new processor designed specifically for running large language models. The chip, called Jalapeno, is optimized for inference workloads—the computationally intensive task of executing already-trained AI models. Specialized inference processors like this can significantly reduce latency and power consumption compared to general-purpose chips, potentially making AI models faster and more efficient to deploy.

Listen to this story

Hear this and more stories in a personalized audio briefing.

Open The Chonkerton