OpenAI and Broadcom have revealed a custom chip, called the Jalapeño chip, which is optimized for large language model (LLM) inference. The chip is reportedly OpenAI's first and was developed in collaboration with Broadcom, with some outlets suggesting it could significantly reduce processing costs. Details about the chip's development and potential impact on the industry are emerging, with some reports indicating a rapid nine-month development cycle and potential implications for other companies in the field.