OpenAI unveils its first AI processor: Jalapeño

OpenAI has just unveiled a new “intelligence processor” for AI servers developed in partnership with Broadcom. The chip, called Jalapeño, is designed to be used in both current and future major languages, according to the announcement on the third.

Jalapeño is an ASIC (Application-Specific Integrated Circuit), which means it was designed for a specific purpose: AI adoption. With AI concepts, models process the user’s request as a Codex or provide an answer from ChatGPT, while AI training includes a model that uses a large amount of data to determine the answers.

They it comes only nine months OpenAI has revealed that it will partner with Broadcom to develop its own chips, helping to reduce the company’s reliance on GPUs supplied by Nvidia that are limited. In a communication with ReutersBroadcom’s CEO, Hock Tan, says it’s similar to the performance of Nvidia’s Blackwell chips and Google’s Tensor processing units. Microsoft, Trimand Amazon is among other AI companies that have recently introduced custom AI chips to power their servers for training or control, while trailing Nvidia’s chips in overall performance.

OpenAI calls Jalapeño “the first step in multi-core computing,” which it hopes to ship by the end of 2026. “Although OpenAI is still testing the final product, preliminary tests show that Jalapeño will provide more work per watt than it currently has,” OpenAI says.