OpenAI, the organization behind ChatGPT and Codex, has partnered with Broadcom to introduce a new chip named Jalapeño. This chip is specifically engineered for large language model (LLM) inference in data centers, marking a significant step in the evolution of AI hardware.
According to both companies, Jalapeño represents the first generation in a long-term initiative to develop chips tailored for LLM applications. Broadcom states that this Application-Specific Integrated Circuit (ASIC) was built from the ground up, leveraging insights gained from discussions with OpenAI researchers. The chip’s design was influenced by OpenAI’s roadmap for future models and products, indicating a close collaboration between the two entities.
The development process for Jalapeño took approximately nine months. Broadcom emphasizes that this chip is more adept at handling the current demands of LLMs compared to existing systems used in data centers. OpenAI claims that preliminary tests indicate Jalapeño will provide superior performance per watt compared to the current state-of-the-art solutions, although they have not yet completed their performance evaluations. A detailed technical report is expected to be released in the coming months.
OpenAI’s strategy appears to be aimed at achieving greater control over its technology stack, potentially reducing reliance on third-party suppliers like Nvidia. This move is part of a broader trend among AI companies to develop custom silicon, which could help alleviate the ongoing compute crunch faced by data centers amid increasing demand.
Broadcom has been actively expanding its business in custom chip production, particularly for hyperscalers and teams developing advanced AI models. Both OpenAI and Broadcom anticipate that Jalapeño chips will be operational in data centers by the end of this year, although specific deployment details remain unclear.
This article was produced by NeonPulse.today using human and AI-assisted editorial processes, based on publicly available information. Content may be edited for clarity and style.
