OpenAI and Broadcom unveil first AI chip amid rising demand

Traders Union
All News
Financial News
OpenAI unveils first chip

Highlights

OpenAI unveiled its first custom AI chip, Jalapeño.
Broadcom will make the chip for OpenAI.
OpenAI is trying to control more of its AI infrastructure.

OpenAI moves into silicon

The chip will be made by Broadcom and used by OpenAI for inference, the process of serving AI models to users after they have been trained. In practice, that means Jalapeño is aimed at the expensive, high-volume work behind ChatGPT responses, enterprise tools, and other applications, CNBC reports.

OpenAI President Greg Brockman said designing more of the technology stack internally would help the company serve more intelligence with greater efficiency and broaden access to advanced AI. The companies described the new processor as an “Intelligence Processor” and the first AI accelerator in a broader platform they plan to expand.

Broadcom shares rose about 2% after the announcement. The chipmaker has become a major beneficiary of the AI boom by helping large technology companies and frontier AI labs design custom silicon.

A push beyond Nvidia

OpenAI has been one of the largest buyers of Nvidia GPUs since ChatGPT helped ignite the generative AI boom in 2022. Nvidia chips remain central to training and running advanced models, but high prices and limited supply have pushed major AI companies to diversify.

OpenAI has already signed deals with Amazon Web Services for Trainium chips, Advanced Micro Devices and Cerebras Systems. The Broadcom partnership goes further because it gives OpenAI a chip designed around its own workloads.

The Jalapeño chip is an ASIC, a type of processor built for specific tasks. ASICs are less flexible than general-purpose GPUs, but they can be cheaper and more efficient when designed for a narrow set of AI workloads. The chip is initially focused on inference rather than training, with deployment for customer use expected later this year.

The infrastructure race gets deeper

OpenAI’s chip debut matters because the AI contest is no longer only about models. It is also about chips, power, data centers and the cost of serving billions of AI requests.

Custom silicon could help OpenAI reduce dependence on Nvidia over time, though it will not replace GPUs overnight. The company is building a broader supplier base while trying to make inference cheaper and more reliable. For Broadcom, the deal strengthens its position as the custom-chip partner of choice for AI companies that want hardware tailored to their own systems.

We also reported OpenAI completes one of banking’s largest AI deployments.

This material may contain third-party opinions, none of the data and information on this webpage constitutes investment advice according to our Disclaimer. While we adhere to strict Editorial Integrity, this post may contain references to products from our partners.