OpenAI enters AI chip race with new processor designed for future ChatGPT models

OpenAI has introduced Jalapeno, its first custom AI chip designed to support future generations of ChatGPT and other AI models. Built in partnership with Broadcom, the processor marks a major step in OpenAI's efforts to control its own AI infrastructure and improve performance, efficiency, and scalability.

Published By: Deepti Ratnam | Published: Jun 25, 2026, 08:00 AM (IST) | Edited: Jun 25, 2026, 08:01 AM (IST)

OpenAI is finally moving beyond the AI infrastructure by launching a new custom chip called 'Jalapeno.' It is considered the first 'Intelligence Processor' from OpenAI in collaboration with Broadcom. This is the first time in the company's plans to build a full AI computing stack to power its own services and products.

According to the company, Jalapeno has been engineered with the company's long-term plans in mind for large language models (LLMs) and future AI systems. The chip is already being used for machine learning workloads within OpenAI's research labs, and is the first step in a multi-generation compute platform it is developing with Broadcom.

OpenAI desires more control over AI infra

In the past, OpenAI has been using third-party hardware vendors, such as Nvidia, to train and operate its AI models. The company aims to have more control over the infrastructure that powers its products like ChatGPT by designing its own chips.

The custom chipset by OpenAI is basically for AI inference, wherein it will have the task of operating a trained AI model and provide results to users. The tech giant's mission is to provide performance at par with top AI accelerators, with a lower latency and enhanced efficiency.

Jalapeno's initial tests show that the processor performs more efficiently than other top-of-the-range AI chips available today, the company said. If this is the case, OpenAI may be able to slash costs and boost the speed of its AI services in real industry applications.

Designed for future use of AI models

One of the best benefits of creating a custom processor is that OpenAI can optimize the hardware and software at the same time. The company claims Jalapeno was designed in consideration for the company's future AI roadmap, which will enable it to better serve future generations of models.

Not only this could help OpenAI to make the future versions of ChatGPT faster, but also more efficient and even cheaper to run. In addition, it can also enhance the efficiency of training and help in the creation of more powerful and sophisticated AI systems.

It comes as part of a larger trend in the tech industry as organisations begin to design specialised chips that can cater to their specific AI computing demands rather than using general-purpose chips.

This was developed in a span of nine months only

OpenAI claims the entire design to manufacturing of tape-out only took nine months. Curiously, the company announced it employed its own AI models to help out with portions of the chip development process.

The quick pace of development shows how AI is starting to influence software development, as well as hardware and semiconductor engineering.

Deployment planned for 2026

The chip has already had some internal testing but OpenAI states it intends to roll it out more widely this year by the end of 2026. The company also said that this is a first step of its plans for chips and that more are on the way in the coming years.

Additional technical information on Jalapeno will be available in a separate technical report in the near future.

If what OpenAI claims about this chipset comes true, then this custom processor could be a crucial component for company's future AI infrastructure, as well as one that helps them cut down on reliance on third parties and speed up the creation of the upcoming AI models.

Get latest Tech and Auto news from Techlusive on our WhatsApp Channel, Facebook, X (Twitter), Instagram and YouTube.