OpenAI and Broadcom Unveil Custom AI Chip 'Jalapeño' to Rival Nvidia
OpenAI, the developer of ChatGPT, has introduced its first custom artificial intelligence chip, developed in collaboration with Broadcom. Named 'Jalapeño', the chip is designed to reduce the company's dependence on Nvidia processors and to gain greater control over the infrastructure that powers its AI products.
The chip, announced on Wednesday, is an Application-Specific Integrated Circuit (ASIC) tailored for inference workloads. Inference involves running trained AI models on new data to generate responses, a critical function for applications like ChatGPT. The Jalapeño chip aims to improve the efficiency and speed of these operations.
This move places OpenAI among a growing list of tech giants—including Google, Amazon, and Microsoft—that have developed their own custom chips for AI inference. For Broadcom, the partnership reinforces its role as a leading provider of custom silicon for hyperscale data centres.
Greg Brockman, president and co-founder of OpenAI, stated: 'The world is moving to a compute-powered economy. Jalapeño is part of our long-term full-stack infrastructure strategy to make compute more abundant, resulting in AI which is faster, more reliable, more affordable for people and businesses.'
Broadcom's CEO Hock Tan added: 'Our collaboration with OpenAI represents a fundamental commitment to scaling the physical infrastructure required for the next decade of AI. This is just the beginning of a multi-generation roadmap.'
The Jalapeño chip is expected to be deployed in data centres owned or leased by OpenAI starting in 2026. It is designed to work with all large language models, not just OpenAI's own. The chip architecture reduces data movement and balances compute, memory, and networking resources to achieve high utilization. Broadcom's networking technologies, including the Tomahawk silicon, will be used in the platform.
OpenAI described the chip as an 'intelligence processor' that is part of a larger AI accelerator platform. The company aims to make advanced AI faster, more reliable, and more accessible. The development also reflects a broader industry trend where companies seek to reduce reliance on Nvidia GPUs, which have dominated AI training but are less optimized for inference.
Custom ASICs like Jalapeño represent a shift toward specialized hardware for AI tasks. Other examples include Google's TPUs and Amazon's Trainium chips. While the upfront cost of custom chips is steep, they promise long-term savings and efficiency gains.
