MUMBAI: The next AI race is no longer just about smarter models, it’s about who builds the brains beneath them. OpenAI has unveiled Jalapeño, its first custom-built artificial intelligence chip, marking a major step beyond AI applications and into the heart of computing infrastructure as the company seeks greater control over the technology powering its rapidly expanding services.
Developed in partnership with Broadcom, the processor is designed specifically for inference workloads that power ChatGPT, Codex, OpenAI’s API and future AI agents. Unlike general-purpose processors, Jalapeño has been built from the ground up to optimise large language model workloads, reflecting OpenAI’s growing ambition to own more of its AI technology stack.
Engineering samples are already running machine learning workloads in OpenAI’s laboratories at target production frequencies and power levels, including GPT-5.3-Codex-Spark. While final performance validation is still underway, the company said early testing indicates the chip delivers significantly better performance per watt than today’s leading alternatives.
Broadcom Chief Executive Officer Hock Tan said the first samples have already been delivered to OpenAI, with initial results pointing to around 50 per cent lower costs compared with conventional AI graphics processing units (GPUs).
Jalapeño is the first product to emerge from the companies’ partnership, first announced last year with plans to develop processors capable of supporting 10 gigawatts of AI computing. The chip is expected to enter deployment by the end of 2026, with additional generations already in development.
OpenAI said the processor was designed using its deep understanding of large language models, serving systems and future product requirements, while Broadcom led chip development and Celestica helped industrialise the platform.
The company also revealed that Jalapeño progressed from initial design to manufacturing tape-out in just nine months, describing it as the fastest ASIC development cycle for a high-performance advanced semiconductor. OpenAI added that it used its own AI models to accelerate elements of the chip’s design and optimisation process.
The launch reflects a broader shift across the technology industry, with major AI companies increasingly designing proprietary processors to improve efficiency, lower operating costs and reduce reliance on third-party hardware suppliers.
The move also comes as OpenAI reportedly prepares for a potential initial public offering that could value the company at around $1 trillion, intensifying its focus on expanding infrastructure, improving margins and scaling AI services globally. By designing more of its technology stack in-house, OpenAI aims to deliver faster, more efficient AI while making advanced capabilities accessible to a wider range of users.