OpenAI and Broadcom Reveal ‘Jalapeño’ — A Groundbreaking Custom AI Chip
OpenAI in partnership with Broadcom (NASDAQ: AVGO) proudly introduced Jalapeño on June 24, 2026. This marks OpenAI’s debut in custom-designed AI chips and stands as one of the most pivotal hardware announcements in the company’s history. The chip, an AI inference accelerator (ASIC), is specifically designed from scratch to cater to OpenAI’s large language models in ChatGPT and other applications.
Remarkably, Jalapeño transitioned from initial design to manufacturing tape-out in a mere nine months. This rapid development is claimed by the companies to possibly be the fastest high-performance ASIC development cycle ever achieved. OpenAI’s proprietary AI models played a crucial role in accelerating parts of the chip design process. Early testing indicates that the accelerator provides approximately 50% cost savings compared to conventional AI GPUs, as stated by Broadcom CEO Hock Tan. The chip also demonstrates significantly improved performance per watt compared to current top-tier hardware.
Greg Brockman, President of OpenAI, elaborated that the chip was designed around deeply understood LLM workloads that were previously underserved by general-purpose silicon. Jalapeño is scheduled for initial deployment by the end of 2026, with gigawatt-scale data centers being constructed in collaboration with Microsoft and other partners. This strategic move places OpenAI in the same league as Google, Apple, and Amazon in building custom silicon, marking a significant strategic shift away from exclusive reliance on Nvidia GPUs.
Source: TechCrunch – OpenAI unveils its first custom chip, built by Broadcom