On Wednesday, June 24th, 2026, OpenAI officially unveiled 'Jalapeño,' its first internally designed AI processor, marking a significant pivot in the company’s infrastructure strategy. Developed in a nine-month sprint with Broadcom and manufacturing partner Celestica, the chip is a dedicated 'Intelligence Processor' specifically optimized for inference rather than model training. OpenAI leadership, including hardware lead Richard Ho, indicated that the custom silicon provides the company with full-stack control, potentially reducing its long-standing reliance on Nvidia's GPU ecosystem. Notably, OpenAI utilized its own models, including GPT-5.3-Codex-Spark, to accelerate the hardware design process. While the chip is already running workloads in labs, commercial deployment via Microsoft and other partners is expected by late 2026, with a long-term goal of powering 10 gigawatts of compute by 2029. This move follows similar silicon initiatives by competitors like Google, Amazon, and Microsoft, further establishing Broadcom as a primary enabler of custom AI accelerators.