OpenAI and Broadcom unveil Jalapeño, a custom-built AI inference chip developed in nine months, targeting deployment at gigawatt-scale data centers by end-2026.

OpenAI and Broadcom have unveiled Jalapeño, a custom-designed inference chip developed over nine months. The collaboration marks a strategic move by the AI leader to build specialized silicon for its computational infrastructure.

The chip is engineered specifically for gigawatt-scale data center deployments, with both companies targeting production readiness by the end of 2026. This timeline suggests OpenAI is preparing for substantial inference workload expansion across its services.

The development underscores the broader industry trend of major AI companies moving beyond off-the-shelf GPUs to purpose-built silicon. For the AI buildout, this signals OpenAI's confidence in its inference demand trajectory and its willingness to invest in long-lead hardware to support scaling—a capital-intensive bet on sustained competitive advantage in AI service delivery.