OpenAI unveils Jalapeno AI inference chip
Analysis based on 9 articles · First reported Aug 25, 2026 · Last updated Aug 26, 2026
The announcement signals OpenAI's move to reduce reliance on Nvidia GPUs for inference, potentially increasing competition in the AI chip market and pressuring Nvidia's dominance. It also highlights the growing trend of AI companies designing custom silicon, which could impact semiconductor supply chains and boost demand for Broadcom's custom ASIC services.
OpenAI unveiled performance results for Jalapeño, its first custom AI inference chip, developed in partnership with Broadcom. The chip is designed to accelerate AI inference, the stage where trained models generate answers, and aims to deliver more AI work per watt while reducing response latency. In benchmarks using InferenceX from SemiAnalysis, OpenAI reported that Jalapeño achieved 1.5 to 1.9 times more AI work per watt, 1.7 to 3.6 times lower end-to-end latency, and 2.1 to 4.1 times higher performance for interactive workloads across three models: GPT-OSS 120B, DeepSeek R1 670B, and Kimi K2.5 1T. OpenAI plans to begin deploying Jalapeño in its infrastructure by the end of 2026, with second and third generations in development. The chip was designed with the help of OpenAI's own AI models, reducing the design-to-tapeout cycle to nine months. OpenAI emphasized that it will continue using Nvidia and other partners' accelerators for both training and inference, positioning Jalapeño as an additional compute source rather than an immediate replacement. The announcement was made at the Hot Chips conference in Palo Alto, California, and CEO Sam Altman commented on X: 'we made a chip and it is fast.'
Set up alerts, explore entity relationships, search across thousands of events, and build custom intelligence feeds.
Open Dashboard