OpenAI Debuts Jalapeño AI Inference Chip, Challenging NVIDIA with Reported Performance Gains

OpenAI has unveiled its first custom AI inference chip, 'Jalapeño,' which reportedly delivers 1.5x–1.9x higher throughput per kilowatt and 1.7x–3.6x lower end-to-end latency compared to NVIDIA's GB200 and GB300 rack systems in initial benchmarks. This development could significantly impact the AI infrastructure landscape and the demand for HBM4 memory, with Samsung reportedly supplying the HBM4.