OpenAI and Broadcom Introduce New AI Accelerator Chip

OpenAI and Broadcom have collaborated to launch 'Jalapeño,' a custom AI accelerator chip specifically designed for large language model inference. This project achieved a record-setting nine-month design-to-tape-out timeline, partially attributed to OpenAI's utilization of its own AI models in the development process. Initial evaluations suggest the chip offers superior performance per watt compared to existing solutions, with large-scale deployment anticipated by late 2026.