Hardware4h ago
OpenAI and Broadcom Unveil Jalapeño Inference Chip
The Jalapeño chip, co-developed by OpenAI and Broadcom, targets large language model inference at scale. This custom silicon will enter production by late 2026. By diversifying its hardware stack, OpenAI reduces its reliance on Nvidia GPUs. This move optimizes inference costs and latency for future model deployments across its massive user base.