OpenAI and Broadcom have unveiled Jalapeño, a custom AI chip designed to run ChatGPT more efficiently and at a lower cost. This inference chip, developed in a rapid nine-month cycle, reportedly outperforms current state-of-the-art chips in performance per watt and is already being tested with GPT-5.3-Codex-Spark, with full deployment expected by the end of 2026. Meanwhile, OpenAI faces a multistate probe concerning potential user harm from ChatGPT, focusing on safety, data privacy, and AI's influence on dangerous behaviors, while also seeing a key Google Gemini model co-lead, Noam Shazeer, depart for the company. AI
IMPACT The custom chip could significantly lower operational costs for OpenAI, potentially impacting pricing and accessibility of its AI services.
RANK_REASON Cluster covers a custom AI chip launch by a major AI lab and a regulatory probe into its flagship product.
- ChatGPT
- United States Of America
- Microsoft
- OpenAI
- OpenClaw
- Broadcom
- DaVinci Commerce
- Diaz Nesamony
- Gemini
- GPT-5.3-Codex-Spark
- Jalapeño
- Noam Shazeer
AI-generated summary · Google Gemini · from 13 sources. How we write summaries →