Broadcom Delivers Jalapeño AI Processor in Nine Months with Higher Efficiency
AVGO•Broadcom co-developed Jalapeño, an LLM-optimized intelligence processor for OpenAI built in nine months, delivering performance-per-watt gains over state-of-the-art and slated for gigawatt-scale deployment with data center partners by 2026. Stanley Druckenmiller’s Duquesne Family Office disclosed a position in Broadcom among three AI infrastructure stocks.
1. Collaboration Unveils Jalapeño AI Processor
Broadcom and OpenAI jointly introduced Jalapeño, a dedicated intelligence processor built from scratch for large language model inference. Engineering samples already run GPT-5.3-Codex-Spark at production frequency and power, promising substantial performance-per-watt improvements over competing accelerators.
2. Accelerated Development and Multi-Generation Roadmap
The chip reached manufacturing tape-out in just nine months, leveraging deep software-hardware co-development and Broadcom’s silicon implementation expertise. Jalapeño forms the first step in a multi-generation compute platform, with gigawatt-scale data center deployment planned by the end of 2026 and subsequent generations to follow.
3. Institutional Backing from Duquesne Family Office
Stanley Druckenmiller’s Duquesne Family Office disclosed a position in Broadcom as one of three AI infrastructure investments, signaling institutional confidence in Broadcom’s strategic push into AI processing solutions.
