Wire flash
TechOpenAI unveils AI-designed chip 'Jalapeno', beats Rubin on tokens per megawatt
Editorial responsibility
- No named human review is recorded for this page.
- Source reporting is collected, normalized, translated or condensed automatically when needed.
- Automatically published source-backed update
OpenAI has unveiled Jalapeno, an AI-designed chip that reportedly outperforms the Rubin chip on tokens per megawatt, a key efficiency metric for AI inference. The chip was taped out in just nine months and is capable of running Doom. The announcement was made via a post on X, promoting a detailed breakdown on SemiAnalysis Weekly featuring Myron and Bryan alongside host Jordan Nanos. The episode is expected to cover comparisons between Jalapeno, Rubin, and GB300 on InferenceX, the importance of tokens per megawatt as a metric, Samsung's HBM4 comeback, AI-written RTL and kernels, the potential draining of the CUDA moat, and Anthropic's upcoming ASIC. This development signals a significant advancement in custom AI hardware design and competition in the AI chip market.
Source report
OpenAI's AI-designed chip, codenamed Jalapeno, has been revealed — and it's already making waves. The chip taped out in just nine months, runs Doom, and reportedly beats NVIDIA's upcoming Rubin architecture on tokens per megawatt.
Myron and Bryan joined @JordanNanos on SemiAnalysis Weekly to break down the details.
Key Topics Covered
- Jalapeno vs. Rubin and GB300 on the InferenceX benchmark
- Why tokens per megawatt is the metric that matters most
- Samsung's HBM4 comeback in the memory race
- AI-written RTL and kernels — the rise of automated chip design
- Is the CUDA moat draining? — NVIDIA's competitive position under threat
- Anthropic's ASIC is next in the custom chip pipeline
Watch Now
Full episode on SemiAnalysis Weekly
Source
SemiAnalysis_Neutral / independent
Part of this Story
OpenAI Unveils Custom Jalapeño AI Chip, Challenging Nvidia’s Dominance