GoKawiilOpenAI unveils Jalapeño inference chip, outperforming Nvidia and AMD in tests
OpenAI has revealed details of Jalapeño, a custom inference chip developed with Broadcom over roughly 16 months, at the Hot Chips conference. SemiAnalysis tested the chip using its InferenceX benchmark suite and found it outperforms Nvidia, AMD, and Google chips across multiple open-source models, despite being OpenAI's first hardware effort.
GoKawiil's interpretation of the reporting above, not reported fact.
A first-generation chip beating established GPU makers signals that heavy investment in hardware-software codesign, rather than years of iterative refinement, can shortcut typical chip development timelines. This could pressure Nvidia's dominance in AI inference hardware and push other AI labs toward building their own silicon rather than relying solely on merchant GPUs.
- Jalapeño was designed in about 16 months, an unusually fast ASIC development cycle
- Independent benchmarks show it beats Nvidia, AMD, and Google chips on open-source models
- The chip is built as a general-purpose inference accelerator, not narrowly tuned to OpenAI's own models
NVIDIA GeForce RTX 4060 Ti Graphics Card — If reading about custom LLM inference silicon makes you want to run models on your own desk, an RTX 4060 Ti is an accessible way in. Its CUDA support means most local LLM and image-generation tools work out of the box, so you can benchmark your own inference workloads at home.
See NVIDIA GeForce RTX 4060 Ti Graphics Card on Amazon → Affiliate link — we may earn a commission on purchases, at no extra cost to you. Product picked by AI based on this article; it is not a tested recommendation.Source: newsletter.semianalysis.com — Bryan Shan, 2026-08-25
Published there as: “OpenAI Jalapeño: Better than Nvidia Blackwell”
Read the original report → The summary and analysis above are GoKawiil's own, written from reporting by the source above. Facts and quotes belong to the original publisher.