{"version":1,"type":"story","url":"https://digestai.news/story/openai-says-jalapeno-chip-delivers-up-to-1-9-more-tokens-per-watt-than","json":"https://digestai.news/story/openai-says-jalapeno-chip-delivers-up-to-1-9-more-tokens-per-watt-than.json","markdown":"https://digestai.news/story/openai-says-jalapeno-chip-delivers-up-to-1-9-more-tokens-per-watt-than.md","slug":"openai-says-jalapeno-chip-delivers-up-to-1-9-more-tokens-per-watt-than","headline":"OpenAI says Jalapeño chip delivers up to 1.9× more tokens per watt than Nvidia","summary":"OpenAI unveiled its first in‑house inference chip, Jalapeño, built with Broadcom and assembled by Celestica. The chip targets data‑movement efficiency, claiming 1.5 to 1.9 times more tokens per watt than Nvidia’s Blackwell systems on a 700‑watt package (measured draw at or below 550 W). Specs shared at Hot Chips include 13.4 PFLOPS MXFP4 compute, 216 GiB HBM4 memory at 15.4 TB/s, 64 core slices each with private memory, and the ability to scale to 128 chips in a local domain or 2,048 chips in a 16‑rack pod linked by Broadcom Tomahawk 6 Ethernet.\n\nOpenAI tested Jalapeño on the InferenceX benchmark, reporting about 1.9× more tokens per kilowatt and 1.7× lower latency versus Nvidia GB200 for a 120 B GPT‑OSS model, and similar gains against GB300 for DeepSeek R1 and Kimi K2.5. The tests were run internally, and no independent verification is cited. The chip will first appear in very small volumes at the end of 2026, with broader rollout in 2027, and will power OpenAI services such as ChatGPT, Codex and its API.\n\nThe design was completed in nine months, from RTL code in February 2025 to tape‑out in November 2025, aided by AI‑driven design tools that wrote more than half the core in the XLS language. A team of roughly 100 engineers handled design, verification, packaging and software, but OpenAI still does not trust the models completely, keeping human validation for all changes.","keyPoints":["OpenAI says Jalapeño yields 1.5‑1.9× more tokens per watt than Nvidia Blackwell on a 700 W package","Chip specs: 13.4 PFLOPS MXFP4, 216 GiB HBM4, 64 core slices, up to 2,048 chips per pod","Design completed in nine months using AI‑assisted hardware language XLS"],"whyItMatters":"If OpenAI's efficiency claims hold, tokens‑per‑watt could become a key metric for AI services, influencing data‑center energy costs and prompting vendors to prioritize data‑movement reduction.","category":{"slug":"hardware","name":"Hardware & Compute","url":"https://digestai.news/category/hardware"},"entities":{"companies":["OpenAI","Broadcom","Celestica","Nvidia","AMD","European Commission"],"models":["GPT-OSS 120B","DeepSeek R1","Kimi K2.5"],"people":["Richard Ho","Tae Kim","Ian Cutress"]},"firstPublishedAt":"2026-10-26T14:00:00Z","updatedAt":"2026-10-26T14:00:00Z","sourceCount":1,"hasPrimarySource":false,"sources":[{"outlet":"i-scoop.eu","title":"OpenAI’s Jalapeño chip bets that cheaper AI is about moving less data","url":"https://i-scoop.eu/openai-jalapeno-chip","publishedAt":"2026-10-26T14:00:00Z","type":"press","primary":false,"lead":true}],"sourceNotes":null,"discussions":[],"thread":{"title":"OpenAI Jalapeño Chip Challenges Nvidia Dominance","url":"https://digestai.news/thread/amd-says-96-core-venice-cpu-is-about-20-faster-than-nvidia-vera-in-spec-cpu-2026","storyCount":5},"cite":{"text":"Digest AI, \"OpenAI says Jalapeño chip delivers up to 1.9× more tokens per watt than Nvidia\", 26 October 2026, https://digestai.news/story/openai-says-jalapeno-chip-delivers-up-to-1-9-more-tokens-per-watt-than","publisher":"Digest AI","title":"OpenAI says Jalapeño chip delivers up to 1.9× more tokens per watt than Nvidia","datePublished":"2026-10-26T14:00:00Z","url":"https://digestai.news/story/openai-says-jalapeno-chip-delivers-up-to-1-9-more-tokens-per-watt-than"},"generatedBy":"Written by Digest AI's editorial model from the linked sources; the sources are the record.","license":"Headlines, digests and key points are written by Digest AI and may be quoted with a link to the story page. Linked articles belong to their publishers. Terms: https://digestai.news/terms#reuse"}