[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"$frrZ4rcxqHfT8by99H2Uk1pE2L_ratjkXIdPU61_vtTs":3},{"locale":4,"topic":5,"relatedTrends":114},"fr",{"topic":6,"slug":7,"canonicalSlug":7,"topicAliases":8,"nicheKey":10,"nicheName":11,"nicheNameEn":12,"nicheIcon":13,"country":14,"countries":15,"agentKey":16,"score":17,"type":18,"isFresh":19,"isPublic":20,"detectedAt":21,"sources":22,"evidence":110},"OpenAI and Broadcom's 'Jalapeño' LLM inference ASIC","openai-and-broadcom-s-jalapeno-llm-inference-asic",[9],"OpenAI and Broadcom's 'Jalapeño' LLM inference chip","ia","Intelligence Artificielle","Artificial Intelligence","🧠","JP",[14],"ia-JP",100,"spiking",true,false,"2026-06-26T00:21:33.502Z",[23,29,35,40,45,50,55,60,65,70,75,80,85,90,95,100,105],{"title":24,"url":25,"domain":26,"snippet":27,"content":28},"OpenAIとBroadcom、「AIが設計したAIのためのチップ」である“Jalapeño”を発表","https:\u002F\u002Fxenospectrum.com\u002Fopenai-jalapeno-inference-chip-broadcom\u002F","xenospectrum.com","OpenAI, Broadcom and Celestica unveiled 'Jalapeño', a custom HBM-equipped ASIC designed for LLM inference that reached tape-out in nine months and promises substantial inference cost reductions.","SEARCH ESC \n\n最新記事 →カテゴリ一覧 →ブックマーク →テーマ切替\n\nEnter で検索 · Esc で閉じる · ↑↓ で移動\n\n×\nXenoSpectrumをGoogleの優先ソースに登録しませんか？\n\nGoogle検索で優先的に表示されます\n\n!](https:\u002F\u002Fgoogle.com\u002Fpreferences\u002Fsource?q=https:\u002F\u002Fxenospectrum.com)\n\n!\n\nこの記事のポイント\n*   何が起きた: OpenAIがBroadcom・Celesticaと共同開発したLLM推論専用カスタムASIC「Jalapeño」を2026年6月24日に発表。LLM推論に特化したゼロベース設計でHBM（高帯域幅メモリ）を搭載し、設計から量産テープアウトまで9ヶ月で完了した。\n*   なぜ重要か: OpenAI自身のモデルを設計ツールとして活用する「AIがAIインフラを加速する」サイクルが実証され、カスタムシリコン開発の所要期間が業界標準の半分以下になりうることが示された。\n*   次に見るべき点: 2026年末の初回展開以降、推論コスト50%削減を裏付ける技術レポートの公開時期と性能・消費電力の独立検証が鍵となる。\n\n推論リクエスト1件あたりの電力消費は、LLM（大規模言語モデル）サービスの採算と普及速度を左右する根本的な制約だ。GPUのレンタルコストがサービス収益を圧迫し、NVIDIAへの依存が調達リスクに直結している現実は、AIサービスを運営するすべての事業者が直面している。OpenAIはこの問題に正面から取り組むため、NVIDIAのGPUを借り続ける道を捨てた。推論に特化したシリコンをゼロから設計し、業界標準の1.5〜2年と比較してわずか9ヶ月で量産段階に到達させたのだ。2026年6月24日に公開された「Jalapeño（ハラペーニョ）」がそれだ。この異例のスピードを可能にした背景には、OpenAI自身のモデルを設計ツールとして使うサイクルがあり、AIがAIインフラを加速する新しい開発手法の実証例として業界の注目を集めている。\n\nAD\n\n詳細を見る\n\n光電子、ファイバー\n\nTV ゲーム、PC ゲーム\n\n果物、野菜\n\n企業ニュース\n\nIC・ICチップ\n\nロールプレイング ゲーム\n\n教育資料\n\nカスタムASIC（特定用途向け集積回路）の開発サイクルが長い理由は、設計検証の反復にある。チップの動作はシリコンを実際に焼いてみるまで最終確認できないため、設計エラーが発覚するたびに数ヶ月単位のやり直しが発生する。従来のチームは大量のシミュレーション実行と人力レビューでこのリスクをカバーしてきた。\n\nOpenAIはこのボトルネックに自社のモデルを投入した。Greg Brockman\n\n[Content truncated...]",{"title":30,"url":31,"domain":32,"snippet":33,"content":34},"OpenAI and Broadcom unveil LLM-optimized inference chip","https:\u002F\u002Fopenai.com\u002Findex\u002Fopenai-broadcom-jalapeno-inference-chip\u002F","openai.com","OpenAI and Broadcom introduce Jalapeño, a custom AI chip built for LLM inference to improve performance, efficiency, and scale across AI systems.",null,{"title":36,"url":37,"domain":38,"snippet":39,"content":34},"OpenAI and Broadcom Unveil LLM-Optimized Intelligence Processor","https:\u002F\u002Fwww.broadcom.com\u002Fcompany\u002Fnews\u002Fproduct-releases\u002F64506","broadcom.com","Built from the ground up for current and future LLMs across the industry. Developed from design to production in nine months, accelerated by OpenAI's models.",{"title":41,"url":42,"domain":43,"snippet":44,"content":34},"OpenAI, Broadcom Unveil Jalapeño AI Chip: A Seismic Shift Coming to AI Hardware Market?","https:\u002F\u002Fwww.businesskorea.co.kr\u002Fnews\u002FarticleView.html?idxno=271972","businesskorea.co.kr","On June 24 (local time), generative artificial intelligence (AI) leader OpenAI and global telecommunications and semiconductor design giant Broadcom...",{"title":46,"url":47,"domain":48,"snippet":49,"content":34},"OpenAI unveils first chip as part of Broadcom deal in effort to 'build the full stack'","https:\u002F\u002Fwww.cnbc.com\u002F2026\u002F06\u002F24\u002Fopenai-and-broadcom-reveal-jalapeno-first-ai-chip-in-partnership.html","cnbc.com","Eight months after announcing a custom chip deal, OpenAI and Broadcom are revealing their first joint project: Jalapeño.",{"title":51,"url":52,"domain":53,"snippet":54,"content":34},"OpenAI and Broadcom unveil \"Jalapeño,\" a custom chip built for LLM inference","https:\u002F\u002Fthe-decoder.com\u002Fopenai-and-broadcom-unveil-jalapeno-a-custom-chip-built-for-llm-inference\u002F","the-decoder.com","OpenAI is adding custom hardware to its tech stack. The \"Jalapeño\" chip, developed with Broadcom, is tailored for large language model inference and is set...",{"title":56,"url":57,"domain":58,"snippet":59,"content":34},"OpenAI Unveils Custom AI Chip","https:\u002F\u002Fwww.startuphub.ai\u002Fai-news\u002Fartificial-intelligence\u002F2026\u002Fopenai-unveils-custom-ai-chip","startuphub.ai","OpenAI and Broadcom unveil Jalapeño, a custom LLM inference chip designed for higher performance and efficiency, marking a key step in OpenAI's full-stack...",{"title":61,"url":62,"domain":63,"snippet":64,"content":34},"OpenAI unveils first custom AI inference chip, Jalapeño, with Broadcom — and its development was sped-up with OpenAI's own models","https:\u002F\u002Fventurebeat.com\u002Finfrastructure\u002Fopenai-unveils-first-custom-ai-inference-chip-jalapeno-with-broadcom-and-its-development-was-sped-up-with-openais-own-models","venturebeat.com","The companies attributed this speed to a deep software-hardware co-development process that actively used OpenAI's own models to accelerate parts of the...",{"title":66,"url":67,"domain":68,"snippet":69,"content":34},"OpenAI Unveils Jalapeno LLM Inferencing Accelerator Built in Collaboration with Broadcom","https:\u002F\u002Fwww.techpowerup.com\u002F350253\u002Fopenai-unveils-jalapeno-llm-inferencing-accelerator-built-in-collaboration-with-broadcom","techpowerup.com","OpenAI today unveiled Jalapeño, its in-house, purpose-built LLM inference accelerator chip, developed in collaboration with Broadcom.",{"title":71,"url":72,"domain":73,"snippet":74,"content":34},"OpenAI gets chippy with Broadcom","https:\u002F\u002Fwww.theregister.com\u002Fai-and-ml\u002F2026\u002F06\u002F24\u002Fopenai-gets-chippy-with-broadcom\u002F5261697","theregister.com","Jalapeño is the latest announcement that attempts to portray OpenAI as more than a race-to-the-bottom model maker.",{"title":76,"url":77,"domain":78,"snippet":79,"content":34},"OpenAIがBroadcomと共同開発したLLM最適推論チップ「Jalapeño」を発表","https:\u002F\u002Fgigazine.net\u002Fnews\u002F20260625-openai-broadcom-jalapeno\u002F","gigazine.net","OpenAIが半導体企業のBroadcomと共同で設計・製造した初のカスタムプロセッサ「Jalapeño(ハラペーニョ)」を発表しました。このJalapeñoはOpenAIの推論システムに合わせ...",{"title":81,"url":82,"domain":83,"snippet":84,"content":34},"OpenAI、LLMのための新チップ「ハラペーニョ」を自社開発 年内にGW級","https:\u002F\u002Fwww.watch.impress.co.jp\u002Fdocs\u002Fnews\u002F2119811.html","watch.impress.co.jp","OpenAIは24日(米国時間)、LLM(大規模言語モデル)向けの推論プラットフォームとして設計した新チップ「Jalapeño」(ハラペーニョ)を発表した。米Broadcomとの共同開発...",{"title":86,"url":87,"domain":88,"snippet":89,"content":34},"OpenAI×Broadcom、初の自社推論チップ「Jalapeño」発表─9ヶ月開発でNvidia依存に風穴","https:\u002F\u002Finnovatopia.jp\u002Fsemiconductor\u002Fsemiconductornews\u002F110838\u002F","innovatopia.jp","通常なら数年かかる先端チップの開発を、わずか9ヶ月で。しかも、その設計をAI自身が手伝った——。OpenAIとBroadcomが披露した推論専用チップ「Jalapeño」は、単なる新...",{"title":91,"url":92,"domain":93,"snippet":94,"content":34},"ニュース OpenAI、LLMに最適化したチップ「Jalapeño」をBroadcomと開発","https:\u002F\u002Fai.watch.impress.co.jp\u002Fdocs\u002Fnews\u002F2119900.html","ai.watch.impress.co.jp","OpenAIとBroadcomは6月24日、OpenAI初のインテリジェンスプロセッサーとなる「Jalapeño（ハラペーニョ）」を発表した。 Jalapeñoは、LLM（大規模言語モデル）の処理に最適...",{"title":96,"url":97,"domain":98,"snippet":99,"content":34},"OpenAI, Broadcom debut custom Jalapeño chip for AI inference","https:\u002F\u002Fsiliconangle.com\u002F2026\u002F06\u002F24\u002Fopenai-broadcom-debut-custom-jalapeno-chip-llm-inference\u002F","siliconangle.com","OpenAI Group PBC today revealed a custom chip called Jalapeño that it will use to power its large language models. The processor is the fruit of a...",{"title":101,"url":102,"domain":103,"snippet":104,"content":34},"Broadcom, OpenAI Unveil Jalapeño AI Inference Chip","https:\u002F\u002Fwww.thelec.net\u002Fnews\u002FarticleView.html?idxno=11692","thelec.net","Broadcom and OpenAI on June 24 (local time) unveiled Jalapeño, a custom artificial intelligence (AI) accelerator jointly developed for large language model...",{"title":106,"url":107,"domain":108,"snippet":109,"content":34},"OpenAI Turns Up the Heat With Jalapeño, Its First Custom AI Chip","https:\u002F\u002Fdecrypt.co\u002F371971\u002Fopenai-broadcom-jalapeno-first-custom-ai-chip","decrypt.co","Built with Broadcom, the LLM-focused accelerator is the first step in OpenAI's effort to design the hardware behind ChatGPT and beyond.",{"mentionsLast7Days":111,"mentionsLast30Days":111,"firstSeen":21,"lastSeen":112,"relatedEntities":113},2,"2026-06-26T01:27:16.345Z",[25,31,37,42,47,52,57,62,67,72],[115,118,121,124,127,130],{"topic":116,"slug":117,"score":17,"type":18,"country":14,"nicheIcon":13},"Anthropic shuts newest models after US foreign-use ban","anthropic-shuts-newest-models-after-us-foreign-use-ban",{"topic":119,"slug":120,"score":17,"type":18,"country":14,"nicheIcon":13},"Anthropic shuts latest models after US bans foreign use","anthropic-shuts-latest-models-after-us-bans-foreign-use",{"topic":122,"slug":123,"score":17,"type":18,"country":14,"nicheIcon":13},"development of autonomous medical AI agents for clinical use","development-of-autonomous-medical-ai-agents-for-clinical-use",{"topic":125,"slug":126,"score":17,"type":18,"country":14,"nicheIcon":13},"Development and prospects of autonomous medical AI agents","development-and-prospects-of-autonomous-medical-ai-agents",{"topic":128,"slug":129,"score":17,"type":18,"country":14,"nicheIcon":13},"Open-source multilingual AI model for all 24 EU languages","open-source-multilingual-ai-model-for-all-24-eu-languages",{"topic":131,"slug":132,"score":17,"type":133,"country":14,"nicheIcon":13},"LLM-based autonomous medical AI agents for clinical decision-making","llm-based-autonomous-medical-ai-agents-for-clinical-decision-making","emerging"]