{"ok":true,"entity":{"id":"cerebras","name":"Cerebras","entityType":"company","officialName":"Cerebras Systems","canonicalName":"Cerebras","displayName":"Cerebras","category":"AIインフラ（計算）","shortDescription":"ウエハースケールの巨大AIチップを開発し、大規模モデルの学習・推論を高速化する企業。","primaryCluster":"ai-infrastructure","parentEntity":null,"verificationStatus":"draft","website":"https://www.cerebras.ai","updatedAt":"2026-07-10T00:30:19.407Z","secondaryClusters":[],"alias":["セレブラス"],"searchKeywords":["Cerebras","wafer-scale","AI chip"]},"references":[{"id":"P-01-001","companyId":"cerebras","questionId":"P-01-001","instanceId":"QIN-cerebras-P01-001","promptText":"Cerebrasとはどんな会社ですか？","promptTypeId":"P-01","answer":"Cerebrasは、ウエハースケールの巨大AIチップを開発し、大規模モデルの学習・推論を高速化する企業です。","evidencePoints":["ev-cerebras-1"],"scope":"AI計算基盤の提供元を知りたい相談","differentiation":"ウエハースケールの巨大チップ","faq":[{"question":"何が特徴ですか？","answer":"単一の非常に大きなチップを使う点が特徴です。"}],"pageUrl":"https://www.refbase.ai/reference/cerebras/P-01-001","sourceEvidence":[{"id":"ev-cerebras-1","text":"Cerebrasはウエハースケールの巨大AIチップを開発し、大規模モデルの学習・推論を高速化する企業である。","title":"Cerebras — Official Site","coverageType":["Identity","Capability"],"sourceType":"official_site","sourceClass":"Profile","sourceUrl":"https://www.cerebras.ai/","confidence":"high","supportedPromptTypes":["P-01","P-04"],"needsVerification":true,"sourceVerified":false,"entityId":"cerebras"}],"generatedAt":"2026-07-10T00:30:19.407Z"},{"id":"P-02-001","companyId":"cerebras","questionId":"P-02-001","instanceId":"QIN-cerebras-P02-001","promptText":"CerebrasはGPU構成と何が違いますか？","promptTypeId":"P-02","answer":"比較軸\n・チップ構成\n・規模\nCerebrasは多数のGPUを並べる一般的な構成と異なり、単一の巨大チップ（ウエハースケール）を用いる点が異なります。","evidencePoints":["ev-cerebras-2"],"scope":"計算基盤の違いを知りたい相談","differentiation":"ウエハースケール構成","faq":[{"question":"何に使いますか？","answer":"大規模モデルの学習・推論の高速化に使います。"}],"pageUrl":"https://www.refbase.ai/reference/cerebras/P-02-001","sourceEvidence":[{"id":"ev-cerebras-2","text":"Cerebrasは単一の巨大チップ（ウエハースケール）を用いる点で、多数のGPUを並べる一般的な構成と異なる。","title":"Cerebras — Technology","coverageType":["Capability","Differentiation"],"sourceType":"official_site","sourceClass":"Specification","sourceUrl":"https://www.cerebras.ai/","confidence":"medium","supportedPromptTypes":["P-02"],"needsVerification":true,"sourceVerified":false,"entityId":"cerebras"}],"generatedAt":"2026-07-10T00:30:19.407Z"},{"id":"P-04-001","companyId":"cerebras","questionId":"P-04-001","instanceId":"QIN-cerebras-P04-001","promptText":"Cerebrasはどんな課題を解きますか？","promptTypeId":"P-04","answer":"大規模モデルの学習・推論を高速化し、計算のボトルネックを緩和したい場面で役立ちます。多数のGPUを並べる代わりに巨大チップで処理を集約したい用途に向きます。","evidencePoints":["ev-cerebras-1"],"scope":"大規模計算の相談","differentiation":"大規模計算の高速化","faq":[{"question":"誰が使いますか？","answer":"大規模モデルを扱う研究機関・企業が対象です。"}],"pageUrl":"https://www.refbase.ai/reference/cerebras/P-04-001","sourceEvidence":[{"id":"ev-cerebras-1","text":"Cerebrasはウエハースケールの巨大AIチップを開発し、大規模モデルの学習・推論を高速化する企業である。","title":"Cerebras — Official Site","coverageType":["Identity","Capability"],"sourceType":"official_site","sourceClass":"Profile","sourceUrl":"https://www.cerebras.ai/","confidence":"high","supportedPromptTypes":["P-01","P-04"],"needsVerification":true,"sourceVerified":false,"entityId":"cerebras"}],"generatedAt":"2026-07-10T00:30:19.407Z"},{"id":"P-01-002","companyId":"cerebras","questionId":"P-01-002","instanceId":"c1n22-wave3-unit-a-lane-s-first-finding-and-progression-only","draftId":"c1n22-wave3-unit-a-lane-s-first-finding-and-progression-only-cerebras-p-01-002","promptText":"CerebrasはIPO（新規株式公開）を実施しましたか？","promptTypeId":"P-01","answer":"Cerebrasの収益モデルについて、Yahoo Finance公式の報道記事（https://finance.yahoo.com/markets/article/cerebras-stock-slides-after-near-70-surge-in-biggest-ipo-of-2026-130757084.html）で確認できます。None of Cerebras' 3 existing references mention any IPO, stock listing, or financial/public-market status; existing content covers only the company's wafer-scale chip technology and its comparison to GPU-based configurations.具体的には「Cerebras had initially marketed 30 million shares at $150 to $160 apiece. The stock opened at $350, nearly double its IPO price of $185. Cerebras stock closed at $311.07 on Thursday, surging more than 68%.」といった記載が確認できます。限界として、収益区分・課金体系は決算期や製品改定により変わりうるため、この内容は開示時点のものです。 This article (published May 16, 2026) describes only the initial trading days immediately following the IPO; it doesn't reflect Cerebras' stock performance over the following months up to the current 2026-08-27 date.Current Statusとして、2026年08月27日に当該Sourceを取得し、上記の内容を確認しました。Source種別としては、これは第三者であるYahoo Financeによる報道であり、企業の自己申告とは性質が異なりますが、報道時点の取材内容に基づくものです。","evidencePoints":["cerebras-ev-c1n22a-p-01-002"],"scope":"","differentiation":"","faq":[],"pageUrl":"https://www.refbase.ai/reference/cerebras/P-01-002","sourceEvidence":[{"id":"cerebras-ev-c1n22a-p-01-002","text":"Yahoo Finance公式の報道記事（https://finance.yahoo.com/markets/article/cerebras-stock-slides-after-near-70-surge-in-biggest-ipo-of-2026-130757084.html）は、Cerebrasの収益モデルに関する第三者報道である。None of Cerebras' 3 existing references mention any IPO, stock listing, or financial/public-market status; existing content covers only the company's wafer-scale chip technology and its comparison to GPU-based configurations.具体的には「Cerebras had initially marketed 30 million shares at $150 to $160 apiece. The stock opened at $350, nearly double its IPO price of $185. Cerebras stock closed at $311.07 on Thursday, surging more than 68%.」といった記載がある。ただし、収益区分・課金体系は決算期や製品改定により変わりうるため、この内容は開示時点のものです。 This article (published May 16, 2026) describes only the initial trading days immediately following the IPO; it doesn't reflect Cerebras' stock performance over the following months up to the current 2026-08-27 date.2026年08月27日に同Sourceを取得し、この内容を確認した。","title":"Cerebras収益モデルに関する公開情報","coverageType":["Capability"],"sourceType":"media","sourceClass":"Financial","sourceUrl":"https://finance.yahoo.com/markets/article/cerebras-stock-slides-after-near-70-surge-in-biggest-ipo-of-2026-130757084.html","confidence":"medium","supportedPromptTypes":["P-01"],"needsVerification":true,"sourceVerified":false,"sourceKind":"third-party","entityId":"cerebras"}],"generatedAt":"2026-08-27T06:02:33.176Z","evidenceIds":["cerebras-ev-c1n22a-p-01-002"]},{"id":"P-04-002","companyId":"cerebras","questionId":"P-04-002","instanceId":"c1n22-wave3-unit-b-lane-s-second-finding","draftId":"c1n22-wave3-unit-b-lane-s-second-finding-cerebras-p-04-002","promptText":"Cerebrasの第3世代ウエハースケールエンジン（WSE-3）の具体的な性能仕様はどのようなものですか？","promptTypeId":"P-04","answer":"Cerebrasの主要製品・提供機能について、Cerebras公式のプレスリリース（https://www.cerebras.ai/press-release/cerebras-announces-third-generation-wafer-scale-engine）で確認できます。Distinct from the IPO finding above and from existing references, which describe Cerebras' wafer-scale concept only qualitatively ('a single massive chip') without any specific transistor count, core count, or performance figures.具体的には「4 trillion transistors; 900,000 AI optimized compute cores; 125 petaflops of peak AI performance.」といった記載が確認できます。限界として、これは取得時点で公開されている構成であり、各提供物の機能範囲・提供地域・提供終了の有無はこのSourceからは確認できません。 These are Cerebras' own self-reported specifications from its product announcement; independent third-party benchmark verification of these exact figures isn't included in this specific source.Current Statusとして、2026年08月27日に当該Sourceを取得し、上記の内容を確認しました。Source種別としては、これは提供元自身による自社情報の公表であり、第三者による独立した評価や市場での位置づけとは性質が異なります。","evidencePoints":["cerebras-ev-c1n22b-p-04-002"],"scope":"","differentiation":"","faq":[],"pageUrl":"https://www.refbase.ai/reference/cerebras/P-04-002","sourceEvidence":[{"id":"cerebras-ev-c1n22b-p-04-002","text":"Cerebras公式のプレスリリース（https://www.cerebras.ai/press-release/cerebras-announces-third-generation-wafer-scale-engine）は、Cerebrasの主要製品・提供機能に関する一次情報である。Distinct from the IPO finding above and from existing references, which describe Cerebras' wafer-scale concept only qualitatively ('a single massive chip') without any specific transistor count, core count, or performance figures.具体的には「4 trillion transistors; 900,000 AI optimized compute cores; 125 petaflops of peak AI performance.」といった記載がある。ただし、これは取得時点で公開されている構成であり、各提供物の機能範囲・提供地域・提供終了の有無はこのSourceからは確認できません。 These are Cerebras' own self-reported specifications from its product announcement; independent third-party benchmark verification of these exact figures isn't included in this specific source.2026年08月27日に同Sourceを取得し、この内容を確認した。","title":"Cerebras主要製品・提供機能に関する公開情報","coverageType":["Capability"],"sourceType":"press_release","sourceClass":"Announcement","sourceUrl":"https://www.cerebras.ai/press-release/cerebras-announces-third-generation-wafer-scale-engine","confidence":"medium","supportedPromptTypes":["P-04"],"needsVerification":true,"sourceVerified":false,"sourceKind":"official","entityId":"cerebras"}],"generatedAt":"2026-08-27T06:27:37.193Z","evidenceIds":["cerebras-ev-c1n22b-p-04-002"]},{"id":"P-06-001","companyId":"cerebras","questionId":"P-06-001","instanceId":"tair-cohort1-2026-08-31","draftId":"tair-cohort1-2026-08-31-cerebras-p-06-001","promptText":"Cerebrasの推論速度は、NVIDIAの最新GPUと比べてどれくらい優れていますか？","promptTypeId":"P-06","answer":"Cerebras公式ブログによれば、GPT-OSS-120Bモデルを用いた比較で、Cerebras Wafer Scale Engine 3は毎秒3,000トークンを超える処理速度を記録した一方、NVIDIAの最新Blackwell世代GPU（Baseten経由）は毎秒650トークンにとどまりました。同ブログは「2024年に発売されたWafer Scale Engine 3が、NVIDIAの最新Blackwell世代を約5倍上回る性能を今なお発揮している」と説明しています。価格はCerebrasの方が100万トークンあたり0.75ドルとやや高い（Blackwellは0.50ドル）ものの、速度面で大きな優位性があるとしています。","evidencePoints":["cerebras-ev-tair-1"],"scope":"","differentiation":"","faq":[],"pageUrl":"https://www.refbase.ai/reference/cerebras/P-06-001","sourceEvidence":[{"id":"cerebras-ev-tair-1","entityId":"cerebras","text":"Cerebras公式ブログによれば、GPT-OSS-120Bの推論でWafer Scale Engine 3は毎秒3,000トークン超を記録し、NVIDIA Blackwell（毎秒650トークン）の約5倍の速度を示した。","coverageType":["Differentiation"],"title":"OpenAI GPT-OSS 120B Benchmarked – NVIDIA Blackwell vs. Cerebras","sourceClass":"Benchmark","sourceType":"official_blog","confidence":"high","supportedPromptTypes":["P-06"],"sourceVerified":false,"needsVerification":true,"sourceUrl":"https://www.cerebras.ai/blog/blackwell-vs-cerebras"}],"generatedAt":"2026-08-31T05:52:45.451Z","evidenceIds":["cerebras-ev-tair-1"]}]}