[{"data":1,"prerenderedAt":5430},["ShallowReactive",2],{"category-data-半導体":3},[4,1737,3850],{"id":5,"title":6,"body":7,"description":1714,"extension":1715,"meta":1716,"navigation":1732,"ogImage":1718,"path":1733,"seo":1734,"stem":1735,"__hash__":1736},"content/blogs/6-semiconductor-vc-series-1.md","Tier1 VCの半導体投資はAI需要のボトルネックへ収斂している｜連載①概観と主要プレイヤー | Semiconductor VC Series①: Converging on AI Bottlenecks",{"type":8,"value":9,"toc":1681},"minimark",[10,871],[11,12,14,19,24,27,48,54,57,60,63,85,87,90,158,161,207,210,212,216,219,230,233,235,239,346,348,352,357,360,363,389,395,397,401,404,406,420,425,427,431,434,436,450,455,457,461,464,466,492,497,499,503,506,508,528,531,536,538,541,848,850,854],"lang-block",{"lang":13},"ja",[15,16,18],"h1",{"id":17},"tier1-vcの半導体投資はai需要のボトルネックへ収斂している","Tier1 VCの半導体投資はAI需要のボトルネックへ収斂している",[20,21,23],"h2",{"id":22},"連載概観と主要プレイヤー","——連載①：概観と主要プレイヤー",[20,25,26],{"id":26},"はじめに",[28,29,30,31,35,36,39,40,43,44,47],"p",{},"2025年、世界半導体市場は過去最高となる ",[32,33,34],"strong",{},"$791.7B（WSTS）"," に達した。2026年には ",[32,37,38],{},"$975B"," への拡大が見込まれ、AIインフラへの投資が市場全体を牽引している。同時に、300mmファブ装置の支出は2026年に ",[32,41,42],{},"$133B","、2027年には ",[32,45,46],{},"$151B"," に達するとSEMIは予測する。数字だけ見れば「半導体ブーム」に見えるが、Tier1 VCが実際に賭けているのはチップそのものではない。",[28,49,50,53],{},[32,51,52],{},"計算能力の向上が生み出す「ボトルネック」","——設計スループット、データ移動（インターコネクト）、パッケージング、そして次世代の計算原理——こそが、トップVCの投資が集まる場所だ。",[28,55,56],{},"この連載では、Sequoia、Andreessen Horowitz（a16z）、Kleiner Perkins、Bessemer Venture Partners、SoftBank Vision Fund/Groupの5社が半導体産業スタック全体にどう張っているかを、公開情報に基づいて分析する。",[58,59],"hr",{},[20,61,62],{"id":62},"エグゼクティブサマリ",[64,65,66,73,79],"ul",{},[67,68,69,72],"li",{},[32,70,71],{},"成長ドライバーはAIインフラ。"," 市場規模の拡大と装置投資の増加は「供給制約が緩む」より「投資負荷が続く」シグナルに近い。スタートアップ側からすれば、量産・テスト能力の確保が評価の分岐点になる。",[67,74,75,78],{},[32,76,77],{},"VCの差分は「ステージの取り方」と「提携構造」。"," 巨大シードでアーキテクチャ賭けに出る（a16z）、設計自動化で開発リードタイム短縮（Sequoia/Bessemer）、M&Aで垂直統合（SoftBank）——資本政策が根本的に異なる。",[67,80,81,84],{},[32,82,83],{},"地理は依然米国中心。"," だが、インドのRISC-V/IP、欧州のAIアクセラレータ/フォトニクス、イスラエルのAIチップなど、サプライチェーン分散に沿った投資が可視化されている。",[58,86],{},[20,88,89],{"id":89},"世界半導体市場の現在地",[91,92,93,112],"table",{},[94,95,96],"thead",{},[97,98,99,103,106,109],"tr",{},[100,101,102],"th",{},"年",[100,104,105],{},"市場規模",[100,107,108],{},"前年比",[100,110,111],{},"主な成長領域",[113,114,115,130,144],"tbody",{},[97,116,117,121,124,127],{},[118,119,120],"td",{},"2024",[118,122,123],{},"$611.2B",[118,125,126],{},"—",[118,128,129],{},"メモリ回復、ロジック",[97,131,132,135,138,141],{},[118,133,134],{},"2025",[118,136,137],{},"$791.7B",[118,139,140],{},"+29.4%",[118,142,143],{},"AIデータセンター、HBM",[97,145,146,149,152,155],{},[118,147,148],{},"2026（予測）",[118,150,151],{},"$975.0B",[118,153,154],{},"+23.2%",[118,156,157],{},"AIインフラ継続拡大",[28,159,160],{},"装置投資（300mmファブ・SEMI予測）：",[91,162,163,175],{},[94,164,165],{},[97,166,167,169,172],{},[100,168,102],{},[100,170,171],{},"装置支出",[100,173,174],{},"背景",[113,176,177,187,197],{},[97,178,179,181,184],{},[118,180,134],{},[118,182,183],{},"〜$120B",[118,185,186],{},"AI需要立ち上がり",[97,188,189,192,194],{},[118,190,191],{},"2026",[118,193,42],{},[118,195,196],{},"AI需要×各国供給網再編",[97,198,199,202,204],{},[118,200,201],{},"2027",[118,203,46],{},[118,205,206],{},"供給投資の継続",[28,208,209],{},"この拡大は「半導体全体が儲かる」ではなく、Memory/Logicなど特定カテゴリに偏りやすい。EDA・テスト・パッケージングといった周辺領域は、ボトルネックとして間接的に恩恵を受ける構造だ。",[58,211],{},[20,213,215],{"id":214},"aiインフラのボトルネック構造","AIインフラのボトルネック構造",[28,217,218],{},"GPUやアクセラレータの演算能力が伸び続けるほど、「その周辺」が限界要因になる。",[220,221,226],"pre",{"className":222,"code":224,"language":225},[223],"language-text","AIインフラ需要の増加\n        │\n        ▼ ボトルネックの再定義\n┌───────────┬─────────────┬──────────────┬────────────┐\n│ 設計生産性 │ データ移動   │ パッケージング │ 新計算原理─ │\n│ EDA/検証  │SerDes/光I/O │ CPO/Chiplet  │アナログ/量子─│\n└────┬─────┴─────┬───────┴───────┬──────┴─────┬───────┘\n     ↓           ↓               ↓            ↓\n シード〜A     Series B〜E     レイト          長期\n ソフト比率高  顧客/製造連携    資本集約     技術/標準不確実\n","text",[227,228,224],"code",{"__ignoreMap":229},"",[28,231,232],{},"VCはこの4つの「制約領域」にリスク分散しながらポジションを取っている。",[58,234],{},[20,236,238],{"id":237},"vc別の型比較","VC別の\"型\"比較",[91,240,241,260],{},[94,242,243],{},[97,244,245,248,251,254,257],{},[100,246,247],{},"VC",[100,249,250],{},"重点技術領域",[100,252,253],{},"ステージ傾向",[100,255,256],{},"地理",[100,258,259],{},"提携・LP構造",[113,261,262,279,296,312,329],{},[97,263,264,267,270,273,276],{},[118,265,266],{},"Sequoia",[118,268,269],{},"設計AI（EDA/検証）、RISC-V/IP、CPO",[118,271,272],{},"Seed〜Growth",[118,274,275],{},"米国・インド",[118,277,278],{},"設計AI領域に公式コミット。関連ビークルがCPOに参加",[97,280,281,284,287,290,293],{},[118,282,283],{},"a16z",[118,285,286],{},"新計算（アナログ/混載）、AIネットワーク",[118,288,289],{},"巨大Seed＋大型B",[118,291,292],{},"米国",[118,294,295],{},"Strategic Partnershipを明示、GTM設計が特徴",[97,297,298,301,304,307,309],{},[118,299,300],{},"Kleiner Perkins",[118,302,303],{},"RISC-V/IP、コヒレントDSP",[118,305,306],{},"Early〜Series D",[118,308,292],{},[118,310,311],{},"TSMC支援ファンド報道など「供給側との接続」示唆",[97,313,314,317,320,323,326],{},[118,315,316],{},"Bessemer",[118,318,319],{},"SerDes/IP、量子、EDA自動化",[118,321,322],{},"Series C中心＋アーリー",[118,324,325],{},"米国・欧州・イスラエル",[118,327,328],{},"AIチップ/量子/EDAで「深技術の商用化」",[97,330,331,334,337,340,343],{},[118,332,333],{},"SoftBank",[118,335,336],{},"AIチップ/CPU/DPU、垂直統合",[118,338,339],{},"Growth〜M&A",[118,341,342],{},"米国・英国",[118,344,345],{},"Arm核のAI Computing Segment。LPに戦略資本（Apple等）",[58,347],{},[20,349,351],{"id":350},"各vc解説","各VC解説",[353,354,356],"h3",{"id":355},"sequoia設計運用ボトルネックを連結する","Sequoia：設計→運用ボトルネックを連結する",[28,358,359],{},"Sequoiaは「AIが高性能化するほど、チップ設計のスループットがボトルネックになる」という仮説を前面に出す。",[28,361,362],{},"代表案件：",[64,364,365,371,377,383],{},[67,366,367,370],{},[32,368,369],{},"Ricursive Intelligence（$35M、2025年シード、リード）","：AIによるチップ設計加速。Sequoiaが初回ラウンドをリードした象徴的案件。",[67,372,373,376],{},[32,374,375],{},"InCore Semiconductors（$3M、2023年シード）","：インドのRISC-V IP。現地の産業政策×設計IPの立ち上がりへの早期賭け。",[67,378,379,382],{},[32,380,381],{},"Mindgrove Technologies（約$2.3M、2023年シード）","：インドSoCスタートアップ。同様にRISC-V文脈。",[67,384,385,388],{},[32,386,387],{},"Ayar Labs（$500M、2026年Series E、関連ビークル参加）","：Co-Packaged Optics（CPO）。NVIDIA・AMD等の戦略投資家と共に参加。評価額$3.75B。",[28,390,391,394],{},[32,392,393],{},"読み解き","：シード段階で「設計の生産性」を押さえつつ、大型レイトでは「次のデータ移動（光I/O）」にも資本を延ばす。設計から運用ボトルネックまでを連結する投資設計になっている。",[58,396],{},[353,398,400],{"id":399},"a16z計算原理そのものへの巨大シード","a16z：計算原理そのものへの巨大シード",[28,402,403],{},"a16zは「計算原理を変えるような勝ち筋に、研究開発段階から大口でアンダーライトする」のが特徴だ。",[28,405,362],{},[64,407,408,414],{},[67,409,410,413],{},[32,411,412],{},"Unconventional AI（$475M、2025年シード、共同リード）","：アナログ/ミックスドシグナルで確率的計算に最適化。製造難易度は高いが、実現すれば電力効率の桁違い改善を狙える。",[67,415,416,419],{},[32,417,418],{},"Nexthop AI（$500M、2026年Series B）","：AIクラスタ向けイーサネットスイッチ。スケールアウトのボトルネックがネットワークに移ったという認識を明示。",[28,421,422,424],{},[32,423,393],{},"：同じAIインフラ投資でも、a16zは「カテゴリを再定義できる」と判断したらシード段階から数百億円を投下する。Strategic Partnerships（顧客・公共・大企業への導入経路の制度化）を明示している点も特徴で、資本だけでなくGTMを設計する色が濃い。",[58,426],{},[353,428,430],{"id":429},"kleiner-perkins水平方向のipデータセンター光","Kleiner Perkins：\"水平方向\"のIP×データセンター光",[28,432,433],{},"Kleiner PerkinsはAIインフラの水平方向——IP（RISC-V）とデータセンターの光/インターコネクト——に選球眼が集まっている。",[28,435,362],{},[64,437,438,444],{},[67,439,440,443],{},[32,441,442],{},"Akeana（累計$100M超、2024年ステルス解除）","：RISC-V CPU/IPポートフォリオ。Kleiner Perkins等が支援。",[67,445,446,449],{},[32,447,448],{},"Retym（$75M、2025年Series D）","：クラウド/AI向けコヒレントDSP。「距離×帯域×消費電力」の実制約に沿った設計。",[28,451,452,454],{},[32,453,393],{},"：ハイパースケーラーを顧客とする領域で「設計の差別化（IP/DSP）」を取りにいく投資スタイル。TSMC支援ファンドの立ち上げ報道は、「資金×ファブ供給（エコシステム）」の結びつきが強まる流れを示唆している。",[58,456],{},[353,458,460],{"id":459},"bessemer深技術の商用化を組み合わせて張る","Bessemer：深技術の商用化を組み合わせて張る",[28,462,463],{},"BessemerはSerDes/IP、量子、EDA自動化という一見バラバラな領域を、「深い技術負債と長いR&Dサイクルを伴うが、当たれば大きい」という共通軸で捉えている。",[28,465,362],{},[64,467,468,474,480,486],{},[67,469,470,473],{},[32,471,472],{},"ChipAgents（$21M、2025年Series A、リード）","：設計・検証のAIエージェント化。EDA工程の自動化。",[67,475,476,479],{},[32,477,478],{},"Kandou（$92.3M、2020年Series C、リード）","：SerDes/IPとリタイマー。データ移動系の商用化を前提にした資本投下。",[67,481,482,485],{},[32,483,484],{},"Rigetti Computing（$79M、2020年Series C、リード）","：量子ハードウェア+クラウド提供。産業化フェーズへの投資。",[67,487,488,491],{},[32,489,490],{},"Habana Labs（$75M Series Bに参加、2018年、Intelが2019年に約$2B で買収）","：AIチップ。大手コーポレートと同じラウンドに入る投資スタイル。",[28,493,494,496],{},[32,495,393],{},"：イスラエルネットワーク（Habana Labs）、スイス（Kandou）など、地域性も活用しながら「深技術×商用化の道筋」が立つ案件を選ぶ。",[58,498],{},[353,500,502],{"id":501},"softbank-vision-fundgroup垂直統合という別次元のリスクテイク","SoftBank Vision Fund/Group：垂直統合という別次元のリスクテイク",[28,504,505],{},"SoftBankは他のTier1 VCと決定的に異なる。ファンド投資だけでなく、買収を含む「垂直統合（Arm×周辺シリコン）」へ踏み込む。",[28,507,362],{},[64,509,510,516,522],{},[67,511,512,515],{},[32,513,514],{},"Fungible（$200M、2019年Series C、SVFリード）","：DPU（データ処理ユニット）。後にMicrosoft傘下へ。",[67,517,518,521],{},[32,519,520],{},"Graphcore（2024年買収、金額非開示）","：英国AIチップ企業。",[67,523,524,527],{},[32,525,526],{},"Ampere Computing（$6.5B買収、2025年発表）","：ArmベースのクラウドサーバーCPU。",[28,529,530],{},"AI Computing SegmentとしてArm・Ampere・Graphcoreを同一セグメントで運営し、半導体事業を中核に据える姿勢を公式に明確化している。2017年のファンド初回クロージングでは、PIF・Mubadala・Apple・Foxconn・Qualcomm・Sharpといった戦略資本がLPとして参画しており、資本構造レベルで「産業としての半導体」に深く組み込まれている。",[28,532,533,535],{},[32,534,393],{},"：これはVCというより「戦略金融×事業会社」の設計に近い。リスクテイクの桁も期間も通常のVCファンドとは異なる。",[58,537],{},[20,539,540],{"id":540},"代表投資案件タイムライン",[91,542,543,563],{},[94,544,545],{},[97,546,547,549,551,554,557,560],{},[100,548,102],{},[100,550,247],{},[100,552,553],{},"投資先",[100,555,556],{},"ラウンド",[100,558,559],{},"金額（$M）",[100,561,562],{},"技術領域",[113,564,565,584,603,621,638,657,674,691,708,727,744,761,779,796,813,832],{},[97,566,567,570,572,575,578,581],{},[118,568,569],{},"2018",[118,571,316],{},[118,573,574],{},"Habana Labs",[118,576,577],{},"Series B（参加）",[118,579,580],{},"75",[118,582,583],{},"AIアクセラレータ",[97,585,586,589,591,594,597,600],{},[118,587,588],{},"2019",[118,590,333],{},[118,592,593],{},"Fungible",[118,595,596],{},"Series C",[118,598,599],{},"200",[118,601,602],{},"DPU",[97,604,605,608,610,613,615,618],{},[118,606,607],{},"2020",[118,609,316],{},[118,611,612],{},"Kandou",[118,614,596],{},[118,616,617],{},"92.3",[118,619,620],{},"SerDes/IP",[97,622,623,625,627,630,632,635],{},[118,624,607],{},[118,626,316],{},[118,628,629],{},"Rigetti",[118,631,596],{},[118,633,634],{},"79",[118,636,637],{},"量子",[97,639,640,643,645,648,651,654],{},[118,641,642],{},"2023",[118,644,266],{},[118,646,647],{},"InCore Semiconductors",[118,649,650],{},"Seed",[118,652,653],{},"3",[118,655,656],{},"RISC-V IP",[97,658,659,661,663,666,668,671],{},[118,660,642],{},[118,662,266],{},[118,664,665],{},"Mindgrove Technologies",[118,667,650],{},[118,669,670],{},"2.3",[118,672,673],{},"SoC",[97,675,676,678,680,683,686,689],{},[118,677,120],{},[118,679,266],{},[118,681,682],{},"Quantum Circuits",[118,684,685],{},"Series B",[118,687,688],{},">60",[118,690,637],{},[97,692,693,695,697,700,703,706],{},[118,694,120],{},[118,696,300],{},[118,698,699],{},"Akeana",[118,701,702],{},"累計",[118,704,705],{},">100",[118,707,656],{},[97,709,710,712,715,718,721,724],{},[118,711,120],{},[118,713,714],{},"SoftBank Group",[118,716,717],{},"Graphcore",[118,719,720],{},"M&A",[118,722,723],{},"非開示",[118,725,726],{},"AIチップ",[97,728,729,731,733,736,738,741],{},[118,730,134],{},[118,732,266],{},[118,734,735],{},"Ricursive Intelligence",[118,737,650],{},[118,739,740],{},"35",[118,742,743],{},"設計AI/EDA",[97,745,746,748,750,753,755,758],{},[118,747,134],{},[118,749,283],{},[118,751,752],{},"Unconventional AI",[118,754,650],{},[118,756,757],{},"475",[118,759,760],{},"アナログ新計算",[97,762,763,765,767,770,773,776],{},[118,764,134],{},[118,766,316],{},[118,768,769],{},"ChipAgents",[118,771,772],{},"Series A",[118,774,775],{},"21",[118,777,778],{},"EDA自動化",[97,780,781,783,785,788,791,793],{},[118,782,134],{},[118,784,300],{},[118,786,787],{},"Retym",[118,789,790],{},"Series D",[118,792,580],{},[118,794,795],{},"コヒレントDSP",[97,797,798,800,802,805,807,810],{},[118,799,134],{},[118,801,714],{},[118,803,804],{},"Ampere Computing",[118,806,720],{},[118,808,809],{},"6,500",[118,811,812],{},"Arm CPU",[97,814,815,817,820,823,826,829],{},[118,816,191],{},[118,818,819],{},"Sequoia（関連）",[118,821,822],{},"Ayar Labs",[118,824,825],{},"Series E",[118,827,828],{},"500",[118,830,831],{},"CPO/光I/O",[97,833,834,836,838,841,843,845],{},[118,835,191],{},[118,837,283],{},[118,839,840],{},"Nexthop AI",[118,842,685],{},[118,844,828],{},[118,846,847],{},"AIスイッチ",[58,849],{},[20,851,853],{"id":852},"zyl0の視点","ZYL0の視点",[855,856,857,862,865,868],"blockquote",{},[28,858,859],{},[32,860,861],{},"半導体投資は「チップを買う」のではなく「AIの制約を買う」",[28,863,864],{},"今回のリサーチで一番印象に残ったのは、Tier1 VCがほぼ一様に「計算そのもの」ではなく「その周辺の制約」を狙っているという点だ。GPU性能が指数関数的に伸び続けるなら、設計・接続・実装・供給の各レイヤーが次々とボトルネックに変わっていく。",[28,866,867],{},"投資家としての自分の視点で言うと、日本からこの波に乗るとしたら、製造装置・材料・テスト（国内に強いプレイヤーが存在する）と、設計IP・EDA自動化（ソフトウェア寄りで参入障壁が低い）の2つの切り口が現実的ではないか。",[28,869,870],{},"次回（連載②）では、各技術領域の具体的な投資ロジックとケーススタディを深掘りする。",[11,872,874,878,882,900,907,910,912,916,936,938,942,996,999,1043,1046,1048,1052,1055,1061,1064,1066,1070,1171,1173,1177,1181,1188,1191,1217,1223,1225,1229,1236,1238,1252,1257,1259,1263,1270,1272,1286,1295,1297,1301,1308,1310,1336,1341,1343,1347,1353,1355,1375,1382,1387,1389,1393,1654,1656,1660],{"lang":873},"en",[15,875,877],{"id":876},"semiconductor-vc-series-converging-on-ai-bottlenecks","Semiconductor VC Series①: Converging on AI Bottlenecks",[20,879,881],{"id":880},"introduction","Introduction",[28,883,884,885,888,889,891,892,895,896,899],{},"In 2025, the global semiconductor market reached an all-time high of ",[32,886,887],{},"$791.7B (WSTS)",", with a forecast expansion to ",[32,890,38],{}," in 2026 driven by AI infrastructure investment. At the same time, SEMI projects 300mm fab equipment spending of ",[32,893,894],{},"$133B in 2026"," and ",[32,897,898],{},"$151B in 2027",". While the numbers suggest a semiconductor boom, the world's top VCs are not simply betting on chips.",[28,901,902,903,906],{},"They are betting on ",[32,904,905],{},"bottlenecks"," — design throughput, data movement (interconnect), packaging, and next-generation computing paradigms — the areas that emerge as limiting factors precisely because GPU compute keeps improving.",[28,908,909],{},"This series analyzes how Sequoia, Andreessen Horowitz (a16z), Kleiner Perkins, Bessemer Venture Partners, and SoftBank Vision Fund/Group position themselves across the full semiconductor stack, based on publicly available information.",[58,911],{},[20,913,915],{"id":914},"executive-summary","Executive Summary",[64,917,918,924,930],{},[67,919,920,923],{},[32,921,922],{},"AI infrastructure is the growth driver."," The rise in market size and equipment investment signals sustained capital intensity, not easing supply constraints. For startups, securing production and testing capability is increasingly the key evaluation criterion.",[67,925,926,929],{},[32,927,928],{},"The differentiation among VCs lies in stage selection and partnership structure."," Deploying massive seed rounds for architectural bets (a16z), improving design productivity (Sequoia/Bessemer), or pursuing vertical integration via M&A (SoftBank) — these represent fundamentally different capital strategies.",[67,931,932,935],{},[32,933,934],{},"Geography remains U.S.-centric but diversifying."," India (RISC-V/IP), Europe (AI accelerators/photonics), and Israel (AI chips) are becoming visible as supply chain diversification plays.",[58,937],{},[20,939,941],{"id":940},"global-semiconductor-market-snapshot","Global Semiconductor Market Snapshot",[91,943,944,960],{},[94,945,946],{},[97,947,948,951,954,957],{},[100,949,950],{},"Year",[100,952,953],{},"Market Size",[100,955,956],{},"YoY Growth",[100,958,959],{},"Key Drivers",[113,961,962,973,984],{},[97,963,964,966,968,970],{},[118,965,120],{},[118,967,123],{},[118,969,126],{},[118,971,972],{},"Memory recovery, logic",[97,974,975,977,979,981],{},[118,976,134],{},[118,978,137],{},[118,980,140],{},[118,982,983],{},"AI data centers, HBM",[97,985,986,989,991,993],{},[118,987,988],{},"2026 (est.)",[118,990,151],{},[118,992,154],{},[118,994,995],{},"Continued AI infrastructure",[28,997,998],{},"300mm equipment spending (SEMI forecast):",[91,1000,1001,1013],{},[94,1002,1003],{},[97,1004,1005,1007,1010],{},[100,1006,950],{},[100,1008,1009],{},"Equipment Spend",[100,1011,1012],{},"Context",[113,1014,1015,1025,1034],{},[97,1016,1017,1019,1022],{},[118,1018,134],{},[118,1020,1021],{},"~$120B",[118,1023,1024],{},"AI ramp",[97,1026,1027,1029,1031],{},[118,1028,191],{},[118,1030,42],{},[118,1032,1033],{},"AI demand + geopolitical supply reshaping",[97,1035,1036,1038,1040],{},[118,1037,201],{},[118,1039,46],{},[118,1041,1042],{},"Continued investment cycle",[28,1044,1045],{},"This expansion is concentrated in specific categories (Memory/Logic). Peripheral areas — EDA, test, packaging — benefit indirectly as bottlenecks, not as direct growth plays.",[58,1047],{},[20,1049,1051],{"id":1050},"the-ai-infrastructure-bottleneck-map","The AI Infrastructure Bottleneck Map",[28,1053,1054],{},"As GPU and accelerator compute continues to scale, the surrounding layers become the limiting factors:",[220,1056,1059],{"className":1057,"code":1058,"language":225},[223],"Growing AI Infrastructure Demand\n        │\n        ▼ Bottlenecks Redefined\n┌───────────────┬───────────────────────────┬────────────────────────┬──────────────────────┐\n│ Design Prod.  │        Data Movement      │       Packaging        │    New Computing     │\n│ EDA/Verif.    │        SerDes/Optics      │       CPO/Chiplet      │    Analog/Quantum    │\n└───────┬───────┴────────────┬──────────────┴─────────────┬──────────┴───────────┬──────────┘\n       ↓                     ↓                            ↓                     ↓\n Seed→A: high           Series B→E                       Late                  Long\n software ratio  Customer/Manufacturing Connection  capital intensive     tech/standard risk\n",[227,1060,1058],{"__ignoreMap":229},[28,1062,1063],{},"VCs diversify across these four constraint layers to spread risk while capturing upside.",[58,1065],{},[20,1067,1069],{"id":1068},"vc-strategy-comparison","VC Strategy Comparison",[91,1071,1072,1090],{},[94,1073,1074],{},[97,1075,1076,1078,1081,1084,1087],{},[100,1077,247],{},[100,1079,1080],{},"Focus Areas",[100,1082,1083],{},"Stage Preference",[100,1085,1086],{},"Geography",[100,1088,1089],{},"Partnership / LP Structure",[113,1091,1092,1108,1124,1139,1155],{},[97,1093,1094,1096,1099,1102,1105],{},[118,1095,266],{},[118,1097,1098],{},"Design AI (EDA/verification), RISC-V/IP, CPO",[118,1100,1101],{},"Seed through Growth",[118,1103,1104],{},"US, India",[118,1106,1107],{},"Publicly committed to design AI; related vehicle in CPO",[97,1109,1110,1112,1115,1118,1121],{},[118,1111,283],{},[118,1113,1114],{},"New compute (analog/mixed-signal), AI networking",[118,1116,1117],{},"Mega-seed + large Series B",[118,1119,1120],{},"US",[118,1122,1123],{},"Explicit Strategic Partnerships program for GTM",[97,1125,1126,1128,1131,1134,1136],{},[118,1127,300],{},[118,1129,1130],{},"RISC-V/IP, coherent DSP",[118,1132,1133],{},"Early through Series D",[118,1135,1120],{},[118,1137,1138],{},"TSMC-backed fund reports suggest supply-side ties",[97,1140,1141,1143,1146,1149,1152],{},[118,1142,316],{},[118,1144,1145],{},"SerDes/IP, quantum, EDA automation",[118,1147,1148],{},"Series C core + early",[118,1150,1151],{},"US, Europe, Israel",[118,1153,1154],{},"Deep tech commercialization across AI chips/quantum/EDA",[97,1156,1157,1159,1162,1165,1168],{},[118,1158,333],{},[118,1160,1161],{},"AI chips/CPU/DPU, vertical integration",[118,1163,1164],{},"Growth through M&A",[118,1166,1167],{},"US, UK",[118,1169,1170],{},"Strategic LPs (Apple, etc.) in Fund I; Arm-centered AI Computing Segment",[58,1172],{},[20,1174,1176],{"id":1175},"individual-vc-profiles","Individual VC Profiles",[353,1178,1180],{"id":1179},"sequoia-linking-design-to-operational-bottlenecks","Sequoia: Linking Design to Operational Bottlenecks",[28,1182,1183,1184,1187],{},"Sequoia leads with the thesis that as AI scales, ",[32,1185,1186],{},"chip design throughput becomes the bottleneck",".",[28,1189,1190],{},"Key deals:",[64,1192,1193,1199,1205,1211],{},[67,1194,1195,1198],{},[32,1196,1197],{},"Ricursive Intelligence ($35M, 2025 Seed, lead)",": AI-accelerated chip design. Sequoia-led founding round.",[67,1200,1201,1204],{},[32,1202,1203],{},"InCore Semiconductors ($3M, 2023 Seed)",": India RISC-V IP. Early bet on India industrial policy × design IP.",[67,1206,1207,1210],{},[32,1208,1209],{},"Mindgrove Technologies (~$2.3M, 2023 Seed)",": India SoC startup, same RISC-V context.",[67,1212,1213,1216],{},[32,1214,1215],{},"Ayar Labs ($500M, 2026 Series E, related vehicle)",": Co-Packaged Optics (CPO) for AI scale-out. Co-investors include NVIDIA and AMD.",[28,1218,1219,1222],{},[32,1220,1221],{},"Takeaway",": Sequoia plants seeds in design productivity, then extends capital to data movement (optical I/O) at the growth stage — connecting the design-to-operations bottleneck chain.",[58,1224],{},[353,1226,1228],{"id":1227},"a16z-mega-seeds-on-computing-paradigms","a16z: Mega-Seeds on Computing Paradigms",[28,1230,1231,1232,1235],{},"a16z's signature move is ",[32,1233,1234],{},"writing massive checks at the seed stage"," for companies that could redefine entire compute categories.",[28,1237,1190],{},[64,1239,1240,1246],{},[67,1241,1242,1245],{},[32,1243,1244],{},"Unconventional AI ($475M, 2025 Seed, co-lead)",": Analog/mixed-signal for probabilistic computation. High manufacturing risk, but potential for orders-of-magnitude power efficiency gains.",[67,1247,1248,1251],{},[32,1249,1250],{},"Nexthop AI ($500M, 2026 Series B)",": AI-era Ethernet switches. Explicitly framed around network becoming the cluster bottleneck.",[28,1253,1254,1256],{},[32,1255,1221],{},": Where others spread early-stage bets, a16z concentrates capital into paradigm-level bets with massive up-front rounds. The formal Strategic Partnerships program — institutionalizing GTM into government and enterprise — is a distinctive structural advantage.",[58,1258],{},[353,1260,1262],{"id":1261},"kleiner-perkins-ip-datacenter-optics","Kleiner Perkins: IP + Datacenter Optics",[28,1264,1265,1266,1269],{},"Kleiner Perkins gravitates toward the ",[32,1267,1268],{},"horizontal infrastructure layer",": IP (RISC-V) and datacenter optical/interconnect.",[28,1271,1190],{},[64,1273,1274,1280],{},[67,1275,1276,1279],{},[32,1277,1278],{},"Akeana (>$100M cumulative, 2024 stealth launch)",": RISC-V CPU/IP portfolio. Kleiner Perkins among listed backers.",[67,1281,1282,1285],{},[32,1283,1284],{},"Retym ($75M, 2025 Series D)",": Cloud/AI-optimized coherent DSP. Designed around real-world \"distance × bandwidth × power\" constraints.",[28,1287,1288,1290,1291,1294],{},[32,1289,1221],{},": Target companies that win through ",[32,1292,1293],{},"design differentiation"," (IP/DSP) in markets where the customer base is hyperscalers. The reported TSMC-backed fund suggests deepening supply-side integration.",[58,1296],{},[353,1298,1300],{"id":1299},"bessemer-commercializing-deep-tech","Bessemer: Commercializing Deep Tech",[28,1302,1303,1304,1307],{},"Bessemer treats semiconductor investment as a portfolio of ",[32,1305,1306],{},"SerDes/IP + quantum + EDA automation",", unified by the thesis that deep technical moats can be commercialized if the path to market is clear.",[28,1309,1190],{},[64,1311,1312,1318,1324,1330],{},[67,1313,1314,1317],{},[32,1315,1316],{},"ChipAgents ($21M, 2025 Series A, lead)",": Agentic AI for chip design and verification.",[67,1319,1320,1323],{},[32,1321,1322],{},"Kandou ($92.3M, 2020 Series C, lead)",": SerDes/IP and retimers for USB4 and server platforms.",[67,1325,1326,1329],{},[32,1327,1328],{},"Rigetti Computing ($79M, 2020 Series C, lead)",": Quantum hardware + cloud delivery.",[67,1331,1332,1335],{},[32,1333,1334],{},"Habana Labs ($75M Series B participant, 2018; acquired by Intel ~$2B, 2019)",": AI chips, co-invested with major corporate.",[28,1337,1338,1340],{},[32,1339,1221],{},": Deep European (Switzerland/Israel) networks enable Bessemer to source technically differentiated companies early. The pattern is consistent: deep tech + a credible path to large-scale deployment.",[58,1342],{},[353,1344,1346],{"id":1345},"softbank-vertical-integration-as-strategy","SoftBank: Vertical Integration as Strategy",[28,1348,1349,1350,1187],{},"SoftBank operates in a fundamentally different mode — not just investing in chip startups, but ",[32,1351,1352],{},"assembling an integrated AI computing segment through acquisitions",[28,1354,1190],{},[64,1356,1357,1363,1369],{},[67,1358,1359,1362],{},[32,1360,1361],{},"Fungible ($200M, 2019 Series C, SVF lead)",": DPU. Subsequently acquired by Microsoft.",[67,1364,1365,1368],{},[32,1366,1367],{},"Graphcore (M&A, 2024, amount undisclosed)",": UK AI chip company.",[67,1370,1371,1374],{},[32,1372,1373],{},"Ampere Computing ($6.5B acquisition announced, 2025)",": Arm-based cloud server CPU.",[28,1376,1377,1378,1381],{},"SoftBank has publicly organized Arm, Ampere, and Graphcore into a single ",[32,1379,1380],{},"AI Computing Segment",", articulating semiconductor as a core business pillar. The 2017 fund featured strategic LPs including PIF, Mubadala, Apple, Foxconn, Qualcomm, and Sharp — a structure that embeds semiconductor strategy at the fund capital level.",[28,1383,1384,1386],{},[32,1385,1221],{},": This is closer to strategic finance + operating company than traditional VC. The risk and time horizon are categorically different.",[58,1388],{},[20,1390,1392],{"id":1391},"investment-timeline","Investment Timeline",[91,1394,1395,1415],{},[94,1396,1397],{},[97,1398,1399,1401,1403,1406,1409,1412],{},[100,1400,950],{},[100,1402,247],{},[100,1404,1405],{},"Portfolio Company",[100,1407,1408],{},"Round",[100,1410,1411],{},"Amount ($M)",[100,1413,1414],{},"Tech Domain",[113,1416,1417,1433,1447,1461,1476,1490,1504,1518,1533,1549,1564,1579,1594,1609,1623,1639],{},[97,1418,1419,1421,1423,1425,1428,1430],{},[118,1420,569],{},[118,1422,316],{},[118,1424,574],{},[118,1426,1427],{},"Series B (participant)",[118,1429,580],{},[118,1431,1432],{},"AI accelerator",[97,1434,1435,1437,1439,1441,1443,1445],{},[118,1436,588],{},[118,1438,333],{},[118,1440,593],{},[118,1442,596],{},[118,1444,599],{},[118,1446,602],{},[97,1448,1449,1451,1453,1455,1457,1459],{},[118,1450,607],{},[118,1452,316],{},[118,1454,612],{},[118,1456,596],{},[118,1458,617],{},[118,1460,620],{},[97,1462,1463,1465,1467,1469,1471,1473],{},[118,1464,607],{},[118,1466,316],{},[118,1468,629],{},[118,1470,596],{},[118,1472,634],{},[118,1474,1475],{},"Quantum",[97,1477,1478,1480,1482,1484,1486,1488],{},[118,1479,642],{},[118,1481,266],{},[118,1483,647],{},[118,1485,650],{},[118,1487,653],{},[118,1489,656],{},[97,1491,1492,1494,1496,1498,1500,1502],{},[118,1493,642],{},[118,1495,266],{},[118,1497,665],{},[118,1499,650],{},[118,1501,670],{},[118,1503,673],{},[97,1505,1506,1508,1510,1512,1514,1516],{},[118,1507,120],{},[118,1509,266],{},[118,1511,682],{},[118,1513,685],{},[118,1515,688],{},[118,1517,1475],{},[97,1519,1520,1522,1524,1526,1529,1531],{},[118,1521,120],{},[118,1523,300],{},[118,1525,699],{},[118,1527,1528],{},"Cumulative",[118,1530,705],{},[118,1532,656],{},[97,1534,1535,1537,1539,1541,1543,1546],{},[118,1536,120],{},[118,1538,714],{},[118,1540,717],{},[118,1542,720],{},[118,1544,1545],{},"Undisclosed",[118,1547,1548],{},"AI chip",[97,1550,1551,1553,1555,1557,1559,1561],{},[118,1552,134],{},[118,1554,266],{},[118,1556,735],{},[118,1558,650],{},[118,1560,740],{},[118,1562,1563],{},"Design AI/EDA",[97,1565,1566,1568,1570,1572,1574,1576],{},[118,1567,134],{},[118,1569,283],{},[118,1571,752],{},[118,1573,650],{},[118,1575,757],{},[118,1577,1578],{},"Analog new compute",[97,1580,1581,1583,1585,1587,1589,1591],{},[118,1582,134],{},[118,1584,316],{},[118,1586,769],{},[118,1588,772],{},[118,1590,775],{},[118,1592,1593],{},"EDA automation",[97,1595,1596,1598,1600,1602,1604,1606],{},[118,1597,134],{},[118,1599,300],{},[118,1601,787],{},[118,1603,790],{},[118,1605,580],{},[118,1607,1608],{},"Coherent DSP",[97,1610,1611,1613,1615,1617,1619,1621],{},[118,1612,134],{},[118,1614,714],{},[118,1616,804],{},[118,1618,720],{},[118,1620,809],{},[118,1622,812],{},[97,1624,1625,1627,1630,1632,1634,1636],{},[118,1626,191],{},[118,1628,1629],{},"Sequoia (related)",[118,1631,822],{},[118,1633,825],{},[118,1635,828],{},[118,1637,1638],{},"CPO/Optical I/O",[97,1640,1641,1643,1645,1647,1649,1651],{},[118,1642,191],{},[118,1644,283],{},[118,1646,840],{},[118,1648,685],{},[118,1650,828],{},[118,1652,1653],{},"AI switch",[58,1655],{},[20,1657,1659],{"id":1658},"zyl0s-perspective","ZYL0's Perspective",[855,1661,1662,1667,1675,1678],{},[28,1663,1664],{},[32,1665,1666],{},"Semiconductor investing means buying AI's constraints, not its compute",[28,1668,1669,1670,1674],{},"The most striking pattern from this research: top-tier VCs are almost uniformly targeting the ",[1671,1672,1673],"em",{},"constraints around"," compute rather than compute itself. If GPU performance keeps scaling exponentially, design, interconnect, packaging, and supply chain each become the next limiting factor in turn.",[28,1676,1677],{},"From my perspective as an investor, the most actionable angles for Japan-based exposure are: (1) manufacturing equipment and materials, where Japan has globally competitive incumbents, and (2) design IP and EDA automation, where the software-heavy nature lowers entry barriers relative to hardware.",[28,1679,1680],{},"Part 2 of this series goes deeper on each technology domain — the investment logic, the specific case studies, and what separates winners from losers in each vertical.",{"title":229,"searchDepth":1682,"depth":1682,"links":1683},2,[1684,1685,1686,1687,1688,1689,1690,1698,1699,1700,1701,1702,1703,1704,1705,1712,1713],{"id":22,"depth":1682,"text":23},{"id":26,"depth":1682,"text":26},{"id":62,"depth":1682,"text":62},{"id":89,"depth":1682,"text":89},{"id":214,"depth":1682,"text":215},{"id":237,"depth":1682,"text":238},{"id":350,"depth":1682,"text":351,"children":1691},[1692,1694,1695,1696,1697],{"id":355,"depth":1693,"text":356},3,{"id":399,"depth":1693,"text":400},{"id":429,"depth":1693,"text":430},{"id":459,"depth":1693,"text":460},{"id":501,"depth":1693,"text":502},{"id":540,"depth":1682,"text":540},{"id":852,"depth":1682,"text":853},{"id":880,"depth":1682,"text":881},{"id":914,"depth":1682,"text":915},{"id":940,"depth":1682,"text":941},{"id":1050,"depth":1682,"text":1051},{"id":1068,"depth":1682,"text":1069},{"id":1175,"depth":1682,"text":1176,"children":1706},[1707,1708,1709,1710,1711],{"id":1179,"depth":1693,"text":1180},{"id":1227,"depth":1693,"text":1228},{"id":1261,"depth":1693,"text":1262},{"id":1299,"depth":1693,"text":1300},{"id":1345,"depth":1693,"text":1346},{"id":1391,"depth":1682,"text":1392},{"id":1658,"depth":1682,"text":1659},"市場規模$975Bへ拡大する半導体業界でTier1 VCはどこに賭けているのか。Sequoia、a16z、Kleiner Perkins、Bessemer、SoftBankの戦略を比較分析する。連載全3回の初回。","md",{"date":1717,"image":1718,"alt":1719,"tags":1720,"tagsEn":1726,"published":1732},"11th Apr 2026","/blogs-img/blog-semiconductor-vc-1.png","Tier1 VCと半導体投資の全体像",[1721,1722,1723,1724,1725],"半導体","VC投資","AI投資","スタートアップ","投資",[1727,1728,1729,1730,1731],"Semiconductor","VC Investment","AI Investment","Startup","Investment",true,"/blogs/6-semiconductor-vc-series-1",{"title":6,"description":1714},"blogs/6-semiconductor-vc-series-1","miuoEmt5k8rSY0Bw17smZSjj0jduhDrxgBpxM0yR8Zc",{"id":1738,"title":1739,"body":1740,"description":3838,"extension":1715,"meta":3839,"navigation":1732,"ogImage":3840,"path":3846,"seo":3847,"stem":3848,"__hash__":3849},"content/blogs/7-semiconductor-vc-series-2.md","「設計」「データ移動」「パッケージング」に資金が集まる理由｜連載②技術別深掘り | Why Design, Data Movement & Packaging Attract VC Capital — Series②",{"type":8,"value":1741,"toc":3780},[1742,2783],[11,1743,1744,1748,1752,1754,1757,1760,1763,1765,1767,1787,1789,1793,1995,1997,2001,2005,2008,2012,2015,2022,2026,2029,2032,2037,2091,2093,2097,2100,2103,2109,2113,2116,2119,2125,2129,2132,2135,2139,2142,2145,2149,2152,2155,2160,2211,2213,2217,2221,2224,2239,2246,2250,2253,2256,2276,2279,2285,2289,2292,2340,2343,2345,2349,2353,2356,2363,2367,2370,2373,2377,2380,2383,2388,2432,2434,2438,2763,2765,2767],{"lang":13},[15,1745,1747],{"id":1746},"設計データ移動パッケージングに資金が集まる理由","「設計」「データ移動」「パッケージング」に資金が集まる理由",[20,1749,1751],{"id":1750},"連載技術別深掘りとケーススタディ","——連載②：技術別深掘りとケーススタディ",[20,1753,26],{"id":26},[28,1755,1756],{},"連載第2回は技術領域の深掘りだ。前回（連載①）で確認した通り、Tier1 VCが半導体に賭けるのは「計算そのもの」ではなく「AIの制約領域」だ。",[28,1758,1759],{},"なぜか。GPUの演算能力が指数関数的に伸びれば、必ずその「周辺」がボトルネックに移る。1万枚のGPUを繋ぐにはスイッチとインターコネクトが要る。チップ単体の性能限界を超えるにはパッケージング（CPO/Chiplet）が要る。設計の複雑性が増せばEDAと検証の自動化が要る。",[28,1761,1762],{},"それぞれの領域に、なぜ今、大型の資金が流れているのか。代表案件をケーススタディとして読み解く。",[58,1764],{},[20,1766,62],{"id":62},[64,1768,1769,1775,1781],{},[67,1770,1771,1774],{},[32,1772,1773],{},"投資のROI設計は「単一チップ勝負」から「スタック勝負（設計→実装→運用）」へ。"," CPOやコヒレントDSP、スイッチングなど、運用上の制約を解く領域で$500M超の大型ラウンドが立ち上がっている。",[67,1776,1777,1780],{},[32,1778,1779],{},"生成AIの普及で設計の複雑性が増し、EDA自動化（Agentic AI）が新たな投資テーマに。"," 設計・検証のリードタイム短縮は、製造リスクが相対的に低いソフト寄りの案件として投資回転が早い。",[67,1782,1783,1786],{},[32,1784,1785],{},"「新計算（アナログ/量子）」は長期テーマとして存続するが、資金は「研究→プロダクト→市場」への接続ができる会社に集まる。"," 技術の実現可能性だけでなく、顧客・クラウド・政府との連携が評価基準になっている。",[58,1788],{},[20,1790,1792],{"id":1791},"技術スタック-投資マッピング","技術スタック × 投資マッピング",[91,1794,1795,1813],{},[94,1796,1797],{},[97,1798,1799,1801,1804,1807,1810],{},[100,1800,562],{},[100,1802,1803],{},"代表企業（VC）",[100,1805,1806],{},"主要ラウンド",[100,1808,1809],{},"投資額",[100,1811,1812],{},"技術の役割",[113,1814,1815,1830,1845,1860,1875,1890,1905,1919,1934,1949,1965,1980],{},[97,1816,1817,1819,1822,1824,1827],{},[118,1818,743],{},[118,1820,1821],{},"Ricursive（Sequoia）",[118,1823,650],{},[118,1825,1826],{},"$35M",[118,1828,1829],{},"チップ設計サイクルの短縮",[97,1831,1832,1834,1837,1839,1842],{},[118,1833,743],{},[118,1835,1836],{},"ChipAgents（Bessemer）",[118,1838,772],{},[118,1840,1841],{},"$21M",[118,1843,1844],{},"設計・検証のエージェント化",[97,1846,1847,1849,1852,1854,1857],{},[118,1848,656],{},[118,1850,1851],{},"Akeana（Kleiner Perkins）",[118,1853,702],{},[118,1855,1856],{},">$100M",[118,1858,1859],{},"自由に使えるCPU命令セット",[97,1861,1862,1864,1867,1869,1872],{},[118,1863,620],{},[118,1865,1866],{},"Kandou（Bessemer）",[118,1868,596],{},[118,1870,1871],{},"$92.3M",[118,1873,1874],{},"高速チップ間通信の物理層",[97,1876,1877,1879,1882,1884,1887],{},[118,1878,795],{},[118,1880,1881],{},"Retym（Kleiner Perkins）",[118,1883,790],{},[118,1885,1886],{},"$75M",[118,1888,1889],{},"長距離光通信の信号処理",[97,1891,1892,1894,1897,1899,1902],{},[118,1893,831],{},[118,1895,1896],{},"Ayar Labs（Sequoia系）",[118,1898,825],{},[118,1900,1901],{},"$500M",[118,1903,1904],{},"銅配線の限界をシリコンフォトニクスで超える",[97,1906,1907,1909,1912,1914,1916],{},[118,1908,847],{},[118,1910,1911],{},"Nexthop AI（a16z）",[118,1913,685],{},[118,1915,1901],{},[118,1917,1918],{},"AI大規模クラスタのネットワーク",[97,1920,1921,1923,1926,1928,1931],{},[118,1922,760],{},[118,1924,1925],{},"Unconventional AI（a16z）",[118,1927,650],{},[118,1929,1930],{},"$475M",[118,1932,1933],{},"電力効率の桁違い改善を狙う確率的計算",[97,1935,1936,1939,1942,1944,1946],{},[118,1937,1938],{},"AIチップ（買収）",[118,1940,1941],{},"Graphcore（SoftBank）",[118,1943,720],{},[118,1945,723],{},[118,1947,1948],{},"英国発グラフプロセッサ",[97,1950,1951,1954,1957,1959,1962],{},[118,1952,1953],{},"CPU/Arm（買収）",[118,1955,1956],{},"Ampere（SoftBank）",[118,1958,720],{},[118,1960,1961],{},"$6.5B",[118,1963,1964],{},"クラウドサーバー向けArmベースCPU",[97,1966,1967,1969,1972,1974,1977],{},[118,1968,637],{},[118,1970,1971],{},"Rigetti（Bessemer）",[118,1973,596],{},[118,1975,1976],{},"$79M",[118,1978,1979],{},"量子クラウド提供+ハードウェア",[97,1981,1982,1984,1987,1989,1992],{},[118,1983,637],{},[118,1985,1986],{},"Quantum Circuits（Sequoia）",[118,1988,685],{},[118,1990,1991],{},">$60M",[118,1993,1994],{},"フォールトトレラント量子コンピュータ",[58,1996],{},[20,1998,2000],{"id":1999},"領域設計ipeda-開発リードタイム短縮への賭け","領域①：設計・IP/EDA — 「開発リードタイム短縮」への賭け",[353,2002,2004],{"id":2003},"なぜ今設計自動化にvcが入るのか","なぜ今、設計自動化にVCが入るのか",[28,2006,2007],{},"先端チップの設計工数は年々増大している。7nm→3nm→2nmと微細化するほど、設計規則の複雑性は指数関数的に増え、テープアウト1回のコストも億円単位になる。検証・シミュレーション・デバッグが開発期間の大半を占める現状に対し、AIエージェントを当てる仮説が立ち上がっている。",[353,2009,2011],{"id":2010},"ケーススタディricursive-intelligence35m2025年seed","ケーススタディ：Ricursive Intelligence（$35M、2025年Seed）",[28,2013,2014],{},"Sequoiaがリードした$35Mシードは、「AIがチップ設計そのものを加速する」テーマをTier1が最上流から取りにいった案件だ。",[28,2016,2017,2018,2021],{},"注目点は",[32,2019,2020],{},"ソフトウェア比率が高い","ことだ。製造工程（ファブ）を持たず、設計フローにAIを組み込むアプローチは、相対的に製造リスクが低い。一方で、Synopsys・Cadenceというグローバルなツールチェーン巨人のプロダクト戦略・M&A戦略に強く影響される。勝ち筋は「差別化されたワークフロー統合」か「特定の設計フロー（例：検証/STA）での圧倒的優位」になりやすい。",[353,2023,2025],{"id":2024},"ケーススタディchipagents21m2025年series-a","ケーススタディ：ChipAgents（$21M、2025年Series A）",[28,2027,2028],{},"Bessemerがリードしたこの案件は、既存EDAチェーンの「検証/デバッグ」をAIエージェントで置き換える仮説だ。",[28,2030,2031],{},"「設計工数の相当部分は検証とバグ修正に費やされる」という現場の現実を起点に、Agentic AIのループ（タスク定義→実行→検証→修正）を設計フローに組み込む。ChipAgentsが述べる通り、目標は「エンジニアが書く必要があるコードを減らす」ことではなく「検証→修正のサイクルを自動化する」ことだ。",[28,2033,2034],{},[32,2035,2036],{},"EDA自動化の投資ロジックまとめ：",[91,2038,2039,2049],{},[94,2040,2041],{},[97,2042,2043,2046],{},[100,2044,2045],{},"観点",[100,2047,2048],{},"内容",[113,2050,2051,2059,2067,2075,2083],{},[97,2052,2053,2056],{},[118,2054,2055],{},"市場の大きさ",[118,2057,2058],{},"全世界のチップ設計者数×設計サイクル短縮価値",[97,2060,2061,2064],{},[118,2062,2063],{},"製造リスク",[118,2065,2066],{},"低（ソフトウェア主体）",[97,2068,2069,2072],{},[118,2070,2071],{},"競合リスク",[118,2073,2074],{},"EDA大手（Synopsys/Cadence）のM&A対象になりうる",[97,2076,2077,2080],{},[118,2078,2079],{},"出口",[118,2081,2082],{},"M&A（EDA大手/チップ設計会社）またはIPO",[97,2084,2085,2088],{},[118,2086,2087],{},"タイムライン",[118,2089,2090],{},"比較的短い（2〜5年）",[58,2092],{},[20,2094,2096],{"id":2095},"領域インターコネクトデータ移動-gpuを繋ぐが主戦場","領域②：インターコネクト/データ移動 — 「GPUを繋ぐ」が主戦場",[353,2098,2099],{"id":2099},"銅配線の限界とシリコンフォトニクスの台頭",[28,2101,2102],{},"AI大規模クラスタ（数千〜数万GPU）では、チップ間・ラック間・データセンター間のデータ移動が帯域・遅延・電力の全てでボトルネックになっている。",[220,2104,2107],{"className":2105,"code":2106,"language":225},[223],"従来の銅配線：帯域 ↑ → 電力消費 ↑↑↑、距離 → 減衰\nシリコンフォトニクス：光で通信 → 高帯域・低電力・長距離\nCPO（Co-Packaged Optics）：光I/OをチップパッケージにCo-Packageng\n        → 銅配線の物理限界を迂回する\n",[227,2108,2106],{"__ignoreMap":229},[353,2110,2112],{"id":2111},"ケーススタディayar-labs500m2026年series-e","ケーススタディ：Ayar Labs（$500M、2026年Series E）",[28,2114,2115],{},"評価額$3.75Bに達したこのラウンドは、CPO（Co-Packaged Optics）分野で最大規模の資金調達の一つだ。",[28,2117,2118],{},"戦略投資家にNVIDIA・AMDが名を連ねる構造は重要だ。彼らがAyar Labsに投資するのは「財務的リターン」だけでなく「次世代パッケージング技術を自分たちの製品ロードマップに組み込む」ためだ。Sequoia系ビークルが参加していることは、ピュアファイナンシャルのリターンも十分あるという判断を示す。",[28,2120,2121,2124],{},[32,2122,2123],{},"投資ロジック："," 銅配線の物理限界（帯域密度・電力）は設計上の解決が難しい。光I/OをCo-Packageするというアーキテクチャ転換が起きれば、その製造技術・IP・テスト能力を持つ会社の価値は急増する。",[353,2126,2128],{"id":2127},"ケーススタディretym75m2025年series-d","ケーススタディ：Retym（$75M、2025年Series D）",[28,2130,2131],{},"クラウド/AI向けコヒレントDSPを専門とするRetymのSeries Dには、Kleiner Perkinsが参加している。",[28,2133,2134],{},"特徴は「距離×帯域×消費電力」の現実制約に設計思想を合わせた点だ。データセンター内（数十〜数百m）、メトロ（数十km）、長距離（数百km〜）でそれぞれ最適なDSPアーキテクチャが異なる。汎用ではなく特定の「距離レンジ」に最適化したシリコンを作ることで、差別化を図る。",[353,2136,2138],{"id":2137},"ケーススタディkandou923m2020年series-c","ケーススタディ：Kandou（$92.3M、2020年Series C）",[28,2140,2141],{},"Bessemerがリードしたこの案件は、SerDes（シリアライザ/デシリアライザ）とリタイマーの商用化を前提にした投資だ。",[28,2143,2144],{},"KandouのMatterhorn IPはUSB4やThunderbolt実装で実績を持つ。サーバー・デバイスの世代交代（USB4 Gen3など）に合わせて実装が変わる局面を「実装の勝ち筋」として取りにいくスタイルだ。スタンダードの世代交代とともに需要が立ち上がる、周期性のあるビジネスモデルと言える。",[353,2146,2148],{"id":2147},"ケーススタディnexthop-ai500m2026年series-b","ケーススタディ：Nexthop AI（$500M、2026年Series B）",[28,2150,2151],{},"a16zが主要投資家として$500Mを投じたNexthop AIは、「AI時代のイーサネットスイッチ」を標榜する。",[28,2153,2154],{},"投資ロジックは明快だ。「AIクラスタが大規模化するほど、スイッチのボトルネックが顕在化する。既存のスイッチはAIの通信パターン（全-全接続的な集合通信）に最適化されていない」という認識から出発し、AI専用スイッチングのアーキテクチャを取りにいく。",[28,2156,2157],{},[32,2158,2159],{},"インターコネクト投資の共通ロジック：",[91,2161,2162,2170],{},[94,2163,2164],{},[97,2165,2166,2168],{},[100,2167,2045],{},[100,2169,2048],{},[113,2171,2172,2180,2188,2196,2204],{},[97,2173,2174,2177],{},[118,2175,2176],{},"技術ドライバー",[118,2178,2179],{},"GPU/アクセラレータの性能スケールがI/Oを際立たせる",[97,2181,2182,2185],{},[118,2183,2184],{},"顧客",[118,2186,2187],{},"ハイパースケーラー（Google/AWS/Microsoft/Meta等）",[97,2189,2190,2193],{},[118,2191,2192],{},"資本集約度",[118,2194,2195],{},"中〜高（量産・認定が必要）",[97,2197,2198,2201],{},[118,2199,2200],{},"リスク",[118,2202,2203],{},"量産歩留まり、サプライヤー認定、地政学",[97,2205,2206,2208],{},[118,2207,2079],{},[118,2209,2210],{},"M&A（チップ大手）またはIPO",[58,2212],{},[20,2214,2216],{"id":2215},"領域aiアクセラレータ新計算-計算原理への賭け","領域③：AIアクセラレータ/新計算 — 計算原理への賭け",[353,2218,2220],{"id":2219},"パラダイム賭け-vs-垂直統合","パラダイム賭け vs. 垂直統合",[28,2222,2223],{},"「計算原理を変える」というテーマには2つのアプローチが存在する。",[2225,2226,2227,2233],"ol",{},[67,2228,2229,2232],{},[32,2230,2231],{},"スタートアップへの巨大シード","（a16z × Unconventional AI）：リスクが高いが、実現すれば市場を再定義できる。",[67,2234,2235,2238],{},[32,2236,2237],{},"買収による垂直統合","（SoftBank × Graphcore/Ampere）：技術の実現可能性リスクは低いが、統合コストと市場タイミングリスクがある。",[28,2240,2241,2242,2245],{},"どちらも「単なるチップ投資」ではなく、",[32,2243,2244],{},"エコシステムとしてのポジション取り","だという点で共通している。",[353,2247,2249],{"id":2248},"ケーススタディunconventional-ai475m2025年seed","ケーススタディ：Unconventional AI（$475M、2025年Seed）",[28,2251,2252],{},"a16zが共同リードした$475Mシードは、半導体投資史上最大規模のシードラウンドの一つだ。",[28,2254,2255],{},"アナログ/ミックスドシグナルで確率的計算を扱うというアプローチは、デジタルCMOSによる現行アーキテクチャとは根本的に異なる。主な課題は：",[64,2257,2258,2264,2270],{},[67,2259,2260,2263],{},[32,2261,2262],{},"量産・再現性","：アナログ回路は製造ばらつきの影響を受けやすい",[67,2265,2266,2269],{},[32,2267,2268],{},"ツールチェーン","：設計・検証ツールが整っていない",[67,2271,2272,2275],{},[32,2273,2274],{},"エコシステム","：ソフトウェア/フレームワーク側の対応が必要",[28,2277,2278],{},"一方、実現すれば「電力効率の桁違い改善」というインパクトは現行GPUに対して圧倒的な競争優位になりうる。",[28,2280,2281,2284],{},[32,2282,2283],{},"a16zがこの賭けをする理由","：Strategic Partnershipsで企業・政府導入の経路を制度化しているa16zにとって、「カテゴリを再定義するアーキテクチャ」を囲い込む初期コストは、後で払う「選択肢なしコスト」より安い。",[353,2286,2288],{"id":2287},"softbankの垂直統合armgraphcoreampere","SoftBankの垂直統合：Arm×Graphcore×Ampere",[28,2290,2291],{},"SoftBankのアプローチは、個別案件の勝負ではなくセグメント全体の設計だ。",[91,2293,2294,2307],{},[94,2295,2296],{},[97,2297,2298,2301,2304],{},[100,2299,2300],{},"会社",[100,2302,2303],{},"役割",[100,2305,2306],{},"SoftBankの取得形態",[113,2308,2309,2320,2330],{},[97,2310,2311,2314,2317],{},[118,2312,2313],{},"Arm",[118,2315,2316],{},"CPU IP、命令セット標準、エコシステムの核",[118,2318,2319],{},"2016年買収（$32B）",[97,2321,2322,2324,2327],{},[118,2323,717],{},[118,2325,2326],{},"グラフ最適化AI処理（GC200/BOW等）",[118,2328,2329],{},"2024年買収",[97,2331,2332,2334,2337],{},[118,2333,804],{},[118,2335,2336],{},"Armベースクラウドサーバーへの最適化",[118,2338,2339],{},"2025年$6.5Bで買収発表",[28,2341,2342],{},"このセグメントは「Armのエコシステムを基盤に、AIワークロードに特化したシリコンを垂直統合する」という事業戦略だ。VCリターンではなく事業IRR（グループシナジー含む）で評価される性質のものだ。",[58,2344],{},[20,2346,2348],{"id":2347},"領域量子-長期だが資本が戻り始めている","領域④：量子 — 長期だが資本が戻り始めている",[353,2350,2352],{"id":2351},"量子投資の商用化フェーズへの移行","量子投資の「商用化フェーズ」への移行",[28,2354,2355],{},"2020年前後の量子ブームは、多くのハードウェアスタートアップが誕生し、その後の市場との乖離で一部が撤退・合併した。しかし2024〜2025年には「プラットフォーム競争」「クラウド提供」を前提にした大型ラウンドが戻り始めている。",[28,2357,2358,2359,2362],{},"変化の核心は：「量子コンピュータが量子優位を達成できるか」ではなく、",[32,2360,2361],{},"「エラー訂正・クラウド提供・ソフトウェアスタックをセットで商用化できるか」"," という評価軸への移行だ。",[353,2364,2366],{"id":2365},"ケーススタディquantum-circuitsseries-b-60m2024年","ケーススタディ：Quantum Circuits（Series B > $60M、2024年）",[28,2368,2369],{},"Sequoiaがリード投資家として参加したQuantum Circuitsのシリーズ B は、フォールトトレラント量子コンピュータの商用化を狙う。",[28,2371,2372],{},"同社の特徴は超伝導量子回路の設計とエラー訂正への集中だ。「ノイズのある中規模量子（NISQ）」ではなく、長期的にスケールするフォールトトレラントアーキテクチャへの賭けとして位置づけられる。",[353,2374,2376],{"id":2375},"ケーススタディrigetti-computing79m2020年series-c","ケーススタディ：Rigetti Computing（$79M、2020年Series C）",[28,2378,2379],{},"Bessemerがリードしたこの案件は、量子ハードウェア+クラウド提供（Quantum Cloud Services）という垂直統合モデルだ。",[28,2381,2382],{},"後にSPACでNASDAQ上場（2022年）し、量子スタートアップとして初期の公開市場実績を残した。クラウド提供という「ソフト的な入り口」を持つことで、企業顧客へのアクセスを維持する戦略は、量子の商用化において重要な設計判断だった。",[28,2384,2385],{},[32,2386,2387],{},"量子投資の評価基準（現在）：",[91,2389,2390,2400],{},[94,2391,2392],{},[97,2393,2394,2397],{},[100,2395,2396],{},"評価軸",[100,2398,2399],{},"詳細",[113,2401,2402,2410,2418,2425],{},[97,2403,2404,2407],{},[118,2405,2406],{},"技術",[118,2408,2409],{},"エラー率・量子ビット数・コヒーレンス時間",[97,2411,2412,2415],{},[118,2413,2414],{},"商用化",[118,2416,2417],{},"クラウドAPI・SDK・顧客導入実績",[97,2419,2420,2422],{},[118,2421,2274],{},[118,2423,2424],{},"アルゴリズム・ソフトウェアパートナー",[97,2426,2427,2429],{},[118,2428,2079],{},[118,2430,2431],{},"大手クラウド（AWS/Azure/GCP）への買収またはIPO",[58,2433],{},[20,2435,2437],{"id":2436},"代表投資案件-完全版テーブル","代表投資案件 完全版テーブル",[91,2439,2440,2462],{},[94,2441,2442],{},[97,2443,2444,2446,2448,2451,2453,2455,2457,2460],{},[100,2445,247],{},[100,2447,553],{},[100,2449,2450],{},"投資年",[100,2452,556],{},[100,2454,559],{},[100,2456,562],{},[100,2458,2459],{},"ステージ",[100,2461,256],{},[113,2463,2464,2483,2502,2520,2538,2557,2575,2593,2612,2631,2650,2670,2689,2708,2726,2745],{},[97,2465,2466,2468,2470,2472,2474,2476,2478,2481],{},[118,2467,266],{},[118,2469,682],{},[118,2471,120],{},[118,2473,685],{},[118,2475,688],{},[118,2477,637],{},[118,2479,2480],{},"B",[118,2482,292],{},[97,2484,2485,2487,2489,2491,2493,2495,2497,2499],{},[118,2486,266],{},[118,2488,647],{},[118,2490,642],{},[118,2492,650],{},[118,2494,653],{},[118,2496,656],{},[118,2498,650],{},[118,2500,2501],{},"インド",[97,2503,2504,2506,2508,2510,2512,2514,2516,2518],{},[118,2505,266],{},[118,2507,665],{},[118,2509,642],{},[118,2511,650],{},[118,2513,670],{},[118,2515,673],{},[118,2517,650],{},[118,2519,2501],{},[97,2521,2522,2524,2526,2528,2530,2532,2534,2536],{},[118,2523,266],{},[118,2525,735],{},[118,2527,134],{},[118,2529,650],{},[118,2531,740],{},[118,2533,743],{},[118,2535,650],{},[118,2537,292],{},[97,2539,2540,2542,2544,2546,2548,2550,2552,2555],{},[118,2541,819],{},[118,2543,822],{},[118,2545,191],{},[118,2547,825],{},[118,2549,828],{},[118,2551,831],{},[118,2553,2554],{},"E",[118,2556,292],{},[97,2558,2559,2561,2563,2565,2567,2569,2571,2573],{},[118,2560,283],{},[118,2562,752],{},[118,2564,134],{},[118,2566,650],{},[118,2568,757],{},[118,2570,760],{},[118,2572,650],{},[118,2574,292],{},[97,2576,2577,2579,2581,2583,2585,2587,2589,2591],{},[118,2578,283],{},[118,2580,840],{},[118,2582,191],{},[118,2584,685],{},[118,2586,828],{},[118,2588,847],{},[118,2590,2480],{},[118,2592,292],{},[97,2594,2595,2597,2599,2601,2603,2605,2607,2610],{},[118,2596,300],{},[118,2598,699],{},[118,2600,120],{},[118,2602,702],{},[118,2604,705],{},[118,2606,656],{},[118,2608,2609],{},"Early",[118,2611,292],{},[97,2613,2614,2616,2618,2620,2622,2624,2626,2629],{},[118,2615,300],{},[118,2617,787],{},[118,2619,134],{},[118,2621,790],{},[118,2623,580],{},[118,2625,795],{},[118,2627,2628],{},"D",[118,2630,292],{},[97,2632,2633,2635,2637,2639,2641,2643,2645,2648],{},[118,2634,316],{},[118,2636,769],{},[118,2638,134],{},[118,2640,772],{},[118,2642,775],{},[118,2644,778],{},[118,2646,2647],{},"A",[118,2649,292],{},[97,2651,2652,2654,2656,2658,2660,2662,2664,2667],{},[118,2653,316],{},[118,2655,612],{},[118,2657,607],{},[118,2659,596],{},[118,2661,617],{},[118,2663,620],{},[118,2665,2666],{},"C",[118,2668,2669],{},"スイス",[97,2671,2672,2674,2677,2679,2681,2683,2685,2687],{},[118,2673,316],{},[118,2675,2676],{},"Rigetti Computing",[118,2678,607],{},[118,2680,596],{},[118,2682,634],{},[118,2684,637],{},[118,2686,2666],{},[118,2688,292],{},[97,2690,2691,2693,2695,2697,2699,2701,2703,2705],{},[118,2692,316],{},[118,2694,574],{},[118,2696,569],{},[118,2698,577],{},[118,2700,580],{},[118,2702,583],{},[118,2704,2480],{},[118,2706,2707],{},"イスラエル",[97,2709,2710,2712,2714,2716,2718,2720,2722,2724],{},[118,2711,333],{},[118,2713,593],{},[118,2715,588],{},[118,2717,596],{},[118,2719,599],{},[118,2721,602],{},[118,2723,2666],{},[118,2725,292],{},[97,2727,2728,2730,2732,2734,2736,2738,2740,2742],{},[118,2729,714],{},[118,2731,717],{},[118,2733,120],{},[118,2735,720],{},[118,2737,723],{},[118,2739,726],{},[118,2741,720],{},[118,2743,2744],{},"英国",[97,2746,2747,2749,2751,2753,2755,2757,2759,2761],{},[118,2748,714],{},[118,2750,804],{},[118,2752,134],{},[118,2754,720],{},[118,2756,809],{},[118,2758,812],{},[118,2760,720],{},[118,2762,292],{},[58,2764],{},[20,2766,853],{"id":852},[855,2768,2769,2774,2777,2780],{},[28,2770,2771],{},[32,2772,2773],{},"「大型ラウンド＝確実性が高い」は誤解だ",[28,2775,2776],{},"Unconventional AIの$475Mシードも、Nexthop/Ayar Labsの$500Mも、「確度が高いから大金を出している」わけではない。「外れたとき以上に、当たったときのカテゴリ創造価値が大きいから大金を出している」のだ。",[28,2778,2779],{},"技術観点で個人的に最も興味深いのはCPOだ。光I/OとチップをCo-Packageするという思想は、半導体・光学・パッケージングという全く異なる産業の交差点にある。日本には半導体装置・光部品・精密実装という3つの強みがあり、このCPO領域は日本企業にとっても参入余地のある数少ない「最先端の交差点」だと思う。",[28,2781,2782],{},"次回（連載③）は戦略的示唆と投資提案——どうポートフォリオを組むか、短中長期でどこに賭けるかを整理する。",[11,2784,2785,2789,2793,2795,2798,2801,2808,2810,2812,2832,2834,2838,3031,3033,3037,3041,3044,3048,3051,3058,3062,3065,3068,3073,3127,3129,3133,3137,3140,3146,3150,3153,3156,3162,3166,3169,3172,3176,3179,3186,3190,3193,3196,3201,3252,3254,3258,3262,3265,3279,3286,3290,3293,3296,3316,3323,3329,3333,3379,3386,3388,3392,3396,3402,3408,3412,3415,3419,3422,3427,3471,3473,3477,3760,3762,3764],{"lang":873},[15,2786,2788],{"id":2787},"why-design-data-movement-packaging-attract-vc-capital","Why Design, Data Movement & Packaging Attract VC Capital",[20,2790,2792],{"id":2791},"series-technical-deep-dive-and-case-studies","— Series②: Technical Deep Dive and Case Studies",[20,2794,881],{"id":880},[28,2796,2797],{},"Part 2 goes deeper. As established in Part 1, top-tier VCs are not betting on compute itself — they're betting on the constraints that emerge around compute.",[28,2799,2800],{},"The logic is straightforward: as GPU performance scales exponentially, the surrounding layers become the limiting factor. Connecting 10,000 GPUs requires switches and interconnects. Exceeding the limits of individual chips requires packaging (CPO/Chiplet). Growing design complexity requires EDA and verification automation.",[28,2802,2803,2804,2807],{},"Why is large capital flowing into each of these areas ",[1671,2805,2806],{},"right now","? We read the representative deals as case studies.",[58,2809],{},[20,2811,915],{"id":914},[64,2813,2814,2820,2826],{},[67,2815,2816,2819],{},[32,2817,2818],{},"Investment ROI is shifting from single-chip to full-stack (design→packaging→operations)."," CPO, coherent DSP, and AI switching — areas that solve operational constraints — are now commanding $500M+ rounds.",[67,2821,2822,2825],{},[32,2823,2824],{},"Generative AI has increased design complexity, making EDA automation (Agentic AI) a new investment theme."," With lower manufacturing risk and faster iteration, design automation is an attractive early-stage semiconductor play.",[67,2827,2828,2831],{},[32,2829,2830],{},"New compute (analog/quantum) remains a long-term theme, but capital flows to companies that can connect research to product to market."," Technical feasibility alone is insufficient — customer, cloud, and government partnerships are now evaluation criteria.",[58,2833],{},[20,2835,2837],{"id":2836},"technology-stack-investment-mapping","Technology Stack × Investment Mapping",[91,2839,2840,2857],{},[94,2841,2842],{},[97,2843,2844,2846,2849,2851,2854],{},[100,2845,1414],{},[100,2847,2848],{},"Representative Co. (VC)",[100,2850,1408],{},[100,2852,2853],{},"Amount",[100,2855,2856],{},"Role",[113,2858,2859,2873,2887,2901,2915,2929,2943,2958,2973,2988,3003,3017],{},[97,2860,2861,2863,2866,2868,2870],{},[118,2862,1563],{},[118,2864,2865],{},"Ricursive (Sequoia)",[118,2867,650],{},[118,2869,1826],{},[118,2871,2872],{},"Accelerate chip design cycles",[97,2874,2875,2877,2880,2882,2884],{},[118,2876,1563],{},[118,2878,2879],{},"ChipAgents (Bessemer)",[118,2881,772],{},[118,2883,1841],{},[118,2885,2886],{},"Agentic AI for design/verification",[97,2888,2889,2891,2894,2896,2898],{},[118,2890,656],{},[118,2892,2893],{},"Akeana (Kleiner Perkins)",[118,2895,1528],{},[118,2897,1856],{},[118,2899,2900],{},"Open CPU instruction set",[97,2902,2903,2905,2908,2910,2912],{},[118,2904,620],{},[118,2906,2907],{},"Kandou (Bessemer)",[118,2909,596],{},[118,2911,1871],{},[118,2913,2914],{},"High-speed chip-to-chip PHY",[97,2916,2917,2919,2922,2924,2926],{},[118,2918,1608],{},[118,2920,2921],{},"Retym (Kleiner Perkins)",[118,2923,790],{},[118,2925,1886],{},[118,2927,2928],{},"Signal processing for optical links",[97,2930,2931,2933,2936,2938,2940],{},[118,2932,1638],{},[118,2934,2935],{},"Ayar Labs (Sequoia related)",[118,2937,825],{},[118,2939,1901],{},[118,2941,2942],{},"Silicon photonics to bypass copper limits",[97,2944,2945,2948,2951,2953,2955],{},[118,2946,2947],{},"AI Switch",[118,2949,2950],{},"Nexthop AI (a16z)",[118,2952,685],{},[118,2954,1901],{},[118,2956,2957],{},"Networking for large-scale AI clusters",[97,2959,2960,2963,2966,2968,2970],{},[118,2961,2962],{},"Analog compute",[118,2964,2965],{},"Unconventional AI (a16z)",[118,2967,650],{},[118,2969,1930],{},[118,2971,2972],{},"Probabilistic computing for power efficiency",[97,2974,2975,2978,2981,2983,2985],{},[118,2976,2977],{},"AI chip (M&A)",[118,2979,2980],{},"Graphcore (SoftBank)",[118,2982,720],{},[118,2984,1545],{},[118,2986,2987],{},"UK graph processor",[97,2989,2990,2993,2996,2998,3000],{},[118,2991,2992],{},"CPU/Arm (M&A)",[118,2994,2995],{},"Ampere (SoftBank)",[118,2997,720],{},[118,2999,1961],{},[118,3001,3002],{},"Arm-based cloud server CPU",[97,3004,3005,3007,3010,3012,3014],{},[118,3006,1475],{},[118,3008,3009],{},"Rigetti (Bessemer)",[118,3011,596],{},[118,3013,1976],{},[118,3015,3016],{},"Quantum cloud + hardware",[97,3018,3019,3021,3024,3026,3028],{},[118,3020,1475],{},[118,3022,3023],{},"Quantum Circuits (Sequoia)",[118,3025,685],{},[118,3027,1991],{},[118,3029,3030],{},"Fault-tolerant quantum",[58,3032],{},[20,3034,3036],{"id":3035},"domain-designipeda-betting-on-development-lead-time-reduction","Domain①: Design/IP/EDA — Betting on Development Lead Time Reduction",[353,3038,3040],{"id":3039},"why-vcs-are-entering-design-automation-now","Why VCs Are Entering Design Automation Now",[28,3042,3043],{},"Advanced chip design complexity grows with each process node. At 7nm→3nm→2nm, design rule complexity increases exponentially, and a single tapeout can cost tens of millions of dollars. Verification, simulation, and debugging now consume the majority of the design cycle. The thesis: apply AI agents to automate the most time-intensive parts of this workflow.",[353,3045,3047],{"id":3046},"case-study-ricursive-intelligence-35m-2025-seed","Case Study: Ricursive Intelligence ($35M, 2025 Seed)",[28,3049,3050],{},"Sequoia's lead on this $35M seed is the clearest signal that a top-tier VC is entering chip design automation at the earliest stage.",[28,3052,3053,3054,3057],{},"The key advantage: ",[32,3055,3056],{},"high software content",". Without owning fabrication, the approach of embedding AI into design flows carries relatively lower manufacturing risk. The main competitive risk is from Synopsys and Cadence — EDA giants with deep customer relationships and M&A firepower. The winning position is likely \"differentiated workflow integration\" or \"dominant capability in a specific design step (e.g., verification, static timing analysis).\"",[353,3059,3061],{"id":3060},"case-study-chipagents-21m-2025-series-a","Case Study: ChipAgents ($21M, 2025 Series A)",[28,3063,3064],{},"Bessemer-led, ChipAgents automates the verification and debug loop using agentic AI.",[28,3066,3067],{},"The framing: \"a significant fraction of design engineering time is spent in verification and bug-fixing cycles.\" An agentic AI system that can define tasks, execute, verify, and iterate autonomously in the EDA toolchain reduces that cycle. The goal is not reducing code written, but automating the verification-to-fix loop itself.",[28,3069,3070],{},[32,3071,3072],{},"EDA Automation Investment Logic:",[91,3074,3075,3085],{},[94,3076,3077],{},[97,3078,3079,3082],{},[100,3080,3081],{},"Dimension",[100,3083,3084],{},"Detail",[113,3086,3087,3095,3103,3111,3119],{},[97,3088,3089,3092],{},[118,3090,3091],{},"Market size",[118,3093,3094],{},"Global chip designers × design cycle cost reduction",[97,3096,3097,3100],{},[118,3098,3099],{},"Manufacturing risk",[118,3101,3102],{},"Low (primarily software)",[97,3104,3105,3108],{},[118,3106,3107],{},"Competitive risk",[118,3109,3110],{},"M&A target for EDA incumbents (Synopsys/Cadence)",[97,3112,3113,3116],{},[118,3114,3115],{},"Exit paths",[118,3117,3118],{},"M&A (EDA giants or chip design companies) or IPO",[97,3120,3121,3124],{},[118,3122,3123],{},"Timeline",[118,3125,3126],{},"Relatively short (2–5 years)",[58,3128],{},[20,3130,3132],{"id":3131},"domain-interconnectdata-movement-connecting-gpus-is-the-main-arena","Domain②: Interconnect/Data Movement — \"Connecting GPUs\" is the Main Arena",[353,3134,3136],{"id":3135},"the-copper-bottleneck-and-the-rise-of-silicon-photonics","The Copper Bottleneck and the Rise of Silicon Photonics",[28,3138,3139],{},"In large-scale AI clusters (thousands to tens of thousands of GPUs), data movement between chips, racks, and data centers is the limiting factor across bandwidth, latency, and power simultaneously.",[220,3141,3144],{"className":3142,"code":3143,"language":225},[223],"Copper interconnect: bandwidth ↑ → power ↑↑↑, distance → signal loss\nSilicon photonics: optical transmission → high bandwidth, low power, long distance\nCo-Packaged Optics (CPO): integrate optical I/O into the chip package\n        → bypass the physical limits of copper\n",[227,3145,3143],{"__ignoreMap":229},[353,3147,3149],{"id":3148},"case-study-ayar-labs-500m-2026-series-e","Case Study: Ayar Labs ($500M, 2026 Series E)",[28,3151,3152],{},"This round, valuing Ayar Labs at $3.75B, is one of the largest fundraises in the CPO space.",[28,3154,3155],{},"The presence of NVIDIA and AMD as strategic investors is critical. Their participation is not purely financial — it signals that CPO technology will be integrated into next-generation product roadmaps. The participation of a Sequoia-related vehicle validates that pure financial returns are also available at this valuation.",[28,3157,3158,3161],{},[32,3159,3160],{},"Investment logic",": The physical limits of copper (bandwidth density, power) are fundamentally hard to engineer around. If CPO becomes the standard packaging architecture for AI compute, companies with the manufacturing technology, IP, and test capability will command significant value.",[353,3163,3165],{"id":3164},"case-study-retym-75m-2025-series-d","Case Study: Retym ($75M, 2025 Series D)",[28,3167,3168],{},"Kleiner Perkins-backed Retym specializes in coherent DSP optimized for cloud and AI workloads.",[28,3170,3171],{},"The differentiator: tailoring the DSP architecture to specific \"distance ranges\" — intra-datacenter (tens to hundreds of meters), metro (tens of km), and long-haul (hundreds of km+). Each range has different optimal DSP design. Rather than a universal chip, Retym targets specific operational regimes with purpose-built silicon.",[353,3173,3175],{"id":3174},"case-study-kandou-923m-2020-series-c","Case Study: Kandou ($92.3M, 2020 Series C)",[28,3177,3178],{},"Bessemer-led, Kandou's Matterhorn IP has demonstrated deployment in USB4 and Thunderbolt implementations.",[28,3180,3181,3182,3185],{},"The strategy rides the ",[32,3183,3184],{},"standard upgrade cycle",": as device and server standards evolve (USB4 Gen3, etc.), implementation requirements change, and Kandou's IP captures value at each transition. A recurring, standards-cycle-driven business model with a proven IP track record.",[353,3187,3189],{"id":3188},"case-study-nexthop-ai-500m-2026-series-b","Case Study: Nexthop AI ($500M, 2026 Series B)",[28,3191,3192],{},"a16z's $500M investment in Nexthop AI frames AI-era Ethernet switching as the next infrastructure layer.",[28,3194,3195],{},"The thesis: \"AI clusters are growing in scale faster than traditional switching architectures can adapt. Existing switches are not optimized for AI communication patterns (all-to-all collective communication).\" Nexthop builds AI-native switching to address this.",[28,3197,3198],{},[32,3199,3200],{},"Interconnect Investment Common Logic:",[91,3202,3203,3211],{},[94,3204,3205],{},[97,3206,3207,3209],{},[100,3208,3081],{},[100,3210,3084],{},[113,3212,3213,3221,3229,3237,3245],{},[97,3214,3215,3218],{},[118,3216,3217],{},"Technical driver",[118,3219,3220],{},"GPU/accelerator scaling makes I/O the bottleneck",[97,3222,3223,3226],{},[118,3224,3225],{},"Customer base",[118,3227,3228],{},"Hyperscalers (Google/AWS/Microsoft/Meta)",[97,3230,3231,3234],{},[118,3232,3233],{},"Capital intensity",[118,3235,3236],{},"Medium to high (manufacturing qualification required)",[97,3238,3239,3242],{},[118,3240,3241],{},"Risks",[118,3243,3244],{},"Production yield, supplier qualification, geopolitics",[97,3246,3247,3249],{},[118,3248,3115],{},[118,3250,3251],{},"M&A (major chip companies) or IPO",[58,3253],{},[20,3255,3257],{"id":3256},"domain-ai-accelerators-new-compute-betting-on-computing-paradigms","Domain③: AI Accelerators / New Compute — Betting on Computing Paradigms",[353,3259,3261],{"id":3260},"two-approaches-mega-seed-vs-vertical-integration","Two Approaches: Mega-Seed vs. Vertical Integration",[28,3263,3264],{},"Two distinct approaches exist for the \"change the computing paradigm\" thesis:",[2225,3266,3267,3273],{},[67,3268,3269,3272],{},[32,3270,3271],{},"Mega-seed in a startup"," (a16z × Unconventional AI): High risk, but success redefines the market.",[67,3274,3275,3278],{},[32,3276,3277],{},"Acquisition-based vertical integration"," (SoftBank × Graphcore/Ampere): Lower technology realization risk, but integration cost and market timing risk remain.",[28,3280,3281,3282,3285],{},"Both are ",[32,3283,3284],{},"ecosystem positioning plays"," — not simply chip investments.",[353,3287,3289],{"id":3288},"case-study-unconventional-ai-475m-2025-seed","Case Study: Unconventional AI ($475M, 2025 Seed)",[28,3291,3292],{},"One of the largest seed rounds in semiconductor investment history, co-led by a16z.",[28,3294,3295],{},"The analog/mixed-signal approach for probabilistic computing differs fundamentally from current digital CMOS architectures. Key challenges:",[64,3297,3298,3304,3310],{},[67,3299,3300,3303],{},[32,3301,3302],{},"Manufacturing reproducibility",": Analog circuits are sensitive to process variation",[67,3305,3306,3309],{},[32,3307,3308],{},"Toolchain",": Design and verification tools are immature",[67,3311,3312,3315],{},[32,3313,3314],{},"Ecosystem",": Software/framework adaptation required",[28,3317,3318,3319,3322],{},"But if successful, the potential for ",[32,3320,3321],{},"orders-of-magnitude power efficiency gains"," represents an overwhelming competitive advantage over current GPU architectures.",[28,3324,3325,3328],{},[32,3326,3327],{},"Why a16z makes this bet",": With its Strategic Partnerships program institutionalizing enterprise and government deployment paths, a16z can afford the early-stage cost of locking in a category-defining architecture. The cost of not owning the option is higher than the cost of the bet.",[353,3330,3332],{"id":3331},"softbanks-vertical-integration-arm-graphcore-ampere","SoftBank's Vertical Integration: Arm × Graphcore × Ampere",[91,3334,3335,3347],{},[94,3336,3337],{},[97,3338,3339,3342,3344],{},[100,3340,3341],{},"Company",[100,3343,2856],{},[100,3345,3346],{},"SoftBank Acquisition",[113,3348,3349,3359,3369],{},[97,3350,3351,3353,3356],{},[118,3352,2313],{},[118,3354,3355],{},"CPU IP, ISA standard, ecosystem core",[118,3357,3358],{},"Acquired 2016 ($32B)",[97,3360,3361,3363,3366],{},[118,3362,717],{},[118,3364,3365],{},"Graph-optimized AI processing",[118,3367,3368],{},"Acquired 2024",[97,3370,3371,3373,3376],{},[118,3372,804],{},[118,3374,3375],{},"Arm-based cloud server optimization",[118,3377,3378],{},"$6.5B acquisition announced 2025",[28,3380,3381,3382,3385],{},"This is a ",[32,3383,3384],{},"business strategy",", not a fund strategy. Evaluated on group synergy IRR, not VC fund returns.",[58,3387],{},[20,3389,3391],{"id":3390},"domain-quantum-long-term-but-capital-is-returning","Domain④: Quantum — Long-Term, But Capital Is Returning",[353,3393,3395],{"id":3394},"the-shift-to-commercialization-phase-quantum-investment","The Shift to \"Commercialization Phase\" Quantum Investment",[28,3397,3398,3399,1187],{},"The quantum boom of the early 2020s saw many hardware startups emerge, then face market mismatch. However, 2024–2025 has seen large rounds return — now structured around ",[32,3400,3401],{},"platform competition + cloud delivery",[28,3403,3404,3405],{},"The evaluation shift: from \"can quantum achieve quantum advantage?\" to ",[32,3406,3407],{},"\"can the company commercialize error correction, cloud delivery, and software stacks as an integrated system?\"",[353,3409,3411],{"id":3410},"case-study-quantum-circuits-series-b-60m-2024","Case Study: Quantum Circuits (Series B >$60M, 2024)",[28,3413,3414],{},"Sequoia's participation in this round signals conviction in fault-tolerant quantum architecture over near-term NISQ approaches. The bet is on the architecture that will scale — not the first to market.",[353,3416,3418],{"id":3417},"case-study-rigetti-computing-79m-2020-series-c","Case Study: Rigetti Computing ($79M, 2020 Series C)",[28,3420,3421],{},"Bessemer-led, with Quantum Cloud Services as the vertical integration model. Subsequently NASDAQ-listed via SPAC (2022). Having a cloud API entry point maintains enterprise access even before hardware reaches quantum advantage — a critical commercialization design choice.",[28,3423,3424],{},[32,3425,3426],{},"Quantum Investment Evaluation Criteria (Current):",[91,3428,3429,3438],{},[94,3430,3431],{},[97,3432,3433,3436],{},[100,3434,3435],{},"Axis",[100,3437,3084],{},[113,3439,3440,3448,3456,3463],{},[97,3441,3442,3445],{},[118,3443,3444],{},"Technology",[118,3446,3447],{},"Error rates, qubit count, coherence time",[97,3449,3450,3453],{},[118,3451,3452],{},"Commercialization",[118,3454,3455],{},"Cloud API, SDK, customer deployments",[97,3457,3458,3460],{},[118,3459,3314],{},[118,3461,3462],{},"Algorithm and software partners",[97,3464,3465,3468],{},[118,3466,3467],{},"Exit",[118,3469,3470],{},"Acquisition by major cloud (AWS/Azure/GCP) or IPO",[58,3472],{},[20,3474,3476],{"id":3475},"complete-investment-reference-table","Complete Investment Reference Table",[91,3478,3479,3498],{},[94,3480,3481],{},[97,3482,3483,3485,3488,3490,3492,3494,3496],{},[100,3484,247],{},[100,3486,3487],{},"Portfolio",[100,3489,950],{},[100,3491,1408],{},[100,3493,1411],{},[100,3495,1414],{},[100,3497,1086],{},[113,3499,3500,3516,3533,3549,3565,3581,3597,3613,3629,3645,3661,3678,3694,3711,3727,3744],{},[97,3501,3502,3504,3506,3508,3510,3512,3514],{},[118,3503,266],{},[118,3505,682],{},[118,3507,120],{},[118,3509,685],{},[118,3511,688],{},[118,3513,1475],{},[118,3515,1120],{},[97,3517,3518,3520,3522,3524,3526,3528,3530],{},[118,3519,266],{},[118,3521,647],{},[118,3523,642],{},[118,3525,650],{},[118,3527,653],{},[118,3529,656],{},[118,3531,3532],{},"India",[97,3534,3535,3537,3539,3541,3543,3545,3547],{},[118,3536,266],{},[118,3538,665],{},[118,3540,642],{},[118,3542,650],{},[118,3544,670],{},[118,3546,673],{},[118,3548,3532],{},[97,3550,3551,3553,3555,3557,3559,3561,3563],{},[118,3552,266],{},[118,3554,735],{},[118,3556,134],{},[118,3558,650],{},[118,3560,740],{},[118,3562,1563],{},[118,3564,1120],{},[97,3566,3567,3569,3571,3573,3575,3577,3579],{},[118,3568,1629],{},[118,3570,822],{},[118,3572,191],{},[118,3574,825],{},[118,3576,828],{},[118,3578,1638],{},[118,3580,1120],{},[97,3582,3583,3585,3587,3589,3591,3593,3595],{},[118,3584,283],{},[118,3586,752],{},[118,3588,134],{},[118,3590,650],{},[118,3592,757],{},[118,3594,1578],{},[118,3596,1120],{},[97,3598,3599,3601,3603,3605,3607,3609,3611],{},[118,3600,283],{},[118,3602,840],{},[118,3604,191],{},[118,3606,685],{},[118,3608,828],{},[118,3610,1653],{},[118,3612,1120],{},[97,3614,3615,3617,3619,3621,3623,3625,3627],{},[118,3616,300],{},[118,3618,699],{},[118,3620,120],{},[118,3622,1528],{},[118,3624,705],{},[118,3626,656],{},[118,3628,1120],{},[97,3630,3631,3633,3635,3637,3639,3641,3643],{},[118,3632,300],{},[118,3634,787],{},[118,3636,134],{},[118,3638,790],{},[118,3640,580],{},[118,3642,1608],{},[118,3644,1120],{},[97,3646,3647,3649,3651,3653,3655,3657,3659],{},[118,3648,316],{},[118,3650,769],{},[118,3652,134],{},[118,3654,772],{},[118,3656,775],{},[118,3658,1593],{},[118,3660,1120],{},[97,3662,3663,3665,3667,3669,3671,3673,3675],{},[118,3664,316],{},[118,3666,612],{},[118,3668,607],{},[118,3670,596],{},[118,3672,617],{},[118,3674,620],{},[118,3676,3677],{},"Switzerland",[97,3679,3680,3682,3684,3686,3688,3690,3692],{},[118,3681,316],{},[118,3683,2676],{},[118,3685,607],{},[118,3687,596],{},[118,3689,634],{},[118,3691,1475],{},[118,3693,1120],{},[97,3695,3696,3698,3700,3702,3704,3706,3708],{},[118,3697,316],{},[118,3699,574],{},[118,3701,569],{},[118,3703,1427],{},[118,3705,580],{},[118,3707,1432],{},[118,3709,3710],{},"Israel",[97,3712,3713,3715,3717,3719,3721,3723,3725],{},[118,3714,333],{},[118,3716,593],{},[118,3718,588],{},[118,3720,596],{},[118,3722,599],{},[118,3724,602],{},[118,3726,1120],{},[97,3728,3729,3731,3733,3735,3737,3739,3741],{},[118,3730,714],{},[118,3732,717],{},[118,3734,120],{},[118,3736,720],{},[118,3738,1545],{},[118,3740,1548],{},[118,3742,3743],{},"UK",[97,3745,3746,3748,3750,3752,3754,3756,3758],{},[118,3747,714],{},[118,3749,804],{},[118,3751,134],{},[118,3753,720],{},[118,3755,809],{},[118,3757,812],{},[118,3759,1120],{},[58,3761],{},[20,3763,1659],{"id":1658},[855,3765,3766,3771,3774,3777],{},[28,3767,3768],{},[32,3769,3770],{},"A large round does not mean high probability — it means high expected value",[28,3772,3773],{},"The $475M seed into Unconventional AI, and the $500M rounds into Nexthop and Ayar Labs, are not bets on certainty. They are bets where \"if wrong, we lose the check; if right, we create a new category worth hundreds of billions.\" The asymmetry justifies the size.",[28,3775,3776],{},"The area I find most technically compelling is CPO. Co-packaging optical I/O with silicon sits at the intersection of semiconductors, photonics, and precision packaging — three industries where Japan has globally competitive players. This may be one of the few \"cutting-edge intersections\" where Japanese industry can participate at the frontier.",[28,3778,3779],{},"Part 3 (the final installment) synthesizes the strategic implications: how to construct a portfolio across these themes, and how to think about the risk-reward across short, medium, and long time horizons.",{"title":229,"searchDepth":1682,"depth":1682,"links":3781},[3782,3783,3784,3785,3786,3791,3798,3803,3808,3809,3810,3811,3812,3813,3814,3819,3826,3831,3836,3837],{"id":1750,"depth":1682,"text":1751},{"id":26,"depth":1682,"text":26},{"id":62,"depth":1682,"text":62},{"id":1791,"depth":1682,"text":1792},{"id":1999,"depth":1682,"text":2000,"children":3787},[3788,3789,3790],{"id":2003,"depth":1693,"text":2004},{"id":2010,"depth":1693,"text":2011},{"id":2024,"depth":1693,"text":2025},{"id":2095,"depth":1682,"text":2096,"children":3792},[3793,3794,3795,3796,3797],{"id":2099,"depth":1693,"text":2099},{"id":2111,"depth":1693,"text":2112},{"id":2127,"depth":1693,"text":2128},{"id":2137,"depth":1693,"text":2138},{"id":2147,"depth":1693,"text":2148},{"id":2215,"depth":1682,"text":2216,"children":3799},[3800,3801,3802],{"id":2219,"depth":1693,"text":2220},{"id":2248,"depth":1693,"text":2249},{"id":2287,"depth":1693,"text":2288},{"id":2347,"depth":1682,"text":2348,"children":3804},[3805,3806,3807],{"id":2351,"depth":1693,"text":2352},{"id":2365,"depth":1693,"text":2366},{"id":2375,"depth":1693,"text":2376},{"id":2436,"depth":1682,"text":2437},{"id":852,"depth":1682,"text":853},{"id":2791,"depth":1682,"text":2792},{"id":880,"depth":1682,"text":881},{"id":914,"depth":1682,"text":915},{"id":2836,"depth":1682,"text":2837},{"id":3035,"depth":1682,"text":3036,"children":3815},[3816,3817,3818],{"id":3039,"depth":1693,"text":3040},{"id":3046,"depth":1693,"text":3047},{"id":3060,"depth":1693,"text":3061},{"id":3131,"depth":1682,"text":3132,"children":3820},[3821,3822,3823,3824,3825],{"id":3135,"depth":1693,"text":3136},{"id":3148,"depth":1693,"text":3149},{"id":3164,"depth":1693,"text":3165},{"id":3174,"depth":1693,"text":3175},{"id":3188,"depth":1693,"text":3189},{"id":3256,"depth":1682,"text":3257,"children":3827},[3828,3829,3830],{"id":3260,"depth":1693,"text":3261},{"id":3288,"depth":1693,"text":3289},{"id":3331,"depth":1693,"text":3332},{"id":3390,"depth":1682,"text":3391,"children":3832},[3833,3834,3835],{"id":3394,"depth":1693,"text":3395},{"id":3410,"depth":1693,"text":3411},{"id":3417,"depth":1693,"text":3418},{"id":3475,"depth":1682,"text":3476},{"id":1658,"depth":1682,"text":1659},"GPU性能が伸びるほど設計・I/O・実装がAIの限界要因になる。CPO・SerDes・EDA自動化・量子にわたる代表案件を技術視点で深掘り。$500M超の大型ラウンドが示す投資ロジックを解読する。",{"date":1717,"image":3840,"alt":3841,"tags":3842,"tagsEn":3844,"published":1732},"/blogs-img/blog-semiconductor-vc-2.png","半導体投資の技術領域別ケーススタディ",[1721,1722,1723,3843,1724],"フィジカルAI",[1727,1728,1729,3845,1730],"Physical AI","/blogs/7-semiconductor-vc-series-2",{"title":1739,"description":3838},"blogs/7-semiconductor-vc-series-2","-ywFilRUg1zg9yFBaMw2NC0snzsGm14MEamxZbdax9E",{"id":3851,"title":3852,"body":3853,"description":5420,"extension":1715,"meta":5421,"navigation":1732,"ogImage":5422,"path":5426,"seo":5427,"stem":5428,"__hash__":5429},"content/blogs/8-semiconductor-vc-series-3.md","半導体×VCの勝ち筋は「ボラティリティを設計する」こと｜連載③戦略的示唆と投資提案 | Semiconductor VC Series③: Designing for Volatility",{"type":8,"value":3854,"toc":5376},[3855,4606,4608],[11,3856,3857,3861,3866,3868,3872,3875,3878,3881,3895,3897,3899,3907,3915,3923,3925,3929,4045,4049,4055,4057,4061,4065,4126,4132,4134,4138,4210,4216,4218,4222,4294,4300,4302,4306,4427,4430,4436,4438,4441,4444,4450,4453,4456,4482,4486,4500,4502,4505,4509,4553,4557,4589,4591],{"lang":13},[15,3858,3860],{"id":3859},"半導体vcの勝ち筋はボラティリティを設計すること","半導体×VCの勝ち筋は「ボラティリティを設計する」こと",[28,3862,3863],{},[32,3864,3865],{},"連載③：戦略的示唆と投資提案",[58,3867],{},[20,3869,3871],{"id":3870},"はじめにボラティリティを設計するとは何か","はじめに：「ボラティリティを設計する」とは何か",[28,3873,3874],{},"半導体産業は本質的にサイクリカルだ。需要は爆発的に膨らみ、供給は2〜3年後に追いつき、価格は暴落する——半導体の歴史はその繰り返しだった。しかし今、Tier1 VCが半導体スタートアップに数億ドルを投じるとき、彼らは「サイクルの波に乗る」のではなく、「サイクル自体を無効化する構造的ポジション」を狙っている。",[28,3876,3877],{},"それが「ボラティリティを設計する（designing for volatility）」という概念だ。チップの価格が上下しても需要が安定する「標準化されたインターフェース技術」（CPO、UCIeなど）、チップ世代交代のたびに価値が高まる「設計インフラ」（EDA自動化、IP再利用）、そして特定のハードウェアベンダーに依存しない「ソフトウェア定義の計算基盤」への投資がそれにあたる。",[28,3879,3880],{},"本連載の最終回では、連載①②で整理したVCの戦略・技術・案件データを踏まえ、以下を体系化する：",[2225,3882,3883,3886,3889,3892],{},[67,3884,3885],{},"Tier1 VC 5社の戦略比較と差異の本質",[67,3887,3888],{},"短期（〜18ヶ月）・中期（2〜4年）・長期（5年〜）の投資機会とリスク",[67,3890,3891],{},"マクロ環境（市場規模・装置投資・政策）のサマリ",[67,3893,3894],{},"個人投資家・機関投資家・事業会社それぞれへの示唆",[58,3896],{},[20,3898,62],{"id":62},[855,3900,3901],{},[28,3902,3903,3906],{},[32,3904,3905],{},"①「AI需要→半導体ボトルネック→スタートアップ機会」という構造は2026年以降も継続する","\nGPU性能のスケーリングが続く限り、設計・データ移動・パッケージングというボトルネックは解消されない。Tier1 VCはこの構造的需要を見越し、すでに2023〜2025年に大型投資を実行済みだ。",[855,3908,3909],{},[28,3910,3911,3914],{},[32,3912,3913],{},"②VC各社の\"賭け方\"は大きく異なる：エコシステム型 vs. 垂直統合型","\nSequoia・a16zがスタートアップ群への分散投資でエコシステムを形成するのに対し、SoftBankはAmpere（$6.5B評価額）など垂直統合型の大型単独案件を選好する。Kleiner PerkinsはIPO経路を重視した米国製造復活テーマに集中し、Bessemerはデータセンターインフラとのシナジーを軸に据える。",[855,3916,3917],{},[28,3918,3919,3922],{},[32,3920,3921],{},"③長期は「新計算原理」が最大の非線形リターン源だが、商用化リスクは高い","\nアナログ計算・量子・フォトニクスはいずれも現在の投資期待値が低い一方、商用化に成功した場合のリターンは既存手法と非線形に差がつく。VCはこれらをポートフォリオの「オプション」として保有し、中期の確実な案件でポートフォリオを安定させる戦略をとっている。",[58,3924],{},[20,3926,3928],{"id":3927},"tier1-vc戦略比較差異の本質","Tier1 VC戦略比較：差異の本質",[91,3930,3931,3949],{},[94,3932,3933],{},[97,3934,3935,3937,3940,3943,3946],{},[100,3936,247],{},[100,3938,3939],{},"投資の\"型\"",[100,3941,3942],{},"代表案件",[100,3944,3945],{},"提携・エコシステム構造",[100,3947,3948],{},"想定出口",[113,3950,3951,3970,3989,4007,4026],{},[97,3952,3953,3958,3961,3964,3967],{},[118,3954,3955],{},[32,3956,3957],{},"Sequoia Capital",[118,3959,3960],{},"分散・エコシステム型。設計自動化・新計算原理を中心に複数スタートアップへ初期投資",[118,3962,3963],{},"Ricursive ($35M)、Unconventional Computing ($475M)",[118,3965,3966],{},"NVIDIA・Synopsys・Coreweaveとのネットワークで被投資企業を繋ぐ",[118,3968,3969],{},"M&A（EDA大手）またはIPO（計算インフラ）",[97,3971,3972,3977,3980,3983,3986],{},[118,3973,3974],{},[32,3975,3976],{},"Andreessen Horowitz (a16z)",[118,3978,3979],{},"テーマ型・大型集中。「AIファクトリー」インフラを横断するデータ移動と新世代メモリ",[118,3981,3982],{},"Ayar Labs ($500M)、Nexthop AI ($500M)",[118,3984,3985],{},"クラウドプロバイダー（AWS・Azure・GCP）とのコマーシャルパイプライン形成",[118,3987,3988],{},"データセンター事業者へのM&AまたはIPO",[97,3990,3991,3995,3998,4001,4004],{},[118,3992,3993],{},[32,3994,300],{},[118,3996,3997],{},"政策・製造復活テーマ型。CHIPS法補助金の受益企業、米国製造回帰",[118,3999,4000],{},"Retym ($75M)、Kandou Technologies ($92.3M)",[118,4002,4003],{},"SEMI協会・米国防総省DARPAプログラムとの連携",[118,4005,4006],{},"IPO（CHIPS法関連企業への市場期待）",[97,4008,4009,4014,4017,4020,4023],{},[118,4010,4011],{},[32,4012,4013],{},"Bessemer Venture Partners",[118,4015,4016],{},"データセンターインフラ連動型。既存クラウド投資との シナジーを重視",[118,4018,4019],{},"Quantum Circuits ($60M+)、ChipAgents ($21M)",[118,4021,4022],{},"既存クラウド・SaaS投資先との顧客共有",[118,4024,4025],{},"M&A（クラウド大手による内製化）",[97,4027,4028,4033,4036,4039,4042],{},[118,4029,4030],{},[32,4031,4032],{},"SoftBank Vision Fund",[118,4034,4035],{},"垂直統合・メガラウンド型。単一企業に大規模資本を集中、グループ内シナジー優先",[118,4037,4038],{},"Ampere Computing ($6.5B評価額)、Graphcore買収",[118,4040,4041],{},"SBグループ（ARM・T-Mobile・WeWork残存網）との垂直統合",[118,4043,4044],{},"長期保有またはIPO（ARM上場の先例）",[353,4046,4048],{"id":4047},"戦略の本質的差異エコシステム型vs垂直統合型","戦略の本質的差異：「エコシステム型」vs.「垂直統合型」",[220,4050,4053],{"className":4051,"code":4052,"language":225},[223],"エコシステム型（Sequoia / a16z / KP / Bessemer）\n────────────────────────────────────────────\nVC\n ├── スタートアップA（設計）\n ├── スタートアップB（インターコネクト）\n ├── スタートアップC（パッケージング）\n └── スタートアップD（新計算）\n        │\n        ▼\n 各社が相互に顧客・サプライヤー関係を形成\n → VC自身がエコシステムのハブになる\n\n垂直統合型（SoftBank）\n────────────────────────────────────────────\nSoftBank\n └── Ampere Computing（独自Arm CPU設計）\n      └── ARM（SB傘下）からIPライセンス\n           └── データセンター向け直販モデル\n                → グループ内でValue Chainを完結\n",[227,4054,4052],{"__ignoreMap":229},[58,4056],{},[20,4058,4060],{"id":4059},"投資機会とリスク3つの時間軸","投資機会とリスク：3つの時間軸",[353,4062,4064],{"id":4063},"短期18ヶ月設計効率化とeda自動化","短期（〜18ヶ月）：設計効率化とEDA自動化",[91,4066,4067,4083],{},[94,4068,4069],{},[97,4070,4071,4074,4077,4080],{},[100,4072,4073],{},"機会領域",[100,4075,4076],{},"投資テーマ",[100,4078,4079],{},"代表スタートアップ",[100,4081,4082],{},"主なリスク",[113,4084,4085,4098,4112],{},[97,4086,4087,4089,4092,4095],{},[118,4088,778],{},[118,4090,4091],{},"LLMによる回路検証・バグ修正の自動化で設計期間を30〜50%短縮",[118,4093,4094],{},"Ricursive（Seq.）、ChipAgents（Bessemer）",[118,4096,4097],{},"Synopsys/Cadenceがin-house LLM開発を加速、M&Aで吸収する可能性",[97,4099,4100,4103,4106,4109],{},[118,4101,4102],{},"IP再利用・モジュール化",[118,4104,4105],{},"既存IP資産のAI検索・カスタマイズにより新チップ設計コスト削減",[118,4107,4108],{},"複数の非公開ステルス案件",[118,4110,4111],{},"大手ファブ（TSMC・Samsung）が独自IPライブラリを強化、差別化困難",[97,4113,4114,4117,4120,4123],{},[118,4115,4116],{},"サプライチェーン可視化",[118,4118,4119],{},"地政学リスクにより部品調達の不確実性が増大、可視化SaaSへの需要急増",[118,4121,4122],{},"（非公開多数）",[118,4124,4125],{},"景気後退局面ではソフトウェア支出が最初に削減される",[28,4127,4128,4131],{},[32,4129,4130],{},"短期の投資判断基準","：収益化期間が短い（18ヶ月以内に初期顧客獲得可能）か、規制の影響を受けにくいソフトウェアファーストモデルかどうかが鍵。EDA自動化はこの条件を満たすが、競合（Cadence・Synopsys）の大型R&D予算には要注意。",[58,4133],{},[353,4135,4137],{"id":4136},"中期24年データ移動cpo先端パッケージング","中期（2〜4年）：データ移動・CPO・先端パッケージング",[91,4139,4140,4152],{},[94,4141,4142],{},[97,4143,4144,4146,4148,4150],{},[100,4145,4073],{},[100,4147,4076],{},[100,4149,4079],{},[100,4151,4082],{},[113,4153,4154,4168,4182,4196],{},[97,4155,4156,4159,4162,4165],{},[118,4157,4158],{},"CPO（コパッケージド光学）",[118,4160,4161],{},"GPU-メモリ間の銅配線を光配線に置換、帯域幅を10倍・消費電力を60〜80%削減",[118,4163,4164],{},"Ayar Labs（a16z、$500M）",[118,4166,4167],{},"量産歩留まり（光コンポーネントの製造難易度）、TSMC/IntelのCPO内製化リスク",[97,4169,4170,4173,4176,4179],{},[118,4171,4172],{},"高速SerDes・電気インターコネクト",[118,4174,4175],{},"銅配線の延命技術としてデータセンター内配線を高効率化",[118,4177,4178],{},"Kandou Technologies（KP、$92.3M）",[118,4180,4181],{},"光学への移行が予想より早まった場合のレガシー化リスク",[97,4183,4184,4187,4190,4193],{},[118,4185,4186],{},"先端パッケージング（Chiplet/3D IC）",[118,4188,4189],{},"Chiplet標準規格（UCIe）による異種チップ統合",[118,4191,4192],{},"Nexthop AI（a16z、$500M）",[118,4194,4195],{},"ファブレスとファブの間のIP所有権交渉の複雑化",[97,4197,4198,4201,4204,4207],{},[118,4199,4200],{},"次世代メモリインターフェース",[118,4202,4203],{},"HBM3/4の後継として帯域幅と電力効率のトレードオフを再設計",[118,4205,4206],{},"（SK Hynix・Micronとの協業スタートアップ）",[118,4208,4209],{},"DRAMメーカーの自社開発能力向上でサードパーティ不要論",[28,4211,4212,4215],{},[32,4213,4214],{},"中期の投資判断基準","：TSMC・Samsung・Intelの先端パッケージング拡張計画と競合するかどうか。競合するなら提携関係（JDA）を確立できているかが生死を分ける。Ayar Labsがa16zの支援のもとTSMCと共同開発体制を構築しているのはこの典型例。",[58,4217],{},[353,4219,4221],{"id":4220},"長期5年新計算原理と量子","長期（5年〜）：新計算原理と量子",[91,4223,4224,4236],{},[94,4225,4226],{},[97,4227,4228,4230,4232,4234],{},[100,4229,4073],{},[100,4231,4076],{},[100,4233,4079],{},[100,4235,4082],{},[113,4237,4238,4252,4266,4280],{},[97,4239,4240,4243,4246,4249],{},[118,4241,4242],{},"アナログ計算",[118,4244,4245],{},"デジタル変換コストを排除し推論電力を100分の1以下に削減、エッジAI向け",[118,4247,4248],{},"Unconventional Computing（Seq.、$475M）",[118,4250,4251],{},"プログラマビリティ（汎用性）の制限、開発者エコシステムの未成熟",[97,4253,4254,4257,4260,4263],{},[118,4255,4256],{},"量子コンピューティング",[118,4258,4259],{},"素因数分解・量子化学シミュレーション・最適化問題での超越性",[118,4261,4262],{},"Quantum Circuits（Bessemer、$60M+）、Rigetti（$79M）",[118,4264,4265],{},"エラー訂正技術の成熟度、「量子優位性」の商用ユースケース特定",[97,4267,4268,4271,4274,4277],{},[118,4269,4270],{},"フォトニック計算",[118,4272,4273],{},"光を用いた行列演算でAI推論の消費電力を抜本的に削減",[118,4275,4276],{},"（非公開複数社）",[118,4278,4279],{},"シリコンフォトニクスの製造インフラ未成熟、スケールアップコスト",[97,4281,4282,4285,4288,4291],{},[118,4283,4284],{},"ニューロモルフィック",[118,4286,4287],{},"脳の神経回路を模倣した非同期計算、超低電力センサー処理",[118,4289,4290],{},"（学術スピンアウト複数）",[118,4292,4293],{},"汎用計算への転用困難、ニッチ市場からの脱出経路不明確",[28,4295,4296,4299],{},[32,4297,4298],{},"長期の投資判断基準","：「論文→プロダクト→市場」の接続性。量子の場合、論文数と特許出願は世界トップクラスだが、商用ユースケースへの接続（特にエラー率の商用許容水準到達）が見えていない企業は評価を保留すべきだ。一方、アナログ計算はエッジ推論という明確な市場セグメントが存在し、中期以降の商用化シナリオが描きやすい。",[58,4301],{},[20,4303,4305],{"id":4304},"マクロ環境市場規模装置投資政策","マクロ環境：市場規模・装置投資・政策",[91,4307,4308,4330],{},[94,4309,4310],{},[97,4311,4312,4315,4318,4321,4324,4327],{},[100,4313,4314],{},"指標",[100,4316,4317],{},"2024実績",[100,4319,4320],{},"2025予測",[100,4322,4323],{},"2026予測",[100,4325,4326],{},"2027予測",[100,4328,4329],{},"出典",[113,4331,4332,4350,4368,4388,4408],{},[97,4333,4334,4337,4340,4342,4344,4347],{},[118,4335,4336],{},"世界半導体市場規模",[118,4338,4339],{},"$627B",[118,4341,137],{},[118,4343,38],{},[118,4345,4346],{},"$1.1T+",[118,4348,4349],{},"WSTS 2025年秋季予測",[97,4351,4352,4355,4358,4361,4363,4365],{},[118,4353,4354],{},"300mm装置支出",[118,4356,4357],{},"$107B",[118,4359,4360],{},"$119B",[118,4362,42],{},[118,4364,46],{},[118,4366,4367],{},"SEMI 2026年予測",[97,4369,4370,4373,4376,4379,4382,4385],{},[118,4371,4372],{},"AI半導体（GPU/NPU/ASIC）",[118,4374,4375],{},"$130B",[118,4377,4378],{},"$200B+",[118,4380,4381],{},"$280B+",[118,4383,4384],{},"$380B+",[118,4386,4387],{},"IDC/Gartner各社推計",[97,4389,4390,4393,4396,4399,4402,4405],{},[118,4391,4392],{},"米CHIPS法補助金執行額",[118,4394,4395],{},"$8B",[118,4397,4398],{},"$30B+",[118,4400,4401],{},"$52B（総額上限）",[118,4403,4404],{},"〃",[118,4406,4407],{},"米商務省",[97,4409,4410,4413,4416,4419,4422,4424],{},[118,4411,4412],{},"中国半導体輸出規制強化",[118,4414,4415],{},"HBM・EDA規制",[118,4417,4418],{},"EUV輸出制限継続",[118,4420,4421],{},"さらなる強化見込み",[118,4423,4404],{},[118,4425,4426],{},"BIS/経産省",[353,4428,4429],{"id":4429},"政策環境が投資判断に与える影響",[220,4431,4434],{"className":4432,"code":4433,"language":225},[223],"CHIPS法・EU半導体法\n    │\n    ├── 補助金の受益企業（Intel Foundry・TSMC米国拠点・Samsung Texas）\n    │        └── → 建設・装置・材料スタートアップに間接需要\n    │\n    └── 地政学リスクの優先度上昇\n             └── → 米国内設計・製造のスタートアップへのVC資金流入加速\n\n中国規制（BIS 輸出規制）\n    │\n    ├── 中国向けAI半導体（H800→H20）の販売制限\n    │        └── → 中国内半導体自給自足投資の爆発（华为/寒武纪等）\n    │\n    └── 先進EDA・製造装置の中国輸出禁止\n             └── → 欧米スタートアップへの反射的恩恵（競合中国企業の能力制限）\n",[227,4435,4433],{"__ignoreMap":229},[58,4437],{},[20,4439,4440],{"id":4440},"ポートフォリオ構成の考え方",[28,4442,4443],{},"Tier1 VCの実際のポートフォリオ配分から読み取れるのは、以下のような「時間軸別の3層構造」だ：",[220,4445,4448],{"className":4446,"code":4447,"language":225},[223],"ポートフォリオの理想的な時間軸配分（イメージ）\n─────────────────────────────────────\n短期（〜18ヶ月）  ████████████████░░░░  40%\n  └── EDA自動化・設計SaaS・サプライチェーン可視化\n\n中期（2〜4年）    ██████████████████░░  45%\n  └── CPO・インターコネクト・先端パッケージング・AIアクセラレータ\n\n長期（5年〜）     ██████░░░░░░░░░░░░░░  15%\n  └── アナログ計算・量子・フォトニクス・ニューロモルフィック\n─────────────────────────────────────\n",[227,4449,4447],{"__ignoreMap":229},[353,4451,4452],{"id":4452},"個人投資家への示唆",[28,4454,4455],{},"個人投資家が直接VCファンドに参加できるケースは限られるが、以下の方法で間接的な半導体スタートアップエクスポージャーを取得できる：",[2225,4457,4458,4464,4470,4476],{},[67,4459,4460,4463],{},[32,4461,4462],{},"上場半導体企業（NVIDIA・ASML・Lam Research・TSMC ADR）","：スタートアップの顧客であり、スタートアップが解決するボトルネックから直接利益を得る",[67,4465,4466,4469],{},[32,4467,4468],{},"半導体特化ETF（SOXX・SMH）","：分散効果はあるが、大型株中心でスタートアップの成長性は反映されにくい",[67,4471,4472,4475],{},[32,4473,4474],{},"SPAC・二次流通市場","：Rigetti（量子）のようにSPAC経由で上場した銘柄は、高リスク・高ボラティリティで個人投資家がアクセスしやすい",[67,4477,4478,4481],{},[32,4479,4480],{},"CVCを持つ事業会社株","：SoftBankグループ・Intel Capital・Qualcomm Ventures母体企業への投資は、VCポートフォリオへの間接アクセスとなる",[353,4483,4485],{"id":4484},"機関投資家事業会社への示唆","機関投資家・事業会社への示唆",[64,4487,4488,4494],{},[67,4489,4490,4493],{},[32,4491,4492],{},"LP（大学基金・年金）","：Sequoia / a16z / Bessemerへのコミットメントは半導体の長期テーマと合致するが、J-Curveリスク（投資後3〜5年はリターンがマイナス）を許容できる長期資金が前提",[67,4495,4496,4499],{},[32,4497,4498],{},"事業会社CVC","：自社バリューチェーンのボトルネックを特定し、そこに対応するスタートアップへ戦略的投資することで、M&Aオプションと市場情報を同時に取得できる",[58,4501],{},[20,4503,4504],{"id":4504},"連載まとめと次の注目ポイント",[353,4506,4508],{"id":4507},"_3回の連載で明らかになったこと","3回の連載で明らかになったこと",[91,4510,4511,4521],{},[94,4512,4513],{},[97,4514,4515,4518],{},[100,4516,4517],{},"連載",[100,4519,4520],{},"核心メッセージ",[113,4522,4523,4533,4543],{},[97,4524,4525,4530],{},[118,4526,4527],{},[32,4528,4529],{},"連載①：概観と主要プレイヤー",[118,4531,4532],{},"Tier1 VCは「AI需要のボトルネック」という構造的欠如部分に集中している。チップ単体ではなく、設計・移動・実装・供給網という4層への分解投資が共通パターン",[97,4534,4535,4540],{},[118,4536,4537],{},[32,4538,4539],{},"連載②：技術別ケーススタディ",[118,4541,4542],{},"EDA自動化・CPO・先端パッケージング・新計算原理という4領域で約$2.5B規模の投資が2022〜2025年に集中執行された。技術的成熟度と商用化パスが投資規模を決定している",[97,4544,4545,4550],{},[118,4546,4547],{},[32,4548,4549],{},"連載③：戦略的示唆（本稿）",[118,4551,4552],{},"VCの「型」はエコシステム型と垂直統合型に二分される。短期はソフトウェア収益性、中期はCPO/パッケージング、長期はアナログ・量子がそれぞれのリターン源",[353,4554,4556],{"id":4555},"今後の注目ポイント2026年後半以降","今後の注目ポイント（2026年後半以降）",[2225,4558,4559,4565,4571,4577,4583],{},[67,4560,4561,4564],{},[32,4562,4563],{},"Ayar Labs量産開始タイミング","：a16zが$500M投じたCPOの量産が2026〜2027年に本格化するか。歩留まりが商用水準（90%以上）に達するかが業界全体のCPO移行速度を決定する",[67,4566,4567,4570],{},[32,4568,4569],{},"Ampere ComputingのIPO計画","：SoftBankが2026年中に示唆しているAmpereのIPOが実現すれば、Arm系CPUのデータセンター向け市場シェアの公開データが初めて入手可能になる",[67,4572,4573,4576],{},[32,4574,4575],{},"CHIPS法補助金の第2フェーズ","：2026年後半に予定される米商務省の第2ラウンド補助金決定（ファウンドリー以外の設計・材料・EDA企業への配分拡大）が、スタートアップへの資金流入を加速させる可能性がある",[67,4578,4579,4582],{},[32,4580,4581],{},"中国の先進ファブリケーション突破口","：HuaweiのKirin/Ascend系チップが7nm相当に到達した実績を踏まえ、5nm以下への突破タイミングが欧米規制強化の速度と「技術ギャップ」の行方を決める",[67,4584,4585,4588],{},[32,4586,4587],{},"量子エラー訂正の商用許容水準到達","：Google（Willow）・IBM・Microsoftが競う論理量子ビットの実装が2027〜2028年に商用ユースケース（金融最適化・創薬）に接続できるか",[58,4590],{},[855,4592,4593,4597,4600,4603],{},[28,4594,4595],{},[32,4596,853],{},[28,4598,4599],{},"半導体投資の本質は「インフラのロックイン」だ。GPUもメモリも、単体では差別化できない。しかし、その間をつなぐデータ移動技術（CPO・SerDes）、上流の設計効率（EDA自動化）、下流の実装技術（先端パッケージング）は、一度採用されたらすぐには切り替えられない粘着性（スティッキネス）がある。",[28,4601,4602],{},"Tier1 VCがこの領域に$2.5B以上を集中させているのは、「チップの価格競争に巻き込まれたくない」という合理的判断の結果だ。価格競争は製造業の宿命だが、インフラはソフトウェア的なネットワーク効果を持てる。",[28,4604,4605],{},"連載を通じて改めて感じたのは、「半導体は地政学と分離できない」という事実だ。CHIPS法・BIS輸出規制・日本の経済安保法——これらの政策変数が、純粋な技術評価と同等以上に投資判断を左右する。個人投資家として半導体セクターに向き合うなら、技術だけでなく政策カレンダーを同時に追うことが不可欠だと考えている。",[58,4607],{},[11,4609,4610,4614,4619,4621,4625,4628,4643,4646,4660,4662,4664,4672,4680,4688,4690,4694,4806,4810,4816,4818,4822,4826,4888,4894,4896,4900,4972,4978,4980,4984,5056,5062,5064,5068,5178,5182,5188,5190,5194,5201,5207,5211,5214,5240,5244,5258,5260,5264,5268,5312,5316,5348,5350],{"lang":873},[15,4611,4613],{"id":4612},"semiconductor-vc-designing-for-volatility","Semiconductor VC: Designing for Volatility",[28,4615,4616],{},[32,4617,4618],{},"Series③: Strategic Implications and Investment Thesis",[58,4620],{},[20,4622,4624],{"id":4623},"introduction-what-does-designing-for-volatility-mean","Introduction: What Does \"Designing for Volatility\" Mean?",[28,4626,4627],{},"The semiconductor industry is inherently cyclical. Demand explodes, supply catches up 2–3 years later, prices collapse — the history of semiconductors has always been one of boom and bust. But when Tier1 VCs commit hundreds of millions of dollars to semiconductor startups today, they are not trying to \"ride the cycle.\" They are targeting structural positions that are immune to the cycle itself.",[28,4629,4630,4631,4634,4635,4638,4639,4642],{},"This is what \"designing for volatility\" means: investing in ",[32,4632,4633],{},"standardized interface technologies"," (CPO, UCIe) whose demand remains stable even when chip prices swing wildly; ",[32,4636,4637],{},"design infrastructure"," (EDA automation, IP reuse) that becomes more valuable with every chip generation transition; and ",[32,4640,4641],{},"software-defined computing foundations"," that are independent of any specific hardware vendor.",[28,4644,4645],{},"In this final installment, I synthesize the strategy, technology, and deal data from Parts 1 and 2 into four key frameworks:",[2225,4647,4648,4651,4654,4657],{},[67,4649,4650],{},"Comparative analysis of five Tier1 VCs and the essence of their strategic differences",[67,4652,4653],{},"Investment opportunities and risks across short (≤18 months), medium (2–4 years), and long (5+ years) time horizons",[67,4655,4656],{},"Macro environment summary: market size, equipment investment, and policy",[67,4658,4659],{},"Implications for individual investors, institutional investors, and corporate strategists",[58,4661],{},[20,4663,915],{"id":914},[855,4665,4666],{},[28,4667,4668,4671],{},[32,4669,4670],{},"① The \"AI demand → semiconductor bottleneck → startup opportunity\" structure will persist beyond 2026","\nAs long as GPU performance scaling continues, bottlenecks in design, data movement, and packaging will remain unsolved. Tier1 VCs have already anticipated this structural demand with major investments executed in 2023–2025.",[855,4673,4674],{},[28,4675,4676,4679],{},[32,4677,4678],{},"② VC firms' investment \"styles\" diverge sharply: ecosystem model vs. vertical integration model","\nWhile Sequoia and a16z build ecosystems through diversified investments in startup clusters, SoftBank concentrates capital in large single-company vertical integration bets (e.g., Ampere at $6.5B valuation). Kleiner Perkins focuses on the U.S. manufacturing revival thesis with clear IPO paths, while Bessemer emphasizes synergies with its existing data center infrastructure portfolio.",[855,4681,4682],{},[28,4683,4684,4687],{},[32,4685,4686],{},"③ Long-term, \"new computing paradigms\" offer the highest non-linear return potential — but commercialization risk is high","\nAnalog computing, quantum, and photonics all carry low near-term expected values, but successful commercialization creates non-linear return gaps versus conventional approaches. VCs hold these as \"options\" within diversified portfolios, using medium-term certainties to stabilize overall returns.",[58,4689],{},[20,4691,4693],{"id":4692},"tier1-vc-strategy-comparison-the-essence-of-differences","Tier1 VC Strategy Comparison: The Essence of Differences",[91,4695,4696,4714],{},[94,4697,4698],{},[97,4699,4700,4702,4705,4708,4711],{},[100,4701,247],{},[100,4703,4704],{},"Investment Style",[100,4706,4707],{},"Signature Deals",[100,4709,4710],{},"Ecosystem/Partnership Structure",[100,4712,4713],{},"Expected Exit",[113,4715,4716,4734,4752,4770,4788],{},[97,4717,4718,4722,4725,4728,4731],{},[118,4719,4720],{},[32,4721,3957],{},[118,4723,4724],{},"Diversified ecosystem. Early-stage bets across design automation and novel computing paradigms",[118,4726,4727],{},"Ricursive ($35M), Unconventional Computing ($475M)",[118,4729,4730],{},"Network bridges portfolio companies to NVIDIA, Synopsys, Coreweave",[118,4732,4733],{},"M&A (EDA majors) or IPO (compute infra)",[97,4735,4736,4740,4743,4746,4749],{},[118,4737,4738],{},[32,4739,3976],{},[118,4741,4742],{},"Thematic, large concentrated bets on \"AI factory\" infrastructure spanning data movement and next-gen memory",[118,4744,4745],{},"Ayar Labs ($500M), Nexthop AI ($500M)",[118,4747,4748],{},"Commercial pipeline formation with cloud providers (AWS, Azure, GCP)",[118,4750,4751],{},"M&A by data center operators or IPO",[97,4753,4754,4758,4761,4764,4767],{},[118,4755,4756],{},[32,4757,300],{},[118,4759,4760],{},"Policy/manufacturing revival theme. CHIPS Act beneficiaries, U.S. onshoring thesis",[118,4762,4763],{},"Retym ($75M), Kandou Technologies ($92.3M)",[118,4765,4766],{},"SEMI association, U.S. DoD/DARPA program linkages",[118,4768,4769],{},"IPO (market enthusiasm for CHIPS Act beneficiaries)",[97,4771,4772,4776,4779,4782,4785],{},[118,4773,4774],{},[32,4775,4013],{},[118,4777,4778],{},"Data center infrastructure synergy. Prioritizes fit with existing cloud investment portfolio",[118,4780,4781],{},"Quantum Circuits ($60M+), ChipAgents ($21M)",[118,4783,4784],{},"Customer sharing with existing cloud/SaaS portfolio companies",[118,4786,4787],{},"M&A (cloud majors internalizing capabilities)",[97,4789,4790,4794,4797,4800,4803],{},[118,4791,4792],{},[32,4793,4032],{},[118,4795,4796],{},"Vertical integration, mega-rounds. Concentrated capital in single companies; group synergy prioritized",[118,4798,4799],{},"Ampere Computing ($6.5B valuation), Graphcore acquisition",[118,4801,4802],{},"Vertical integration with SB group (Arm, T-Mobile, residual WeWork network)",[118,4804,4805],{},"Long-term hold or IPO (Arm IPO as precedent)",[353,4807,4809],{"id":4808},"the-essential-divide-ecosystem-vs-vertical-integration","The Essential Divide: Ecosystem vs. Vertical Integration",[220,4811,4814],{"className":4812,"code":4813,"language":225},[223],"Ecosystem Model (Sequoia / a16z / KP / Bessemer)\n────────────────────────────────────────────────\nVC\n ├── Startup A (Design Automation)\n ├── Startup B (Interconnect/CPO)\n ├── Startup C (Advanced Packaging)\n └── Startup D (New Computing Paradigm)\n        │\n        ▼\n Portfolio companies form mutual customer-supplier relationships\n → VC itself becomes the hub of a self-reinforcing ecosystem\n\nVertical Integration Model (SoftBank)\n────────────────────────────────────────────────\nSoftBank\n └── Ampere Computing (proprietary Arm CPU design)\n      └── IP licensed from Arm (SB subsidiary)\n           └── Direct-to-data-center sales model\n                → Full value chain captured within the group\n",[227,4815,4813],{"__ignoreMap":229},[58,4817],{},[20,4819,4821],{"id":4820},"investment-opportunities-and-risks-three-time-horizons","Investment Opportunities and Risks: Three Time Horizons",[353,4823,4825],{"id":4824},"short-term-18-months-design-efficiency-and-eda-automation","Short-Term (≤18 Months): Design Efficiency and EDA Automation",[91,4827,4828,4844],{},[94,4829,4830],{},[97,4831,4832,4835,4838,4841],{},[100,4833,4834],{},"Opportunity",[100,4836,4837],{},"Investment Theme",[100,4839,4840],{},"Representative Startups",[100,4842,4843],{},"Key Risks",[113,4845,4846,4860,4874],{},[97,4847,4848,4851,4854,4857],{},[118,4849,4850],{},"EDA Automation",[118,4852,4853],{},"LLM-powered circuit verification and bug correction reducing design time by 30–50%",[118,4855,4856],{},"Ricursive (Seq.), ChipAgents (Bessemer)",[118,4858,4859],{},"Synopsys/Cadence accelerating in-house LLM development; acquisition risk",[97,4861,4862,4865,4868,4871],{},[118,4863,4864],{},"IP Reuse/Modularization",[118,4866,4867],{},"AI-powered search and customization of existing IP assets reducing new chip design cost",[118,4869,4870],{},"Multiple undisclosed stealth-stage companies",[118,4872,4873],{},"Leading fabs (TSMC, Samsung) building proprietary IP libraries, crowding out differentiation",[97,4875,4876,4879,4882,4885],{},[118,4877,4878],{},"Supply Chain Visibility",[118,4880,4881],{},"Geopolitical risk driving demand for SaaS platforms that map procurement uncertainty",[118,4883,4884],{},"(Numerous undisclosed companies)",[118,4886,4887],{},"In a recession, software spend is the first to be cut",[28,4889,4890,4893],{},[32,4891,4892],{},"Short-term investment criteria",": Can the company acquire initial customers within 18 months? Is the model software-first and insulated from regulatory risk? EDA automation meets these conditions, but the large R&D budgets of Cadence and Synopsys demand constant vigilance.",[58,4895],{},[353,4897,4899],{"id":4898},"medium-term-24-years-data-movement-cpo-and-advanced-packaging","Medium-Term (2–4 Years): Data Movement, CPO, and Advanced Packaging",[91,4901,4902,4914],{},[94,4903,4904],{},[97,4905,4906,4908,4910,4912],{},[100,4907,4834],{},[100,4909,4837],{},[100,4911,4840],{},[100,4913,4843],{},[113,4915,4916,4930,4944,4958],{},[97,4917,4918,4921,4924,4927],{},[118,4919,4920],{},"CPO (Co-Packaged Optics)",[118,4922,4923],{},"Replacing copper GPU-to-memory interconnect with optical fiber; 10× bandwidth, 60–80% power reduction",[118,4925,4926],{},"Ayar Labs (a16z, $500M)",[118,4928,4929],{},"Production yield (optical component manufacturing complexity); TSMC/Intel CPO internalization risk",[97,4931,4932,4935,4938,4941],{},[118,4933,4934],{},"High-Speed SerDes and Electrical Interconnect",[118,4936,4937],{},"Extending copper's lifespan by improving intra-data-center wiring efficiency",[118,4939,4940],{},"Kandou Technologies (KP, $92.3M)",[118,4942,4943],{},"Legacy risk if optical migration happens faster than expected",[97,4945,4946,4949,4952,4955],{},[118,4947,4948],{},"Advanced Packaging (Chiplet/3D IC)",[118,4950,4951],{},"Heterogeneous chip integration via Chiplet standards (UCIe)",[118,4953,4954],{},"Nexthop AI (a16z, $500M)",[118,4956,4957],{},"IP ownership negotiation complexity between fabless companies and foundries",[97,4959,4960,4963,4966,4969],{},[118,4961,4962],{},"Next-Gen Memory Interfaces",[118,4964,4965],{},"Redesigning bandwidth vs. power tradeoffs as HBM3/4 successors",[118,4967,4968],{},"(Startups co-developing with SK Hynix, Micron)",[118,4970,4971],{},"DRAM vendors' growing in-house development capacity making third parties redundant",[28,4973,4974,4977],{},[32,4975,4976],{},"Medium-term investment criteria",": Does the company compete with TSMC's, Samsung's, or Intel's advanced packaging expansion plans? If so, has it established a JDA (Joint Development Agreement) partnership? This is a matter of survival. Ayar Labs' co-development structure with TSMC, backed by a16z, is the canonical example.",[58,4979],{},[353,4981,4983],{"id":4982},"long-term-5-years-new-computing-paradigms-and-quantum","Long-Term (5+ Years): New Computing Paradigms and Quantum",[91,4985,4986,4998],{},[94,4987,4988],{},[97,4989,4990,4992,4994,4996],{},[100,4991,4834],{},[100,4993,4837],{},[100,4995,4840],{},[100,4997,4843],{},[113,4999,5000,5014,5028,5042],{},[97,5001,5002,5005,5008,5011],{},[118,5003,5004],{},"Analog Computing",[118,5006,5007],{},"Eliminating digital conversion overhead; 100× inference power reduction for edge AI",[118,5009,5010],{},"Unconventional Computing (Seq., $475M)",[118,5012,5013],{},"Programmability (generality) limitations; underdeveloped developer ecosystem",[97,5015,5016,5019,5022,5025],{},[118,5017,5018],{},"Quantum Computing",[118,5020,5021],{},"Exceeding classical computation in prime factorization, quantum chemistry simulation, optimization",[118,5023,5024],{},"Quantum Circuits (Bessemer, $60M+), Rigetti ($79M)",[118,5026,5027],{},"Error correction maturity; identifying commercial use cases that achieve \"quantum advantage\"",[97,5029,5030,5033,5036,5039],{},[118,5031,5032],{},"Photonic Computing",[118,5034,5035],{},"Using light for matrix operations to dramatically reduce AI inference power consumption",[118,5037,5038],{},"(Multiple undisclosed companies)",[118,5040,5041],{},"Immature silicon photonics manufacturing infrastructure; scale-up cost",[97,5043,5044,5047,5050,5053],{},[118,5045,5046],{},"Neuromorphic Computing",[118,5048,5049],{},"Asynchronous computation mimicking neural circuits; ultra-low-power sensor processing",[118,5051,5052],{},"(Multiple academic spinouts)",[118,5054,5055],{},"Difficult to adapt for general-purpose compute; unclear path from niche markets",[28,5057,5058,5061],{},[32,5059,5060],{},"Long-term investment criteria",": Is there a clear \"paper → product → market\" connection? For quantum, the volume of papers and patents is world-class, but companies that cannot show a path to commercial error rates should be approached with caution. Analog computing, by contrast, has a well-defined target market in edge inference, making medium-to-long-term commercialization scenarios more legible.",[58,5063],{},[20,5065,5067],{"id":5066},"macro-environment-market-size-equipment-investment-and-policy","Macro Environment: Market Size, Equipment Investment, and Policy",[91,5069,5070,5092],{},[94,5071,5072],{},[97,5073,5074,5077,5080,5083,5086,5089],{},[100,5075,5076],{},"Metric",[100,5078,5079],{},"2024 Actual",[100,5081,5082],{},"2025 Forecast",[100,5084,5085],{},"2026 Forecast",[100,5087,5088],{},"2027 Forecast",[100,5090,5091],{},"Source",[113,5093,5094,5110,5126,5142,5159],{},[97,5095,5096,5099,5101,5103,5105,5107],{},[118,5097,5098],{},"Global Semiconductor Market",[118,5100,4339],{},[118,5102,137],{},[118,5104,38],{},[118,5106,4346],{},[118,5108,5109],{},"WSTS Autumn 2025",[97,5111,5112,5115,5117,5119,5121,5123],{},[118,5113,5114],{},"300mm Equipment Spending",[118,5116,4357],{},[118,5118,4360],{},[118,5120,42],{},[118,5122,46],{},[118,5124,5125],{},"SEMI 2026 Forecast",[97,5127,5128,5131,5133,5135,5137,5139],{},[118,5129,5130],{},"AI Semiconductors (GPU/NPU/ASIC)",[118,5132,4375],{},[118,5134,4378],{},[118,5136,4381],{},[118,5138,4384],{},[118,5140,5141],{},"IDC/Gartner estimates",[97,5143,5144,5147,5149,5151,5154,5156],{},[118,5145,5146],{},"U.S. CHIPS Act Grants Executed",[118,5148,4395],{},[118,5150,4398],{},[118,5152,5153],{},"$52B (cap)",[118,5155,126],{},[118,5157,5158],{},"U.S. Commerce Dept.",[97,5160,5161,5164,5167,5170,5173,5175],{},[118,5162,5163],{},"China Semiconductor Export Controls",[118,5165,5166],{},"HBM + EDA",[118,5168,5169],{},"EUV export ban continues",[118,5171,5172],{},"Further tightening expected",[118,5174,126],{},[118,5176,5177],{},"BIS/METI",[353,5179,5181],{"id":5180},"how-policy-environment-affects-investment-decisions","How Policy Environment Affects Investment Decisions",[220,5183,5186],{"className":5184,"code":5185,"language":225},[223],"CHIPS Act / EU Chips Act\n    │\n    ├── Direct beneficiaries (Intel Foundry, TSMC Arizona, Samsung Texas)\n    │        └── → Indirect demand for construction/equipment/materials startups\n    │\n    └── Elevated priority on geopolitical risk management\n             └── → Accelerated VC capital flow into U.S.-designed/manufactured startups\n\nChina Export Controls (BIS)\n    │\n    ├── Sales restrictions on AI chips to China (H800 → H20)\n    │        └── → China's domestic semiconductor self-sufficiency investment explodes (Huawei/Cambricon)\n    │\n    └── Advanced EDA/equipment export ban to China\n             └── → Reflexive benefit for Western startups (Chinese competitors' capability constrained)\n",[227,5187,5185],{"__ignoreMap":229},[58,5189],{},[20,5191,5193],{"id":5192},"portfolio-construction-framework","Portfolio Construction Framework",[28,5195,5196,5197,5200],{},"Reading across the actual portfolio allocations of Tier1 VCs reveals a consistent ",[32,5198,5199],{},"three-layer structure by time horizon",":",[220,5202,5205],{"className":5203,"code":5204,"language":225},[223],"Ideal Portfolio Allocation by Time Horizon (Illustrative)\n──────────────────────────────────────────────────────\nShort-term (≤18 months)   ████████████████░░░░  40%\n  └── EDA automation · Design SaaS · Supply chain visibility\n\nMedium-term (2–4 years)   ██████████████████░░  45%\n  └── CPO · Interconnect · Advanced packaging · AI accelerators\n\nLong-term (5+ years)      ██████░░░░░░░░░░░░░░  15%\n  └── Analog computing · Quantum · Photonics · Neuromorphic\n──────────────────────────────────────────────────────\n",[227,5206,5204],{"__ignoreMap":229},[353,5208,5210],{"id":5209},"implications-for-individual-investors","Implications for Individual Investors",[28,5212,5213],{},"Individual investors rarely have direct access to VC funds, but indirect semiconductor startup exposure is available through:",[2225,5215,5216,5222,5228,5234],{},[67,5217,5218,5221],{},[32,5219,5220],{},"Listed semiconductor companies (NVIDIA, ASML, Lam Research, TSMC ADR)",": Direct customers of startups, benefitting directly from the bottlenecks that startups are solving",[67,5223,5224,5227],{},[32,5225,5226],{},"Semiconductor ETFs (SOXX, SMH)",": Diversification benefit, but large-cap weighted — limited startup growth reflection",[67,5229,5230,5233],{},[32,5231,5232],{},"SPACs and secondary markets",": Companies like Rigetti (quantum), which went public via SPAC, offer individual investors high-risk, high-volatility access",[67,5235,5236,5239],{},[32,5237,5238],{},"Corporate parents of active CVCs",": Investing in SoftBank Group, Intel (Intel Capital parent), or Qualcomm (Qualcomm Ventures parent) provides indirect VC portfolio exposure",[353,5241,5243],{"id":5242},"implications-for-institutional-investors-and-corporates","Implications for Institutional Investors and Corporates",[64,5245,5246,5252],{},[67,5247,5248,5251],{},[32,5249,5250],{},"LPs (endowments, pension funds)",": Commitments to Sequoia / a16z / Bessemer align well with long-term semiconductor themes, but require capital that can tolerate the J-Curve (returns typically negative for 3–5 years post-commitment)",[67,5253,5254,5257],{},[32,5255,5256],{},"Corporate CVC",": Identify the bottleneck in your own value chain and strategically invest in the startup addressing it — simultaneously acquiring M&A optionality and competitive intelligence",[58,5259],{},[20,5261,5263],{"id":5262},"series-wrap-up-and-next-focal-points","Series Wrap-Up and Next Focal Points",[353,5265,5267],{"id":5266},"what-the-three-part-series-revealed","What the Three-Part Series Revealed",[91,5269,5270,5280],{},[94,5271,5272],{},[97,5273,5274,5277],{},[100,5275,5276],{},"Part",[100,5278,5279],{},"Core Message",[113,5281,5282,5292,5302],{},[97,5283,5284,5289],{},[118,5285,5286],{},[32,5287,5288],{},"Part 1: Overview and Key Players",[118,5290,5291],{},"Tier1 VCs are concentrating on the \"structural absence\" at AI infrastructure bottlenecks. The common pattern is decomposing investment across four layers — design, movement, packaging, and supply chain — rather than betting on chips themselves",[97,5293,5294,5299],{},[118,5295,5296],{},[32,5297,5298],{},"Part 2: Technology Deep Dives",[118,5300,5301],{},"Approximately $2.5B in investment was concentrated across EDA automation, CPO, advanced packaging, and new computing paradigms from 2022–2025. Technical maturity and commercial path clarity determine deal size",[97,5303,5304,5309],{},[118,5305,5306],{},[32,5307,5308],{},"Part 3: Strategic Implications (this piece)",[118,5310,5311],{},"VC styles bifurcate into ecosystem vs. vertical integration models. Short-term returns come from software profitability; medium-term from CPO/packaging; long-term from analog/quantum",[353,5313,5315],{"id":5314},"focal-points-for-h2-2026-and-beyond","Focal Points for H2 2026 and Beyond",[2225,5317,5318,5324,5330,5336,5342],{},[67,5319,5320,5323],{},[32,5321,5322],{},"Ayar Labs volume production timeline",": Will a16z's $500M CPO bet reach volume manufacturing in 2026–2027? Achieving commercial yield rates (90%+) will determine the pace of CPO adoption across the entire industry",[67,5325,5326,5329],{},[32,5327,5328],{},"Ampere Computing IPO",": If SoftBank's hinted 2026 IPO for Ampere materializes, it will provide the first publicly available data on Arm-based CPU market share in data centers",[67,5331,5332,5335],{},[32,5333,5334],{},"CHIPS Act Phase 2 grants",": U.S. Commerce Department's second-round grant decisions in H2 2026 (expanding beyond foundries to design, materials, and EDA companies) could accelerate capital flows to startups",[67,5337,5338,5341],{},[32,5339,5340],{},"China's advanced fabrication breakthrough",": Given Huawei's demonstrated 7nm-equivalent Kirin/Ascend chips, the timing of any sub-5nm breakthrough will determine both Western regulatory response speed and the persistence of the \"technology gap\"",[67,5343,5344,5347],{},[32,5345,5346],{},"Quantum error correction hitting commercial tolerances",": Whether Google (Willow), IBM, and Microsoft's logical qubit implementations can connect to commercial use cases (financial optimization, drug discovery) by 2027–2028",[58,5349],{},[855,5351,5352,5356,5363,5366,5373],{},[28,5353,5354],{},[32,5355,1659],{},[28,5357,5358,5359,5362],{},"The essence of semiconductor investment is ",[32,5360,5361],{},"infrastructure lock-in",". GPUs alone, memory alone — neither creates sustainable differentiation. But the data movement technology bridging them (CPO, SerDes), the upstream design efficiency layer (EDA automation), and the downstream assembly layer (advanced packaging) all have the kind of stickiness that resists easy substitution once adopted.",[28,5364,5365],{},"Tier1 VCs concentrating $2.5B+ in these areas is the rational outcome of a simple observation: they don't want to be dragged into chip price competition. Price wars are manufacturing's destiny, but infrastructure can achieve software-like network effects.",[28,5367,5368,5369,5372],{},"What this series reinforced for me is that ",[32,5370,5371],{},"semiconductors are inseparable from geopolitics",". The CHIPS Act, BIS export controls, Japan's economic security legislation — these policy variables influence investment decisions at least as much as pure technical assessment.",[28,5374,5375],{},"For individual investors engaging with the semiconductor sector: tracking the policy calendar alongside the technology roadmap is no longer optional. It's table stakes.",{"title":229,"searchDepth":1682,"depth":1682,"links":5377},[5378,5379,5380,5383,5388,5391,5395,5399,5400,5401,5404,5409,5412,5416],{"id":3870,"depth":1682,"text":3871},{"id":62,"depth":1682,"text":62},{"id":3927,"depth":1682,"text":3928,"children":5381},[5382],{"id":4047,"depth":1693,"text":4048},{"id":4059,"depth":1682,"text":4060,"children":5384},[5385,5386,5387],{"id":4063,"depth":1693,"text":4064},{"id":4136,"depth":1693,"text":4137},{"id":4220,"depth":1693,"text":4221},{"id":4304,"depth":1682,"text":4305,"children":5389},[5390],{"id":4429,"depth":1693,"text":4429},{"id":4440,"depth":1682,"text":4440,"children":5392},[5393,5394],{"id":4452,"depth":1693,"text":4452},{"id":4484,"depth":1693,"text":4485},{"id":4504,"depth":1682,"text":4504,"children":5396},[5397,5398],{"id":4507,"depth":1693,"text":4508},{"id":4555,"depth":1693,"text":4556},{"id":4623,"depth":1682,"text":4624},{"id":914,"depth":1682,"text":915},{"id":4692,"depth":1682,"text":4693,"children":5402},[5403],{"id":4808,"depth":1693,"text":4809},{"id":4820,"depth":1682,"text":4821,"children":5405},[5406,5407,5408],{"id":4824,"depth":1693,"text":4825},{"id":4898,"depth":1693,"text":4899},{"id":4982,"depth":1693,"text":4983},{"id":5066,"depth":1682,"text":5067,"children":5410},[5411],{"id":5180,"depth":1693,"text":5181},{"id":5192,"depth":1682,"text":5193,"children":5413},[5414,5415],{"id":5209,"depth":1693,"text":5210},{"id":5242,"depth":1693,"text":5243},{"id":5262,"depth":1682,"text":5263,"children":5417},[5418,5419],{"id":5266,"depth":1693,"text":5267},{"id":5314,"depth":1693,"text":5315},"Tier1 VCがチップ単体ではなく「設計生産性・データ移動・実装・供給網」に分解して投資する理由。短期・中期・長期の投資機会とリスクを体系化する。",{"date":1717,"image":5422,"alt":5423,"tags":5424,"tagsEn":5425,"published":1732},"/blogs-img/blog-semiconductor-vc-3.png","半導体VC投資の戦略フレームワーク",[1721,1722,1723,1725,1724],[1727,1728,1729,1731,1730],"/blogs/8-semiconductor-vc-series-3",{"title":3852,"description":5420},"blogs/8-semiconductor-vc-series-3","b5wNXd_ZiF8_NfgfhIvXkiv6Jg5X50phVnrAqqEEnWY",1775926672889]