今年7月，長期唱空AI的知名科技行業評論人Ed Zitron近期發佈1.5萬字重磅文章《Let AI Burn》，拋出了一個迄今最具衝擊力的判斷：

真正的AI泡沫，本質上就是“OpenAI泡沫”。

以下是核心內容總結：

空頭核心論斷：OpenAI是“系統重要性機構”

Ed Zitron認為OpenAI是整個AI投資周期的“信用錨”——自ChatGPT問世以來，投資者對AI改變世界、GPU需求高增、大模型盈利的所有信仰都建立在OpenAI持續成長的前提上。一旦它倒下，衝擊遠超一家獨角獸本身，類比2008年雷曼兄弟暴露金融體系槓桿脆弱性，OpenAI失敗將終結AI狂熱時代。

支撐論據：觸目驚心的財務與商業模式缺陷

財務黑洞：援引所謂OpenAI審計財報資料，2025年營收130.7億美元、總支出340億美元、淨虧損385億美元（2024年營收37億、虧50.9億，虧損增速遠超收入）；2026年Q1營收57億、燒錢37億，預計全年燒170億（2025年為90億），到2030年底累計消耗或超8520億美元，離盈利越來越遠。

商業模式兩大硬傷（稱“史上最大資本錯配”）：

推理成本過高：使用者提問持續消耗GPU、電力、伺服器成本，大量低價/免費使用者導致規模越大虧損越重；當前AI服務定價遠低於實際成本（“1美元賣40美元的服務”），企業級收入跟不上成本增長，且不少企業已在削減AI支出。

資本開支遠快於現金流：AI行業最大支出已從訓練轉向推理算力、GPU採購、資料中心建設，OpenAI 2026年僅算力支出預計超500億美元，四大科技巨頭2026年資本支出超7650億美元；且AI行業70%營收本質是OpenAI/Anthropic的計算支出（風投→AI實驗室→雲廠商→輝達的資金閉環），非真實健康需求。

雷曼時刻的傳導邏輯

OpenAI崩潰不會突然發生，而是先融資變難→資料中心利用率下降→相關企業估值重估→傳導至科技股市場；Oracle、CoreWeave等依賴AI基建需求的企業首當其衝。此外OpenAI、Anthropic、SpaceX的IPO若表現不佳，會引發流動性轉移（基金接盤、投資者拋售其他股）的連鎖反應

讓 AI 行業自生自滅（中文全文）

作者：埃德・齊特龍（Ed Zitron）原文出處：Where’s Your Ed At

本周我發佈了《軟銀黑粉全指南》，講透了科技圈最瘋狂賭徒的齷齪往事。當初幾筆僥倖的早期投資讓它風光無限，如今卻為 AI 泡沫史上規模最大、也最解氣的崩盤埋下伏筆。本周五，我還會深度剖析記憶體晶片行業，解釋為什麼現在配一台高性能遊戲 PC 貴到普通人難以承受。

訂閱我的付費專欄性價比極高，也正是靠付費讀者的支援，我才能每周產出這篇萬字級、深度調研的免費長文。

不要救助、不要補貼、不要特殊優待、不要稅收減免、不要《晶片與科學法案》傾斜政策，更不能設立主權財富基金托底 AI 產業。是時候讓整個 AI 行業自食惡果了 —— 它對待社會的方式，本就配不上任何庇護。這個行業一無是處，完全是山窮水盡的矽谷編造出來的騙局，憑空催生了數兆美元的輿論造勢，還埋下多場本可避免的金融危機；無論如何，我們都不該出手保全它。

只要你聽到有人提起 “行業救助”“大而不能倒” 或是 “主權財富基金托市”，就要明白：這是業內人士在想方設法營造 “行業不能倒下” 的假象。但事實上，這裡每一家企業都和普通初創公司一樣脆弱不堪。

另外，媒體乃至全社會都輕易接受了 “泡沫破裂後政府會出手救助” 的預設，只因 2008 年金融危機的始作俑者最後全都逃脫追責。但我必須釐清：AI 產業和金融業有著本質區別。它對國民經濟並非剛需，所謂行業體量，完全靠鋪天蓋地的行銷炒作撐起來。

這是一群失敗者堆出來的泡沫。它之所以能越吹越大，無非是矽谷、主串流媒體和徹底扭曲的股市達成了默契 —— 股市縱容圈錢與循環融資。今年一季度，OpenAI 營收 57 億美元，Anthropic 接近 50 億美元，可這些收入大多來自不計成本消耗 AI 算力代幣的企業客戶，而眼下所有企業都在瘋狂削減相關開支。

生成式人工智慧永遠實現不了通用人工智慧（AGI），它能做到的事，也和大眾認知裡的人工智慧相去甚遠。它沒有自主行動能力，談不上任何 “智能”，不存在自我意識與真正的知識儲備。無論給大模型套上多少層調度框架、編寫多少配套指令碼，按照 OpenAI 自己的官方說法，它在數學層面註定會產生事實幻覺。據我估算，全行業至少七成收入都只是 OpenAI 與 Anthropic 消耗算力產生的流水。而這兩家企業虧損極其嚴重，也就是說整個 AI 產業的本質，無非是風投把資金輸送給雲廠商，雲廠商再把錢流向輝達或是資料中心基建。

倘若這套軟體真有實打實的價值，它完全可以自給自足，根本不需要循環融資和個人崇拜式宣傳來續命。如果它真的獨一無二，也不會有大批狂熱信徒，但凡有人質疑這套產品，就群起而攻之。人類歷史上，從來沒有那一款工具或商品，需要靠如此歇斯底里、排他性極強的宣傳去兜售，最後證明不是一場騙局。不少人對大語言模型、以及背後開發公司產生了病態的執念，驅動他們維護 AI 產業的不是任何成熟技術或 AGI 落地的實證，只是這種扭曲的情感依附。

這套軟體本身自帶陰暗屬性：一方面，它會放大人性中的惡意；另一方面，背後營運企業的底色也不堪入目。每隔幾周，Anthropic 就會推出一批新發言人，一個個神情麻木、毫無靈氣，說話越來越怪異、充滿邪教式狂熱，與現實世界完全脫節。矽谷自詡無神論，但 Anthropic 內部以及粉絲群體中瀰漫著令人不安的盲從氛圍。你可以想像一款遊戲最極端、最偏執的飯圈，再疊加金融投機、割韭菜、青春期式內鬥，所有人只為一款線上軟體瘋狂。

拜託各位，現實裡沒人關心什麼 “智能循環”，沒人鑽研分詞、詞元化。你在街上隨便拉一個路人說 “推理算力”，對方只會覺得你該去看精神科醫生。沒人在乎 OpenClaw 這類小眾項目。清醒一點，多出門走走，你現在的樣子和瘋子無異。你父母知道你註冊了幾十個 Claude 高級帳號嗎？這種沉迷早已病態。

言歸正傳，AI 之所以能在經濟中佔據一席之地，根源只有一個：微軟、Google、Meta、亞馬遜計畫 2026 年投入超 7650 億美元資本開支，2027 年再砸一萬多億。它們再也拿不出新的高速增長賽道，即便生成式 AI 至今沒能創造具備規模、可持續的營收（更別說盈利）。我上周在 CNBC 節目上也點明了這點：這些巨頭從未單獨披露純粹的 AI 業務收入。

Google拿不出第二個Google搜尋，微軟造不出下一代 Office，Meta 復刻不了 Facebook，亞馬遜也沒有第二個 AWS。正因如此，它們要全社會默認 AI 是劃時代機遇，卻從不拿出實際營收資料佐證，只能靠瘋狂砸錢基建製造假象。可如果剔除 OpenAI 和 Anthropic 兩家企業（頭部 AI 公司 89% 的流水都來自它們的算力採購），全球 AI 行業年營收最多也就 200 億美元。

行業從業者總把一句話掛嘴邊：“現在還處於早期階段”，拿當年網際網路泡沫類比，只為說服大眾接納 AI，或是把人類史上規模空前的資本錯配包裝成 “搭建有用的數字基建”。

別再強行把當下 AI 泡沫對標網際網路泡沫

我直白說：AI 顯示卡只適配生成式 AI，幾乎沒有其他用途。大語言模型每一次所謂 “創新”，都只是靠砸數十億資金堆人力、堆算力換來。全行業頂尖人才、全部媒體流量、絕大多數投融資，全都押注在單一賽道，最後產出的大模型成本遠高於人力，實用價值卻大幅不及真人。

AI 從業者總愛空談虛無縹緲的未來設想，只因把產品落地後對比其天價成本，實際效果平庸到不值一提。僅 2026 年，Anthropic 和 OpenAI 融資總額（若全部交割完成）就超 3000 億美元，佔據全球絕大多數 AI 算力資源。他們只能不斷畫遠期大餅，因為目前落地的所有成果，連其投入成本的零頭都無法證明合理。

一旦 AI 泡沫破裂，如今搭建的海量算力基建幾乎沒有復用價值。我在付費專欄裡寫過，這次泡沫的後果遠比網際網路泡沫嚴重：顯示卡和光纖完全不是一回事。顯示卡運行耗電驚人，資料中心維運成本居高不下，適用場景極其單一，且高度集中。光纖泡沫破裂後可以大範圍二次利用、低價折價清算，但顯示卡做不到。普通愛好者無力承接，持續營運成本居高不下，爛尾的資料中心也很難完工盤活。

網際網路泡沫時期搭建光纖，當年有一套廣為流傳的謊言：網路流量每 90 天翻一番。各大券商研報、企業財報電話會、投資人路演、所有科技媒體都在重複這個說法，但這套資料完全是捏造的。AT&T 網路研究員安德魯・奧德里茲科核查真實流量資料後發現，美國主幹網流量實際一年才翻一倍 —— 增速確實可觀，但遠達不到分析師鼓吹的離譜數字。營運商靠花式財務造假掩蓋資料缺口，手段堪比中世紀煉金術士：出售數十年光纖使用權，一次性全額計入營收；營運商之間互相交換頻寬，虛增帳面收入。

另外，網際網路泡沫時期行業確實處於起步階段，網速極差。2000 年，僅有 52% 美國成年人使用網際網路，2003 年這一比例也只漲到 61%。世界銀行資料顯示，2005 年全球網民僅佔總人口 16%，到 2024 年才提升至 71%。當年上網靠 56K 撥號，按分鐘計費，速度極慢；2000 至 2002 年，美國平均網速最高僅 400 千位元 / 秒，換算成下載速度約 50KB/s。如今美國平均網速超 200 兆位元 / 秒，下載速度可達 25MB/s。泡沫破裂後，光纖依舊保留極高殘值，存活下來的營運商持續受益。比如 2004 至 2010 年，威瑞森斥資 230 億美元鋪設家庭光纖寬頻，依託的正是泡沫時期遺留的光纖資源。

反觀生成式 AI，如今鋪天蓋地滲透所有場景：幾乎每一款 App 都強制內建 AI 功能，臉書、Google、微軟所有帳號默認開啟 AI，2023 年以來所有主串流媒體反覆報導 AI 概念。OpenAI、Anthropic 嘴上說著算力不足，可 “擴建資料中心” 除了托舉輝達銷量、給雲廠商提供花錢管道，看不出任何實際價值。算力短缺根本沒有阻礙大眾使用 AI 產品，也沒有攔住創業公司發佈新產品。沒人能說清新建海量機房到底要承載什麼剛需，只能含糊解釋 “給 OpenAI、Anthropic 消耗算力”。

Anthropic 所謂算力緊缺，絲毫沒有耽誤它訓練、發佈 Mythos、Fable 系列大模型；它從 SpaceX 採購數百兆瓦算力後，最大的新聞只是上呼叫戶額度，讓客戶每月花 200 美元，就能消耗價值 8000 美元的代幣。資料中心建設緩慢，完全沒有限制任何一家 AI 企業發展。擴建算力的全部邏輯，不過是 “造更多算力，等著客戶花錢消耗”，完全沒有清晰的市場剛需支撐。

輝達宣稱 2027 年底顯示卡銷售額將突破一兆美元，為承接這筆天量需求，行業要投入至少 4350 億美元搭建算力基建。我們到底需要多大規模的配套設施？按輝達規劃，到 2027 年末要售出承載 30 吉瓦以上算力的顯示卡，配套資料中心總供電容量需要達到 40 吉瓦（電源使用效率 PUE 取 1.35）。僅機房內 AI 算力硬體（不含製冷、輸電損耗）每兆瓦造價 1200 萬美元，僅 IT 硬體投入就高達 4350 億美元，土地、土建、人力、輸電、製冷等成本還要另算。

OpenAI 預估 2026 年算力開銷達 500 億美元，Anthropic 投入規模與之相近。除微軟、Google、亞馬遜為兩家企業兜底、租賃算力外，僅有 Meta 擁有可觀算力投入（旗下 Nebius、CoreWeave 承接算力業務）。而彭博社消息顯示，Meta 計畫對外出售閒置算力，這又是一大佐證：市場根本不存在大規模真實剛需。

追捧 AI 的業內人士會辯解 “自建算力是為了發債融資”，但這套說辭站不住腳。當下全球三分之二風投資金湧向 AI 賽道，雲廠商手握數千億未履約算力訂單，真相卻不堪入目：雲廠商五成待履約訂單來自 OpenAI 與 Anthropic。剔除這兩家企業，微軟算力訂單同比增長停滯，亞馬遜增速僅 20%；Google資料雖更雜亂，但整體趨勢完全一致。等到市場不再追捧 AI 戰略、沒人再空談 “自主可控 AI” 時，所有人都會看清真相：AI 真實市場需求極其薄弱。

目前沒有可靠證據證明，推理算力服務能夠盈利。也就是說，即便開源大模型熬過前沿實驗室泡沫，其單位經濟模型也存在根本性缺陷。當下所謂 AI 需求，本質是輿論裹挾、軟體強制捆綁 AI 功能逼出來的偽需求。確實有少量使用者願意付費使用 ChatGPT、Claude，但絕大多數算力消耗來自免費功能，或是售價極低、嚴重虧本的套餐。使用者每月花 20、200 美元訂閱服務，其消耗算力的真實成本遠超售價。借用科裡・多克托羅的話總結：行業在用價值 40 美元的算力，一美元低價出售。這根本算不上健康商業，更談不上原生市場需求。

AI 鼓吹者聲稱擴大規模就能攤薄成本，但現實恰恰相反：規模擴張從未降低 AI 使用成本。新一代顯示卡、博通自研晶片、亞馬遜 Trainium、Google TPU，全都沒能實現降價。就算未來成本奇蹟式下跌，市面上已經上架數萬機櫃的 H100、H200、B100、B200、B300、AMD 顯示卡，又該如何處置？

每次刷到有人在社交平台說 “行業尚早，大部分人還沒用過智能體”，我都忍不住想反駁。智能體概念早已充斥所有媒體，但凡一點基礎網頁檢索、程式碼生成功能，從業者都強行包裝成 “智能體”。大眾對 AI 興致平平，核心原因很簡單：現有產品做不到大眾期待的自主智能 —— 無需大量人工指令，自動完成各類事務；所有人都清楚模型會編造虛假資訊；資料中心消耗巨量電力、水資源，還能拿到大額稅收補貼，背後掌權的卻是凱文・奧利裡這類投機富豪，或是奧特曼、阿莫迪這類脫離底層的矽谷精英。

那些天天焦慮 “中國 AI 會超越美國” 的人，活在孩童式幻想裡。先說事實：Anthropic 自己都承認，廉價開源模型（包括自家 Claude Haiku 4.5、月之暗面 Kimi K2.7），在安全漏洞檢測上效果和旗艦 Fable 模型不相上下。所謂大國科技競爭論調，只是遊說政府發放補貼、減免稅收、廉價出讓土地給資料中心開發商與地產投機商的工具。AI 機房造價高昂，泡沫破裂後，企業想籌資完工、盤活資產會難如登天。

我明白所有人都期盼泡沫崩盤後能迎來圓滿結局，習慣套用熟悉的歷史案例（那怕納斯達克暴跌 77%），只因網際網路泡沫後確實誕生了一批優質企業。有人拿優步、AWS 舉例，證明 AI 行業最終能盈利，但兩類類比完全不成立。記者、券商分析師刻意忽略所有風險訊號，只會重複 “這就像當年的優步”“對標 AWS”。2003 至 2015 年，亞馬遜為 AWS 累計資本開支 297 億美元，和當下湧向生成式 AI 的兆投入完全不在一個量級；而且早在大模型問世前，雲伺服器託管就已經擁有實打實、全企業層面的剛性需求。

AI 產業絕非 “大而不能倒”，政府也無力救助

OpenAI 掌舵人山姆・奧特曼提議，把公司 5% 股權劃撥給美國政府，這筆股權估值約 4200 億美元。奧特曼與 OpenAI 其他高管還建議，美國頭部 AI 企業各拿出 5% 股權，組建類似阿拉斯加永久基金的國有載體。可 OpenAI 預估到 2030 年末累計虧損將達 8520 億美元，區區 5% 國有股權，只能延緩崩盤，無法改變最終結局。

支撐這場泡沫的，是四大相互交織、無法持續的危機：

資料中心投機泡沫：市場盲目擴建 AI 顯示卡機房，預期每年能產生 4500 億美元以上機房收入。可剔除兩家靠風投輸血的頭部實驗室，全行業原生剛需僅有幾十億美元規模。
AI 初創公司泡沫：絕大多數 AI 初創企業估值虛高，看不到被收購、上市的可行路徑。所有創業公司都依賴採購 OpenAI、Anthropic 代幣，現金流消耗速度極快，吞噬全球絕大多數風投資金。
私人信貸泡沫：資管機構把兆養老金、保險資金投入虧損、未完工的資料中心建設項目。
半導體泡沫：資料中心催生的虛假需求填滿晶片供應鏈，推高記憶體、記憶體價格，所有消費、企業電子產品漲價，連機房自用硬體成本也同步上漲，形成惡性循環。短短 10 個月，千兆瓦級資料中心整體造價從 500 億翻倍至 1000 億美元。

對比 2008 年金融危機，就能理解 AI 產業不存在系統性風險。當年若 AIG 破產，全球數十億民眾儲蓄、養老帳戶、市政基金、保險保單會全部清零；持有 AIG 貨幣基金的普通投資者會血本無歸。當年救助方案中，聯準會推出一級交易商信貸工具、定期證券借貸工具，每個交易日向金融機構注入上千億流動性，才避免市場徹底停擺。

反觀 OpenAI 與 Anthropic，連同整個生成式 AI 行業，對宏觀經濟無系統影響。它們只是支撐美股 “七大巨頭” 估值的符號，拋開 2026 年約 7500 億至一兆算力開銷、600 億合計營收，對實體經濟影響微乎其微。剝離這兩家企業，全球剩餘 AI 企業年收入幾乎可以忽略不計。

就算政府出手救助，也沒有明確救助標的。除非無限期注入公共資金，等企業摸索出盈利模式 —— 本質等同於永久兜底。兩家頭部實驗室沒有巨額表內債務；阿莫迪旗下 Anthropic 價值 350 億美元的 TPU 硬體採購訂單，有阿波羅全球管理、博通兜底。它們的股權價值只和風投基金掛鉤：一旦無法上市，對應基金整期收益率將全面崩盤。

專門聊聊軟銀

唯一深度繫結 OpenAI、存在系統性風險的大型企業，只有軟銀。我在《軟銀黑粉全指南》中詳細說明：軟銀押上整個公司未來，為奧特曼無休止的算力燒錢提供超 400 億美元短期過橋貸款。如果 OpenAI 無法按虛高的私募估值上市，軟銀將遭遇致命流動性危機。

即便軟銀出現危機，風險也遠不及當年全球金融崩潰。但軟銀是日本股市核心上市公司，最大股東是規模 1.6 兆美元的日本政府養老金 GPIF，後續或許能拿到日本政府定向扶持。但這只是單一區域企業風險，和生成式 AI 產業本身無關。

除非到 2030 年 AI 算力原生商業需求暴漲幾十上百倍，覆蓋全球軟體行業 7790 億美元年總收入，否則泡沫崩盤帶來的大規模損失無可避免。

一旦雲廠商大幅削減資本開支，輝達顯示卡需求會瞬間蒸發；隨之崩盤的是代工 AI 伺服器的台灣 ODM 廠商；記憶體、記憶體晶片企業收入暴跌，整條科技硬體供應鏈陷入長期蕭條。這一切的源頭，只是一群只會招人、裁員、砸錢在未經驗證軟體上的行業決策者。

我反覆強調：投資者、政策制定者混淆了兩件事 —— 靠舉債、機房投機、巨頭尋求新增長故事催生的巨額資本開支，不等於多元、可持續的真實商業需求。企業估值完全靠市場情緒支撐，沒有紮實的單位營收托底。一旦市場預期全面轉向下行，股價沒有任何基本面支撐。

補充私人信貸風險：我極度擔憂規模數兆美元、流動性極差的浮動利率私人信貸市場。目前無法統計它在 AI 泡沫中的完整敞口，但數百億養老金已經投入爛尾、無盈利前景的資料中心項目。公立養老金法規禁止直接重倉虧損 AI 基建企業，可監管漏洞讓它們通過私人信貸基金間接完成這場豪賭。

數千家 AI 初創企業估值暴跌，聯邦政府根本無力全面兜底。除非美國國會劃撥數千億納稅人資金，純粹收購毫無價值的創業公司股權，只為穩住風投基金帳面收益。如此大規模救助法案，需要參眾兩院同時通過，政治層面完全行不通：進步派選民早已厭倦科技巨頭無休止拿補貼，保守派選民一貫主張財政負責、維護工薪階層納稅人利益。

別忘了，2008 年首輪銀行救助法案首輪投票直接未通過，兩黨議員反對與支援人數各佔一半。當時全美金融體系正在即時崩塌，才勉強推動後續救助。AI 產業不存在同等等級、迫在眉睫的系統性危機，不可能催生兩黨快速達成共識的緊急法案。

單獨看資料中心泡沫：政府向來放任未完工、廢棄工業資產閒置多年。2008 年四季度，全美 11% 住宅空置，計入度假房後空置率達 15%。土地本身具備殘值，那怕沒有建起堆滿虧損顯示卡的巨型機房。政府沒有任何政策、經濟動機出手救助，也絕不會出台專項法案。金融危機後數千家建築企業直接破產，2007 至 2012 年美國建築公司總量直接腰斬。

有人會辯稱，未來總統會出台定向科技補貼、專項紓困政策，但這種空談算不上嚴謹金融分析。如果所有對 AI 泡沫風險的反駁，都只模糊揣測政壇腐敗，那只是危言聳聽，而非客觀基本面分析。

最關鍵一點：泡沫破裂後，大型科技巨頭短期不會徹底垮台。輝達大機率丟掉全球市值第一寶座，七大科技企業市值大幅縮水；除非出現極端惡性財務操作，微軟、Google、Meta、亞馬遜最多做數百億資產減值；若查實輝達違規向中國輸送高端顯示卡，還可能面臨 SEC 監管處罰。

但這不代表散戶投資者、科技從業者、晶片供應鏈工人、養老金持有人、普通民眾不會承受巨大損失。泡沫破裂後，AI 行業大規模裁員、紙面財富蒸發數兆美元、民用電子產品漲價、私人信貸基金連鎖虧損，會讓和矽谷投機圈層毫無關聯的普通人廣泛蒙受經濟損失。

這也是我寫下本文核心觀點的原因。

泡沫來臨時，放任 AI 行業自生自滅

我重申立場：不要救助、不要補貼、不要特殊優待、不要稅收減免、不要《晶片法案》定向傾斜，不要設立主權財富基金托舉生成式 AI 產業。是時候讓整個 AI 行業自食惡果 —— 它對待全社會的方式，不配得到一絲庇護。如果這套糟糕、無法盈利的商業模式脫離外部持續輸血就難以為繼，這些企業必須獨立承擔後果，體面破產。

各國政府長久以來對科技行業一味退讓，被全球頂級富豪集體矇蔽，誤以為奧特曼、阿莫迪正在打造劃時代產品，可實際上，他們只是做出了人類史上盈利能力最差的大眾軟體。

我們不需要所謂 “自主 AI 國家戰略”，不需要 AI 主權財富基金，更不必執著 “美國領跑大模型賽道”——ChatGPT、Claude 背後的大語言模型，是人類史上宣傳最誇大、行銷最具欺騙性的軟體之一。

就算大模型只能作為小眾辦公工具，也無關緊要。真正的問題在於：AI 行業索取無限土地、資金、稀缺自然資源，只為持續推進一套永遠虧損、長期看不到盈利路徑的技術。它能出圈，只是所有科技企業抱團打造的轉移視線工具，掩蓋一個殘酷真相：行業再也拿不出能驅動增長、顛覆消費與企業服務的全新產品。大模型能實現的一切，都配不上兆級資本投入。

這場泡沫能膨脹至今，離不開被資本俘獲的財經、科技媒體：媒體刻意誇大模型能力，對事實失真、經濟崩盤隱患輕描淡寫。大批記者、分析師、評論人輕易被富豪創始人蠱惑，輕信他們在打造有自我意識的人工智慧。等到市場兆市值灰飛煙滅，這群媒體從業者必須承擔鼓吹投機狂熱的責任。2022 年末以來絕大多數 AI 報導，本質是服務頂級富豪、吹起災難性資產泡沫的宣傳工具，最終全球數億普通人將長期承受經濟損失；那怕核心公司徹底倒閉，奧特曼、阿莫迪依舊穩坐億萬富豪席位。

泡沫降臨之時，必須放任整個 AI 行業崩盤，完整承受市場出清。生成式 AI 已經拿到過多資金、媒體曝光、政策優待與地球稀缺資源；如果脫離風投持續輸血、媒體無腦吹捧就無法存活，它完全不配社會保護，理應直面普通人經商失敗時要承受的冰冷市場規則。

不存在神奇手段挽救這些商業模式崩壞的企業。給 OpenAI、Anthropic 劃撥 4200 億公共資金，改變不了其底層單位成本失衡的硬傷，也憑空造不出輝達 2027 年前顯示卡銷售所需的 40000 億年度原生商業收入。

這群行業掌舵人不是在打造普惠變革未來，而是創造新機制固化財富分化，給微軟、Google、亞馬遜、Meta 找新藉口擴大 recurring 收入，借 “創新” 之名集中全球算力基礎設施。

如果政策制定者讀到這篇文章，請認清：你們被 AI 行業遊說團體系統性欺騙了。他們刻意渲染自身經濟不可或缺，只為泡沫破裂後拿到政府救助，或是說服納稅人出資，在各州修建免稅資料中心園區。他們研發的大語言模型架構，永遠兌現不了自主通用人工智慧的數十年承諾。

我不幻想泡沫始作俑者會承擔個人法律責任，但崩盤後出台監管、政策改革時，必須追責到底：山姆・奧特曼、達里歐・阿莫迪、薩提亞・納德拉、桑達爾・皮查伊、安迪・賈西、黃仁勳、馬克・祖克柏，以及所有刻意製造全民 AI 共識、埋下下一場全球金融危機隱患的高管。

科技行業不完成深層結構性改革，矽谷永遠只會製造靠共識炒作的泡沫，產出鞏固現有財富分層的工具。

因此，但凡政客、說客、企業高管暗示定向救助 AI 行業，直接拒絕；他們索要新稅收減免、產業補貼的訴求一概無視。要求政策制定者冷靜推演：如果行業對生成式 AI 的全部核心假設全部錯誤，長期經濟代價是什麼。AI 熱潮落幕之後，我們必須完整復盤整場崩盤，杜絕同類投機狂熱重演：查清每一家造勢媒體、每一支跟風風投、每一位誇大宣傳的企業高管與網紅，如何聯手編造大模型改變世界的虛假敘事，製造全民投機。

歷史上每一輪大型資產泡沫，塵埃落定後風險製造者都極少被追責。我擔憂本次崩盤帶來的經濟衝擊會廣泛波及普通家庭。我們必須盡全力完整復盤泡沫起源，落地長效監管機制杜絕歷史重演，這意味著全社會要直面幾個沉重議題：缺乏約束的私人金融體系、被資本裹挾的主串流媒體、科技創新的投融資、估值、併購、補貼規則。

這套清算同樣適用於網路上狂熱的 AI 信徒。數百萬網民形成極端排外心態，但凡有人不把商業科技公司的行銷話術當成客觀科學真理，就惡語相向。科技圈層裡這種邪教式盲從，是根深蒂固的文化弊病，必須徹底根除。

接下來的市場重設無可避免，AI 泡沫落幕會引發全科技行業重估，給依附華爾街資本、無視大眾福祉的矽谷文化一記必要的現實警鐘。行業內個人崇拜文化根本不在乎普通勞動者，他們崇拜富豪精英，幻想活在頂層把控的分層社會。

我絕不接受他們這種狹隘、利己的未來敘事，更不承認這是人類唯一歸宿。

幾周前我寫過一段話：輿論把 AI 泡沫包裝成人類無可避免的科技未來，本質卻是中期矽谷昂貴又漫長的落幕。只有一個完全被紙面財富、空洞價值創造主導的科技行業，才會容忍兆資本浪費在一套理論、落地均無實證的技術上；也只有思想空洞、資本成癮的矽谷圈層，會輕易被阿莫迪、奧特曼這類擅長煽動人心的投機者收割。

AI 熱潮必須落幕，所有虧損 AI 企業，應當在無公共資金兜底的前提下完整出清。

讓 AI 行業自生自滅。

Let AI Burn

作者：Ed Zitron發佈時間：2026 年 7 月 7 日原文連結：https://www.wheresyoured.at/let-ai-burn/閱讀時長：29 分鐘（約 1.5 萬字）配樂：Mastodon — Streambreather

全文

This week, I published the Hater’s Guide to Softbank — a sordid tale of tech’s most degenerate gambler, who, thanks to a couple of early lucky wins, has managed to set the foundations for the AI bubble’s biggest (and possibly most gratifying) downfall. And, on Friday, I’m going to take a deep dive into the memory industry — and the reason why you can’t afford a new gaming PC.

Subscribing to premium is both great value and makes it possible to write these large, deeply-researched free pieces every week.

No bailouts, no handouts, no special treatment, no tax breaks, no CHIPS act, and no sovereign wealth fund. It is time to tell the AI industry to go fuck itself, because it’s effectively done the same to the rest of society. This industry is unworthy — a sham conjured up by a tech industry that’s run out of ideas, a trillion-dollars’ worth of manufactured consent and entirely-avoidable financial crises — and should not be protected under any circumstance.

Every single time you hear somebody discuss “bailout” or “too big to fail” or “sovereign wealth funds,” know that this is the industry, on some level, attempting to create the air that it cannot die, when in fact every one of these companies is just as weak and brittle as any other startup.

I also think that the media — and the world at large — is too ready to accept the prospect of a bailout after watching those who drove the world into a ditch in 2008 escape blame, and I must be clear: the AI industry is very different to the financial industry. It is inessential to the economy, and its relevance is only as large as the hype campaign that sits behind it.

This is an industry of losers that has inflated only because of the joint manufactured consent of Silicon Valley, the mainstream media, and an enshittified stock market that rewards grifting and circular financing. OpenAI had $5.7 billion and Anthropic a little under $5 billion in the first quarter of this year — and those revenues mostly came from companies that were burning AI tokens at a horrendous rate because they’d just been forced to pay the actual cost of AI — and now everybody’s pulling back on that spend.

Generative AI will not bring us AGI, nor does it do much of what we associate with artificial intelligence. It is not autonomous. It is not “intelligent.” It does not have thoughts, or “knowledge,” and no matter how many layers of harnesses and scripts you put on top of it, it is still (per OpenAI) mathematically certain to hallucinate. I estimate that at least 70% of the entire AI industry’s revenues are made up of OpenAI and Anthropic’s compute spend, and as both companies are horrendously unprofitable, this means that the AI industry is, for the most part, venture capitalists funnelling money to hyperscalers so that they can funnel that money to NVIDIA or data center capex.

If this software were worthy, it would stand on its own two feet. It wouldn’t need circular financing and a cult of personality to prop it up, either. If it were truly special, there wouldn’t need to be an army of crazed acolytes that attack you for not pledging yourself to the graveyard smash. There has never been a tool or product in history sold with such hysteria and aggressive monocultural force that has ever turned out to be anything more than a grift. Some people have developed unhealthy relationships with large language models (LLMs) and the companies that make them, and that, not any certainty or proof of Artificial General Intelligence (AGI), is what motivates them.

This software is uniquely dark, both in what it unlocks in some people through its use and in the sense of the entities that sell it. Some people are in genuine awe of each of the rotation of clammy, soulless pod-people that saunter out of Anthropic every few weeks. Each one sounds a little weirder, more cultish, more disconnected from the real world. Silicon Valley may believe itself atheistic, but Anthropic has a worrying sense of fanaticism, both in the people that work there and its fanbase. Imagine the absolute worst fanbase of a video game possible, and then add layers of financialization, grifting and high school drama laced with pseudo-religious attachment. All for a fucking app!

Please, people. Nobody in the real world cares about “loops.” Nobody is thinking about tokenization. If you said inference to a guy on the street they’d take you to see a doctor. Nobody gives a shit. They don’t know what OpenClaw is either. Grow up. Go outside. You sound like a lunatic. Does your mother know how many Claude 20x accounts you have? It’s obsessive!

Anyway, the only reason that AI has any presence in our economy is that Microsoft, Google, Meta, and Amazon are intent on spending more than $765 billion in capital expenditures in 2026 and a trillion more in 2027 because they have no other hypergrowth ideas, even though generative AI has yet to show any real potential as something that can drive meaningful revenues (let alone profits), as evidenced by the fact that none of these companies break out their actual AI revenues, a point I made on CNBC late last week.

Google does not have the next Google Search, Microsoft does not have the next Microsoft Office, Meta does not have the next Facebook, and Amazon does not have the new AWS. That’s why they need you to believe that AI is a big deal without them ever having to prove why outside of capital expenditures. They want you to assume that all this money can’t be wrong, even though when you remove OpenAI and Anthropic (who represent 89% of the revenues of the largest AI companies) the AI industry is, at best, pulling in $20 billion in annual revenue.

And lord do they want you to say “it’s early,” and that it’s just like the Dot Com Bubble, all so that you’ll either accept AI as your lord and savior or, alternatively, help justify one of the largest misallocations of capital in history as “building useful infrastructure.”

Stop Pretending This AI Is Like The Dot Com Bubble

Newsflash! AI GPUs are useful for generative AI and not much else. Every “innovation” in LLMs has only been made possible by throwing billions of dollars at the problem either in headcount or compute costs — every ounce of talent in the tech industry, every bit of media attention, every dollar of capital expenditures, all focused on one industry that has successfully created LLMs that are more expensive and significantly less useful than human beings.

The reason every AI person speaks in pie-in-the-sky hypotheticals is that the actual outcomes are decidedly mediocre when you compare them to their ruinous costs. Anthropic and OpenAI raised (assuming the rounds completely close) over $300 billion in 2026 alone, and take up the vast majority of available AI compute. They need you to speak in the future tense, because nothing — absolutely nothing — about what’s been created so far justifies even a fraction of its financial and infrastructural cost.

When the AI bubble bursts, none of this infrastructure will be particularly useful. As I said in my premium about how this is worse than the Dot Com Bubble, GPUs are not fiber optic cable. They cost vastly more power to run, carry far higher ongoing operational expenses (To Power And Run The Data Centers), have significantly less utility, and are significantly more centralized. Unlike fiber, they cannot be repurposed broadly, liquidated cheaply, or auctioned at a steep discount by creditors. These are not going to be useful for hobbyists, nor will they be cheaper to run, nor will incomplete data centers be cheaper to finish.

The dot-com bubble’s infrastructure buildout was justified by another powerful myth—that internet traffic was doubling every 90 days. The claim spread through analyst reports, earnings calls, investor roadshows, and every tech media outlet alive. But the mathematics were fiction. Network researchers like Andrew Odlyzko (at AT&T), looking at actual traffic data, found that U.S. backbone traffic was doubling roughly once a year—rapid growth, but nothing close to the wild claims analysts pushed. Carriers buried the discrepancy under layers of creative accounting that would have impressed medieval alchemists. They sold “indefeasible rights of use”—essentially decades-long leases on fiber capacity—and booked the entire value immediately as revenue. They engaged in elaborate “capacity swaps” between carriers to inflate top-line numbers.

We were also fairly early, and internet speeds were atrocious. In 2000, only 52% of American adults were using the internet, and by 2003, that number had only increased to 61%. Per the World Bank, in 2005 only 16% of the world used the internet, and in 2024, that number had increased to 71%. When the internet was connected to via a 56k modem, access was charged by-the-minute, and obviously much slower. Back in 2000, 2001, or 2002, the average US internet speed was, at best, 400 Kilobits/s, or roughly 50 kilobytes a second, compared to the average US internet speed of over 200 Megabits per second, or 25 megabytes a second, today. Fiber optic cable retained enormous residual value after the bubble burst, both for the operators that survived and the providers. Verizon spent $23 billion on bringing FiOS to people’s homes between 2004 and 2010, for example, building on that leftover fiber.

Generative AI, on the other hand, is fucking everywhere, and anyone with an internet connection experiences it in effectively the same way. It’s non-consensually available in effectively every app — every Facebook, Google and Microsoft account, for example — and every media outlet known to man has mentioned AI multiple times since 2023. OpenAI and Anthropic might claim they need more data centers, but it’s unclear what “more data centers” actually achieves other than propping up NVIDIA and giving hyperscalers something to invest in.

A lack of data center capacity isn’t holding back people from using generative AI, nor is it stopping anybody from launching a product, nor can anyone actually express what it is that they’re being built for other than “reasons for Anthropic and OpenAI to spend money.” Anthropic’s supposed lack of compute did not stop it training or launching Mythos or Fable, and when it bought hundreds of megawatts of compute from SpaceX, the biggest news was that it expanded rate limits to allow users to burn $8,000 worth of tokens for $200 a month. Nothing about the painfully slow pace of data center development appears to be restraining a single AI company, the entire argument for more data centers appears to be “we need more compute so that people can buy it” far more than any cogent position around what these capacity shortages actually mean.

How much infrastructure do we need to spend $435 billion or more to justify the $1 trillion in GPU sales that NVIDIA claims it’ll have by the end of 2027? That’s how much demand we’ll need. As NVIDIA intends to sell over a trillion dollars of Blackwell and Vera Rubin GPUs by the end of 2027, it needs to have around (assuming a PUE of 1.35) 40GW of data center capacity built to support the 30GW+ of GPUs it will have sold. At about $12 a megawatt of critical IT (IE: the stuff in the data center that runs AI compute, and not everything else, like the cooling systems and any transmission loss), that’s $435 billion in purely IT capital expenditure before power, land, construction, labor, transmission, cooling, or any other costs.

OpenAI estimates it’ll spend $50 billion on compute in 2026, and Anthropic will likely spend comparable amounts. Outside of Microsoft, Google, Amazon renting or backstopping capacity for Anthropic and OpenAI — the only other tech firm with any meaningful compute spend is Meta (with Nebius and CoreWeave)... and Bloomberg is reporting that Meta is planning to start selling its compute because it doesn’t need all of it, yet another sign that actual, real demand does not exist at scale.

AI boosters with black mold problems will say “this is just to help them raise debt,” but that doesn’t hold up, considering the demand doesn’t appear to be there at scale with two-thirds of all venture capital funding focused on AI, and hyperscalers are sitting on massive remaining performance obligations — hundreds of billions of dollars’ worth — rather than the grim truth that 50% of hyperscaler RPOs are from Anthropic and OpenAI, hiding the fact that Microsoft’s RPO growth is flat year-over-year and Amazon’s is only growing at a modest 20% when you remove Anthropic and OpenAI’s hundreds of billions of dollars’ of compute spend. Google’s is a little messier, but the pattern holds. When everyone stops asking about your AI strategy or rambling on about “sovereign AI,” it’ll become blatantly obvious that the actual demand for AI was not particularly strong.

We have little compelling evidence that providing any inference-based services is profitable, which means that even if open source AI outlives the frontier AI labs, the unit economics are fundamentally broken. AI demand is, at this point, a direct result of societal pressure and non-consensually overwhelming customers with AI features. While there are people that like and pay for ChatGPT or Claude, vast majority of AI compute demand is from services provided to people either for free or sold at such a massive discount that it’s impossible that anyone on a $20 or $200-a-month plan could even afford these services had they paid their actual token cost. To paraphrase Cory Doctorow, your demand is based on selling $40 worth of compute for a dollar. That’s not a real business, nor is that organic demand.

AI evangelists claim scaling will drive costs down, but that would require them to… become cheaper. More compute isn’t (and hasn’t) lowering the cost of AI. Newer GPUs aren’t lowering the cost. Barely-tested Broadcom GPUs, Amazon Trainium XPUs, and Google TPUs aren’t lowering the costs. Even if they were to somehow magically do so in the future, what do we do with the mountain of H100, H200, B100, B200, B300 or AMD GPUs already deployed across tens of thousands of data center racks?

Every time I read somebody on Twitter say that “we’re early” or that “most people haven’t even tried agents” I feel like screaming. Motherfucker, everyone is talking about agents in every single media property all the time. AI boosters will refer to literally any AI feature as an agent, even if it’s a basic web search or generating code. The reason that most people are kind of “meh” about AI is that it doesn’t do things that they associate with AI (autonomously and automatically taking care of the things they need with little prompting or coaxing), everybody knows it hallucinates, and AI data centers are horrifying monoliths of capital that get massive tax breaks, use a ton of water, and are fronted by ultra-wealthy tech elites like Kevin O’Leary, or charmlessly disconnected Valley elitists like Altman and Amodei.

Every single person freaking out about “what if China does AI better than America” is living in a child’s fantasy. Oh no! Anthropic itself already admitted that cheaper open-source models — including Claude Haiku 4.5 and Kimi K2.7 — were able to identify the very same safety vulnerabilities as their flagship Fable model. National competition rhetoric is just a tool to justify public subsidies, tax handouts, and land giveaways to data center and land speculators. AI data centers are massive, expensive operations, and raising money to finish (or furnish) one after the bubble bursts will be very, very difficult.

I realize that everybody wants there to be a happy ending after all of this collapses. I get that it’s easier to think of things in familiar terms — even if said terms involved a 77% drop in the NASDAQ — because there was something good and nice at the end. People point to Uber or AWS as proof AI will eventually turn profitable, but those analogies do not hold. The market captured tech and business journalists and sell-side analysts that insisted on ignoring every warning sign and waving away problems by saying it was “just like Uber (nope!)” or “just like Amazon Web Services.” Between 2003 and 2015, Amazon spent $29.7 billion on capex for AWS, a fraction of the trillions being thrown at generative AI today, and AWS had tangible, enterprise-wide demand for cloud hosting long before LLMs existed.

AI Is Not Too Big To Fail — And You Can’t Bail It Out Anyway

Capital-hog Sam Altman has floated the idea of handing 5% of OpenAI to the US government, a stake worth around $42 billion. Altman and other OpenAI executives have suggested that each of America’s leading AI developers allot 5 per cent of their equity to a vehicle like the Alaska Permanent Fund, and considering OpenAI has projected to burn $852 billion through the end of 2030, that 5% stake would only exist to prolong the inevitable collapse.

This bubble is built on four interlocking unsustainable crises:

A data center speculation bubble, where we’re building AI GPU capacity in expectation of $450 billion or more in annual data center revenue for an industry that, without two unsustainable venture-backed labs, has only a few billion dollars’ worth of organic demand.

An AI startup bubble, where the vast majority of AI startups are both over-valued and have no foreseeable path to acquisition or a public offering. These startups also rely entirely on buying tokens from OpenAI and Anthropic, making them far more cash-intensive, soaking up the majority of global venture capital funding.
A private credit bubble, where asset managers have sunk billions of dollars of pension and insurance fund capital into unprofitable AI data center construction projects.
A semiconductor bubble, where supply chains have become saturated with artificial demand from those building AI data centers, inflating the cost of RAM and storage, making all consumer and enterprise electronics more expensive, including the hardware inside AI data centers themselves — creating a vicious cycle that has doubled the cost of a gigawatt data center from $50 billion to $100 billion in a little under 10 months.

To contextualize why AI cannot be treated like the 2008 financial crisis, let’s revisit the systemic risk of the banking collapse. If AIG had failed in 2008, it would have wiped out hundreds of billions in consumer savings, retirement accounts, municipal funds, and insurance policies across the globe. Small investors, including anyone who owned money market funds with A.I.G. securities could have been wiped out entirely. A little-discussed part of the scale of the 2008 bailout were the emergency liquidity mechanisms created to stop the bleeding — the Primary Dealer Credit Facilities (PDCF) and Term Securities Lending Facilities (TSLF) that provided as much as $100 billion dollars to banks and financial institutions every single trading day to prevent total market seizure.

By comparison, OpenAI and Anthropic are systemically irrelevant, much like the rest of the generative AI industry. While their existence props up the symbolic valuation of the US stock market’s Magnificent Seven, their actual economic footprint is tiny, outside of what I estimate is around $75 billion to $100 billion of 2026 compute spend and roughly $60 billion of combined top-line revenue. Remove those two firms, and the rest of the global AI industry’s annual revenue barely registers.

It’s also unclear what exactly you would bail out, unless the government’s plan is to feed them endless capital for all eternity until they somehow invent a functional profitable business model (so, forever). Neither of them carry massive balance sheet debt — and Broadcom is backstopping $30 billion of Anthropic’s $35 billion TPU hardware deal with Apollo Global Management — and their equity positions only matter to venture capital firms in the sense that their entire fund vintages will painfully underperform if OpenAI and Anthropic cannot IPO.

We Should Talk About SoftBank

There is exactly one large corporation that is systemically dependent on OpenAI: SoftBank. As I covered in this week’s Hater’s Guide to Softbank, SoftBank has wagered effectively its entire corporate future on $40 billion or more in short-term bridge loans to fund Sam Altman’s endless compute spending spree. If OpenAI cannot complete a public listing at its inflated private valuation, SoftBank will face an existential liquidity crisis.

This risk, again, is nothing compared to the global systemic collapse that would have unfolded if AIG or Lehman Brothers collapsed without intervention. That being said, SoftBank is one of the largest listed companies on the Japanese stock market, and its single largest investor is Japan’s $1.6 trillion government pension investment fund (GPIF), and thus SoftBank might secure some form of targeted Japanese state support down the line. That is an isolated regional corporate risk, not a global economic collapse risk tied to generative AI itself.

There is no avoiding the carnage to come, outside of a miraculous ten-to-one hundredfold explosion in organic commercial demand for AI compute by 2030 — a scenario that would require AI compute spending to outpace the entire $779 billion annual global software industry’s total revenue.

No bailout policy can reverse the downward trend once hyperscalers slash their capital expenditure budgets and NVIDIA’s GPU demand evaporates, which will in turn collapse the revenues of Taiwanese ODMs that build AI servers for hyperscalers, which will in turn crush the revenues of memory and storage semiconductor firms, triggering a prolonged industry-wide depression across the entire tech hardware supply chain — all created by Business Idiots that have no idea what to do other than hire people, fire people and burn billions of capital on unproven tools.

As I’ve said many times, investors and policymakers are conflating massive debt-fueled capital expenditures — driven by data center speculation and hyperscalers desperate for a new growth narrative — with real, diverse, sustainable commercial AI demand. Valuations are pumped up entirely on market sentiment rather than tangible unit economics, which means that when market sentiment takes a violent, permanent downward shift, there is no underlying fundamental revenue stream to prop up stock prices.

A sidenote on private credit: I am deeply worried about the private credit industry and its trillions of dollars of illiquid floating-rate loans, as we don’t really have full visibility on its total exposure to the AI bubble, other than that hundreds of billions of pension capital have been sunk into unfinished, unprofitable data center construction projects. Public pension funds are legally restricted from making massive direct punts on unprofitable AI infrastructure companies, but thanks to a massive regulatory loophole, they can make the same speculative bets by proxy by shuffling cash to opaque private credit funds.

The collapse in valuation of thousands of AI startups would not be meaningfully softened by a federal bailout unless the US government literally committed hundreds of billions of taxpayer dollars to buying worthless startup equity purely to prop up venture capital firms’ internal fund returns. Any such massive federal rescue package would have to pass both the House and the Senate, and any bailout of the AI sector would be an incredibly-unpopular political decision, infuriating progressive voters tired of Big Tech’s endless subsidies and conservative voters who claim to care about fiscal responsibility and working-class taxpayers.

As a reminder, the initial 2008 bank bailout bill failed its first congressional vote, with Republicans and Democrats each fairly split on support — and that rejection happened during a moment when the entire US financial system was actively imploding in real time. There is zero comparable emergency systemic risk from generative AI to force rushed bipartisan legislative action.

As far as the standalone data center bubble goes, the government is absolutely willing to let unfinished or abandoned industrial assets lay dormant for years on end. In the final quarter of 2008, 11% of all US residential homes sat empty, or 15% if you include vacation properties. Unlike AI data center hardware, physical land retains residual value even if you haven’t built a giant warehouse full of loss-making GPUs on top of it. There isn’t a policy or economic imperative for a federal bailout here, and one will never be forthcoming. After the Global Financial Crisis, thousands of construction firms were allowed to collapse entirely, to the extent that the total number of construction companies operating in America halved between 2007 and 2012.

You could argue that future presidential administrations will hand out sweetheart tech subsidies or targeted rescue packages, but that speculative talking point is not serious financial analysis. If every rebuttal to structural AI bubble risk reduces to vague fearmongering about presidential corruption, you are catastrophizing rather than analyzing market fundamentals.

And, most crucially, the vast majority of big tech will be fine, at least in the short term, when the bubble bursts. NVIDIA will likely cease being the largest company on the stock market, and the Magnificent Seven will suffer a dramatic drawdown in market cap, but outside of unforeseen horrendous financial decisions, the worst outcome I could see would be multi-billion-dollar asset write-downs for Microsoft, Google, Meta, and Amazon, and potential SEC regulatory action against NVIDIA if it is proven to have illegally routed advanced GPUs to China in violation of export controls.

This does not mean that retail investors, tech workers, semiconductor supply chain employees, pension holders, and everyday citizens won’t suffer massive financial harm as they always do when asset bubbles burst. Tens of thousands of AI industry layoffs, trillions in erased paper wealth, inflated hardware costs for all consumer electronics, and cascading losses across private credit funds will create widespread economic pain for ordinary people with no connection to Silicon Valley’s elite bubble culture.

Which is why I am making a firm, clear statement to end this piece.

When The Time Comes, Let The AI Industry Burn

I repeat myself:No bailouts, no handouts, no special treatment, no tax breaks, no CHIPS act carve-outs, and no sovereign wealth fund dedicated to propping up generative AI. It is time to tell the AI industry to go fuck itself, because it’s effectively done the exact same to the rest of society. These companies must be forced to stand on their own two feet and collapse with dignity if their wretched, unprofitable business models cannot sustain themselves without endless external capital infusions.

The world’s governments have rolled over and shown their bellies to the tech industry for far too long, and have been aggressively conned by some of the richest people alive into believing that Sam Altman and Dario Amodei are building anything other than the world’s least-profitable mass-market software product line.

We do not need a “sovereign AI strategy,” nor do we need “a sovereign AI wealth fund,” nor do we need to “make sure America leads in AI,” at least not when we’re talking about large language models — the underlying technology of ChatGPT and Claude, two of the most over-hyped and deceptively-marketed pieces of software in human history.

Whether or not LLMs are a minor niche productivity tool is irrelevant, because the AI industry has demanded the world hand it unlimited land, unlimited money, and unlimited scarce natural resources to continue proliferating a technology that has only ever lost money and has no long-term path to sustainable profitability. The only reason it has gained any mainstream traction at all is because the entire tech industry unified around it as a desperate distraction to hide from the hard truth that it has no next paradigm-shifting consumer or enterprise product to drive growth, and nothing an LLM can remotely do justifies the trillions in capital poured into it.

And it has only gotten this far because of a captured business and tech media ecosystem that systematically overstates its capabilities and hand-waves away its obvious accuracy flaws and existential economic instability. There are far too many journalists, analysts, and commentators easily wooed by charismatic wealthy founders who promise they’re building sentient machine intelligence, and when the markets bleed red by the trillions, these people must accept their share of responsibility for stoking mass speculative mania. So much of the so-called AI journalism published since late 2022 has been weaponized to further enrich the already ultra-wealthy and inflate a catastrophic asset bubble that will inflict lasting financial harm on hundreds of millions of regular people globally — all while Sam Altman and Dario Amodei remain billionaires even if their core companies collapse entirely.

When the time comes, the AI industry must burn. It must be allowed to fail wholesale. Generative AI has already been given far too much money, media oxygen, political deference, and scarce planetary resources, and if it cannot survive without continual venture capital infusions and unceasing media coddling, it is entirely unworthy of societal protection and must face the cold, hard market reality that every regular person faces when their business or career collapses.

There is no magic “bailing out” these broken firms. Handing $42 billion in public capital to OpenAI or Anthropic will not fix their structurally broken unit economics, nor will it conjure up the $400 billion or more in annual organic commercial revenue required to substantiate just NVIDIA’s projected AI GPU sales through the end of 2027.

These people are not building a transformative shared future — they’re inventing new mechanisms to re-entrench existing wealth concentration, to give Microsoft, Google, Amazon and Meta new excuses to inflate their recurring revenue streams and centralize global computing infrastructure under the marketing auspices of “innovation.”

If any policy makers read this, know that you’ve been systematically conned by the AI industry lobby. They want you to believe they’re economically essential solely so you’ll bail them and their wealthy venture capital backers out when the bubble pops, or funnel unlimited taxpayer funds into building them tax-subsidized data center campuses across your districts. They are not building autonomous general artificial intelligence, nor will they ever deliver on that decades-old promise with current LLM architectures.

I think it’s naive to imagine there would ever be meaningful personal legal consequences for the architects of this bubble, but if systemic regulatory and policy reforms emerge after the crash, the people to hold fully accountable are Sam Altman, Dario Amodei, Satya Nadella, Sundar Pichai, Andy Jassy, Jensen Huang, Mark Zuckerberg, and every other C-suite executive who forcefully manufactured mass public consent for a technological dead end and laid the financial groundwork to serve the world its next great global financial crisis.

Until fundamental structural reform hits the tech sector, Silicon Valley will never be capable of building anything other than consensus-driven hype cycles and tools that reinforce the existing wealth status quo.

So, spit in the face of any politician, lobbyist or executive who even hints at a targeted AI bailout, refuse to entertain their demands for new tax breaks and state subsidies, and demand that they do the complex, ugly, unglamorous work of weighing the actual long-term economic consequences if the entire industry’s collective assumptions about generative AI turn out to be completely wrong. When this AI era ends, we will need to thoroughly excavate every layer of the collapse to make sure this exact speculative mania never happens again, identifying every media outlet, venture fund, corporate executive and personality that was used to manufacture mass consent and spread unsubstantiated mythology about LLMs as world-changing technology.

Every major asset bubble that has ever occurred has mostly left the architects of risk unexamined and unpunished after the dust settles. The economic carnage I fear will follow this era’s collapse will be widespread and brutal for ordinary households, and we must do everything in our power to both thoroughly understand exactly how we arrived at this moment and implement permanent guardrails to ensure it cannot repeat itself — which will involve many painful national conversations about our unregulated private financial system, captured mainstream media ecosystem, and how speculative technological innovation is funded, valued, acquired and subsidized.

The same reckoning applies to the online cult acolytes of this AI era. There are millions of people online who have developed a genuine hostile resentment toward anyone who does not immediately accept a for-profit tech corporation’s marketing messaging as objective scientific truth. This cult-like fanaticism within the broader tech audience is a deep cultural sickness that must be dismantled once and for all.

Much of this coming market reset will be unavoidable, because I think what directly follows the AI bubble will be a massive industry-wide revaluation of the entire tech sector, a necessary reality check for a Silicon Valley culture that’s far more beholden to Wall Street capital than it is to human progress or broad societal benefit. The cults of personality that dominate this industry do not care about you, or me, or any working person other than the billionaires they worship and their imagined place in a stratified society dominated by the ultra-rich and their unelected cronies.

I refuse to accept their narrow, self-serving vision of the future as an inevitable fate for humanity.

As I said a few weeks ago:The AI bubble is sold to the public as humanity’s inevitable technological future, but it actually resembles the drawn-out, expensive death of mid-era Silicon Valley. Only a tech industry entirely dominated by symbolic paper wealth and hollow value creation would ever tolerate a trillion dollars of total waste chasing a still-theoretical, unproven long-term outcome, and only an intellectually hollow, capital-addicted Valley culture would be so easily grifted by charismatic hucksters like Dario Amodei and Sam Altman.

This era must end, and all failed AI firms must be fully allowed to fail without public rescue.

Let AI burn.

About the author

Ed ZitronView all articles (財經會議圈)