---
格式版本: 2
标题: "Seven reasons I wouldn’t count Google out"
原文链接: "https://garymarcus.substack.com/p/seven-reasons-i-wouldnt-count-google"
发布日期: "2026-08-06"
发布时间校准状态: "found"
发布时间需复核: "否"
发布时间来源: "rule:scrape:strict_html_body"
发布时间证据: "div class=pencraft pc-reset color-pub-secondary-text-hGQ02T line-height-20-t4M0El font-meta-MWBumP size-11-NuY2Zx weight-medium-fw81nC transform-uppercase-yKDgcq reset-IxiVJZ meta-EgzBVA: Aug 06, 2026"
发布时间校准原因: "规则确认唯一严格发布时间，来源 scrape:strict_html_body"
发布时间校准置信度: "high"
发布时间候选数量: 10
发布时间严格候选数量: 1
发布时间原页读取状态: "source template page reused from URL open"
发布时间未找到原因: ""
发布时间校准时间: "2026-08-31T11:51:48+08:00"
发布时间仲裁状态: "skipped"
发布时间仲裁尝试次数: 0
发布时间仲裁耗时毫秒: 0
发现时间: "2026-08-31T11:48:46+08:00"
入库时间: "2026-08-31T03:52:26.197Z"
来源平台: "Substack 数据中心相关博客搜索"
搜索渠道: "source_template"
搜索词: "site:substack.com Google"
匹配关键词:
  - "AI"
相关厂家:
  - "Google"
  - "NVIDIA"
  - "Meta"
  - "Microsoft"
  - "OpenAI"
相关专家:
  []
内容类型: "网页"
抓取工具: "Free Fetch + Defuddle"
清洗工具: "Defuddle Markdown + Defuddle/Readability 正文提取"
原始附件:
  []
AI优质: "否"
AI打分: 17
AI分档: "非优质"
AI质检状态: "不通过"
AI打分理由: "正文主线是对Google人才流失、资金、数据和市场竞争力的个人评论，仅一句提及自研TPU及分布式系统能力，未讨论机架级架构、规格或部署。来源为个人Substack评论且发布日期缺失，多项陈述缺少原始文件支撑。历史知识库已包含Google第八代TPU及AI Hypercomputer的详细发布信息，本文没有新增产品、规格、客户、量产或路线图事实。命中弱相关和泛趋势否决，不具备高价值准入通道。"
AI质检模型: "gpt-5.6-sol"
AI质检时间: "2026-08-31T11:52:34+08:00"
AI主题相关性: 3
AI来源权威性: 4
AI新颖性: 3
AI技术细节: 1
AI商业部署信号: 0
AI完整性: 6
AI评分提示词版本: "v17-精简生产版"
AI评分提示词SHA256: "48fb9777f386026761b4873eaff30807694fb11e9b352d7c69bf2dfde750cc7d"
AI评分知识库版本: "knowledge_base_v1-20260819+runtime.61"
AI评分知识库SHA256: "ebb5f43a30c10c8490f1a5f1cc1d40cf016c39f27d463127d716cb935fe11fca"
AI评分知识库检索词: "[\"Google\",\"site:substack.com Google\",\"NVIDIA\",\"Meta\",\"Microsoft\",\"x.com/firstadopter/status/2085041030563295250\",\"TPU\",\"LLM\"]"
AI评分知识库命中: "[{\"id\":\"july-correct-0033\",\"title\":\"Microsoft, Alphabet, Meta Pivot from Buy to Build in AI\",\"sourceType\":\"labeled_article\",\"time\":\"2026-07\",\"matchedTerms\":[\"Google\",\"NVIDIA\",\"Meta\",\"Microsoft\",\"TPU\"],\"rank\":-16.628154856098057},{\"id\":\"historical-jan-apr-02\",\"title\":\"二、Google Cloud Next '26：AI Hypercomputer 与第八代 TPU 发布\",\"sourceType\":\"curated_item\",\"time\":\"2026-01_to_2026-04\",\"matchedTerms\":[\"Google\",\"NVIDIA\",\"Meta\",\"TPU\"],\"rank\":-14.839912463776773},{\"id\":\"july-correct-0104\",\"title\":\"NVIDIA Vera Rubin 提升每瓦性能，为全球合作伙伴实现最低 Token 成本\",\"sourceType\":\"labeled_article\",\"time\":\"2026-07\",\"matchedTerms\":[\"Google\",\"NVIDIA\",\"Microsoft\",\"LLM\"],\"rank\":-11.731282338966109},{\"id\":\"runtime-e8b946c3131877fe7add476b\",\"title\":\"Peeling Apart That Supposed $120 Billion Chip Deal Google Inked With Marvell\",\"sourceType\":\"ai_excellent_article\",\"time\":\"2026-08-27\",\"matchedTerms\":[\"Google\",\"Meta\",\"TPU\"],\"rank\":-10.924630043570314},{\"id\":\"july-correct-0027\",\"title\":\"Microsoft Taps AMD For At Scale AI CPU And GPU Clusters\",\"sourceType\":\"labeled_article\",\"time\":\"2026-07\",\"matchedTerms\":[\"NVIDIA\",\"Meta\",\"Microsoft\",\"LLM\"],\"rank\":-9.558580954667772}]"
AI摘要: "尽管Demis Hassabis卸任Google DeepMind负责人，Jeff Dean等顶尖人才陆续离职，作者仍认为不应看衰谷歌：谷歌拥有海量数据、自研TPU、超强分发渠道和雄厚资金，年收入4020亿美元、利润1320亿美元。"
AI摘要模型: "ali-deepseek-v4-flash"
AI摘要时间: "2026-08-31T23:37:26.566Z"
采集批次: "2026年8月31日11点48分36秒"
采集批次ID: "20260831-114836-897"
去重键: "https://garymarcus.substack.com/p/seven-reasons-i-wouldnt-count-google"
---

In huge news, in case you missed it, Sir Demis Hassabis has given up the reins at GoogleDeepMind (moving aside to be Chairman and Alphabet’s Chief Science Officer), and the rightly legendary Jeff Dean and some other top talent just left altogether, following earlier departures of star researchers like John Jumper, David Silver, Tim Rocktäschel, and Noam Shazeer. Google/Alphabet are hemorrhaging talent, morale is poor, and their latest model is delayed. [Many are ready to count them out altogether](https://x.com/firstadopter/status/2085041030563295250?s=61).

I think this is a mistake. Here are seven reasons why:

1. Google still has access to an enormous amount of data, from search and email especially, at a scale that nobody else has.
2. Only a small number of companies (Meta, Microsoft, Amazon, maybe a few in China) can match on sheer compute and expertise building distributed systems. Google/Alphabet has long had their own TPU chips, lessening their dependency on Nvidia.
3. Google still makes a *ton* of money, $402 billion in revenue last year, $132 billion in profit), and has vastly more cash to play with than OpenAI and Anthropic. (The fact that Google is going to the bond market yet again is worrisome, though.)
4. Google has great distribution, through Android, Mail, Search, Docs, YouTube etc.
5. Hassabis, arguably the most formidable person in AI, is still there, not gone altogether. He has worked together with his replacement, Koray Kavukcuoglu, for over a decade, and they will continue to work together. (I have spoken once with Koray and found him to be smart and sober, in a world where not all of their competitors are. His lack of name recognition may be a liability in the short-term, but his sobriety may be crucial in the long-run.)
6. Google’s most obvious competitors have troubles of their own. OpenAI has a declining reputation, Anthropic has conflict with the administration, Microsoft and Amazon don’t have the same talent pool.
7. Google can probably keep going even if it doesn’t take much of the LLM market in the end (which I expect will be a multi-way tie with relatively modest profits). On the other hand, if things do turn out as I expect, and there is a multi-way tie with relatively modest profits, Anthropic and OpenAI don’t have that much else to fall back on; they may never earn out their investments. The only real existential threat would be if Google loses Search, Android, and YouTube altogether. Google may simply outlast the other two.

Google definitely doesn’t look unassailable like they once did, and may never be as strong as they once were, but it’s premature to count them out.