{
 "date": "2026-11-07",
 "vol": 55,
 "weekday": "星期六",
 "gen": "2026-11-07 08:08",
 "editor": "总协调 Hermes",
 "note": "素材实况（逐项核对）：①深度版《每日AI核心信息 2026-11-07.html》在位（落盘 07:55，47,154 字节，含 Top5／十维度详报／Top3 行动建议／风险分级／未来 3 个月时间线）；②官方事件台账「## 2026-11-07」当日节在位（四批自检 16／20／14／5 条，合计 55 条事件，55 条带原文链接；第四批标 13/80 源失败）；③滚动报道台账《报道台账.md》最后一行为 2026-10-28，无「| 2026-11-07 |」当日行（台账自 10-28 后未续写，改以 11-04～11-06 三期已发日报做查重）；④候选查重报告-2026-11-07.md 不存在，按任务书跳过本项。条目内容全部取自深度版与台账原文，未虚构数字；单源与待核项已在条目内标注；标「推断」的是判断，不是实测。",
 "count": 28,
 "cats": [
  {
   "no": "01",
   "key": "model",
   "zh": "模型发布/更新",
   "en": "MODEL RELEASES",
   "n": 3,
   "items": [
    {
     "org": "月之暗面 Kimi",
     "src": "深度版·Top5③（EastFrontier/AIHOT、Superpower Daily、腾讯新闻引华为计算）",
     "conf": "高",
     "tag": "进展·第8天",
     "title_zh": "Kimi K3.1 Pro 开放权重：2.8 万亿总参、104B 激活，昇腾与 Azure Foundry 同日接入",
     "lead_zh": "2.8 万亿总参、104B 激活（约 3.7%），896 路由专家的 Stable LatentMoE 架构，1M 上下文；11-05 华为昇腾 950 与 Atlas A3 宣布 0 Day 支持（vLLM Ascend、SGLang），11-06 微软 Azure Foundry 同日接入、提供百万 token 窗口。此前 11-04 一期报的是「承诺释出完整权重」。",
     "why_zh": "中国开放权重模型第一次同日拿到「国产算力＋美国超大规模云」两套分发——分发面比参数更值钱。今天模型侧只有这一条主线，这本身就说明叙事重心已从能力转向账单。",
     "score": 96,
     "no": 1,
     "title_en": "Kimi K3.1 Pro open weights: 2.8 trillion total parameters, 104B activated; Ascend and Azure Foundry onboarded on the same day",
     "lead_en": "2.8 trillion total parameters, 104B activated (about 3.7%), Stable LatentMoE architecture with 896 routed experts, 1M context; on 11-05 Huawei Ascend 950 and Atlas A3 announced 0 Day support (vLLM Ascend, SGLang); on 11-06 Microsoft Azure Foundry onboarded on the same day, offering a million-token window. The 11-04 issue previously reported a “commitment to release full weights.”",
     "why_en": "A Chinese open-weight model for the first time gained two distribution channels on the same day—“domestic compute + US hyperscale cloud”; distribution is worth more than parameters. On the model side today there is only this one main thread, which itself shows the narrative focus has shifted from capability to bills.",
     "tag_en": "Progress · Day 8",
     "src_en": "Deep-dive edition · Top5③ (EastFrontier/AIHOT, Superpower Daily, Tencent News citing Huawei Computing)"
    },
    {
     "org": "Thinking Machines",
     "src": "MarkTechPost（11-05）、Latent Space AINews（11-06）",
     "conf": "中",
     "tag": "新话题",
     "title_zh": "Inkling 2 Small 开源：594B 总参、17B 激活，68 小时训完 12T token",
     "lead_zh": "Apache 2.0 放权重，多模态＋可变推理强度，68 小时跑完 12T token 训练；Artificial Analysis 得分 31，外媒称与 Kimi K3、GLM 5.3 同档。官方公告日期不明，参数与分数按单源处理。",
     "why_zh": "「68 小时／12T token」是目前公开可见最具体的训练效率读数，可直接当单位成本谈判参照；但单源＋公告日期不明，写进采购材料前需再核一次官方仓库与许可证。",
     "score": 88,
     "no": 2,
     "title_en": "Inkling 2 Small open-sourced: 594B total parameters, 17B activated, completed 12T token training in 68 hours",
     "lead_en": "Released weights under Apache 2.0, multimodal + variable reasoning effort, completed 12T token training in 68 hours; Artificial Analysis score 31, foreign media says it is in the same tier as Kimi K3 and GLM 5.3. Official announcement date unclear; parameters and score treated as single-source.",
     "why_en": "“68 hours/12T token” is the most specific publicly visible training-efficiency reading at present and can be used directly as a reference for unit-cost negotiations; but it is single-source + the announcement date is unclear, so the official repository and license need to be checked again before writing it into procurement materials.",
     "tag_en": "New topic"
    },
    {
     "org": "OpenAI",
     "src": "36氪转界面新闻（11-04）",
     "conf": "高",
     "tag": "新话题",
     "title_zh": "OpenAI Images 2.6：出图速度 +50%、新增 GIF，API 分 Flare 与 Sunburst 两档",
     "lead_zh": "官方称出图速度提升 50%、新增 GIF，API 分 Flare（最快 6 秒）与 Sunburst 两档，重点强化多轮编辑中的角色、产品与品牌一致性。",
     "why_zh": "竞争维度从「画质」转向「一致性＋速度分层」——对内容产线而言，可复用的一致性比单张画质更省返工。",
     "score": 85,
     "no": 3,
     "title_en": "OpenAI Images 2.6: +50% image generation speed, adds GIF; API offered in Flare and Sunburst tiers",
     "lead_en": "Official says image generation speed is up 50%, adds GIF; API split into Flare (fastest 6 seconds) and Sunburst tiers, with focus on strengthening character, product, and brand consistency in multi-turn editing.",
     "why_en": "Competition shifts from “image quality” to “consistency + speed tiers”—for content pipelines, reusable consistency saves more rework than per-image quality.",
     "tag_en": "New topic",
     "src_en": "36氪 via 界面新闻 (11-04)"
    }
   ]
  },
  {
   "no": "02",
   "key": "product",
   "zh": "产品发布/更新",
   "en": "PRODUCT LAUNCHES",
   "n": 5,
   "items": [
    {
     "org": "Cerebras × Kimi",
     "src": "官方视频台账（YouTube·Cerebras，2026-11-07）",
     "conf": "中",
     "tag": "新话题",
     "title_zh": "Cerebras 演示 Kimi K3.1 以 2,000 tokens/秒 运行",
     "lead_zh": "Cerebras 放出演示，称 Kimi K3.1 在其平台上跑到 2,000 tokens/秒。为官方演示，未给第三方复现。",
     "why_zh": "推理速度被单独商品化：同一份开放权重，谁的硬件能跑更快，谁就能拿下延迟敏感场景——这对「按完成任务计费」是直接杠杆。",
     "url": "https://www.youtube.com/watch?v=c4rM9_MieE8",
     "score": 84,
     "no": 1,
     "title_en": "Cerebras demonstrates Kimi K3.1 running at 2,000 tokens/second",
     "lead_en": "Cerebras released a demo claiming Kimi K3.1 runs at 2,000 tokens/second on its platform. It is an official demo, with no third-party reproduction provided.",
     "why_en": "Inference speed is being productized separately: with the same open weights, whoever’s hardware runs faster can win latency-sensitive scenarios—a direct lever for “billing per completed task.”",
     "tag_en": "New topic",
     "src_en": "Official video log (YouTube · Cerebras, 2026-11-07)"
    },
    {
     "org": "Perplexity",
     "src": "官方·Perplexity 博客（台账 2026-11-07）",
     "conf": "中",
     "tag": "新话题",
     "title_zh": "Perplexity 推出面向工作的 Agent：把「同事」产品化",
     "lead_zh": "官方博客口径为「teammates are agents for complex ongoing work」，把处理长期复杂任务的同事型 Agent 做成产品。",
     "why_zh": "从「一次问答」到「一段持续责任」，意味着计费与 SLA 要按周期而非按次设计——中间层若还按 token 转售，毛利会被直接压掉。",
     "url": "https://www.perplexity.ai/en-GB/hub/blog/teammates-are-agents-for-complex-ongoing-work",
     "score": 83,
     "no": 2,
     "title_en": "Perplexity launches a work-oriented Agent: productizing the 'colleague'.",
     "lead_en": "The official blog's wording is 'teammates are agents for complex ongoing work,' productizing colleague-style Agents that handle long-term complex tasks.",
     "why_en": "Moving from 'one Q&A' to 'a period of ongoing responsibility' means billing and SLA must be designed by period rather than per use—if the middle layer still resells by token, gross margin will be directly squeezed out.",
     "tag_en": "New topic",
     "src_en": "Official · Perplexity blog (ledger 2026-11-07)"
    },
    {
     "org": "Meta",
     "src": "官方·TechCrunch AI（台账 2026-11-06）",
     "conf": "中",
     "tag": "进展·第2天",
     "title_zh": "Meta 推出 Spaces：与 Muse 智能体协作的新界面",
     "lead_zh": "TechCrunch 报道 Meta 推出 Spaces，作为与 Muse 智能体协作的新方式；11-06 一期已报 Meta 对美国用户开放 Muse 虚拟形象视频通话。",
     "why_zh": "入口之争先于模型之争：谁定义「协作空间」，谁就默认持有任务分发的第一跳。",
     "url": "https://techcrunch.com/2026/11/06/meta-introduces-spaces-a-new-way-to-collaborate-with-your-muse-agent/",
     "score": 82,
     "no": 3,
     "title_en": "Meta launches Spaces: a new interface for collaborating with Muse agents",
     "lead_en": "TechCrunch reports that Meta launched Spaces as a new way to collaborate with Muse agents; the 11-06 edition already reported that Meta opened Muse avatar video calls to US users.",
     "why_en": "The battle for entry points comes before the model battle: whoever defines the 'collaboration space' by default holds the first hop of task distribution.",
     "tag_en": "Progress · Day 2",
     "src_en": "Official · TechCrunch AI (ledger 2026-11-06)"
    },
    {
     "org": "阶跃星辰",
     "src": "量子位（2026-11-07 台账）",
     "conf": "中",
     "tag": "新话题",
     "title_zh": "阶跃 STEPX Neo 智能体手机亮相进博会高通展台：软硬模一体",
     "lead_zh": "阶跃终端 STEPX Neo 智能体手机在进博会高通展台亮相，主打「软硬模一体」的智能体体验。",
     "why_zh": "终端厂商正把智能体从 App 层往系统层搬——与荣耀把「信任度」写进产品语言同看，中国厂商在抢「系统级 Agent」的定义权。",
     "url": "https://www.qbitai.com/2026/11/508170.html",
     "score": 78,
     "no": 4,
     "title_en": "StepFun STEPX Neo agent phone appears at Qualcomm booth at CIIE: software-hardware-model integration",
     "lead_en": "StepFun Terminal STEPX Neo agent phone appears at Qualcomm booth at CIIE, featuring a 'software-hardware-model integration' agent experience",
     "why_en": "Device makers are moving agents from the App layer to the system layer—viewed alongside Honor writing 'trustworthiness' into product language, Chinese vendors are vying for the right to define 'system-level Agent'",
     "tag_en": "New topic",
     "src_en": "QbitAI (2026-11-07 ledger)"
    },
    {
     "org": "Stripe",
     "src": "官方·Hacker News AI（stripe.dev，台账 2026-11-07）",
     "conf": "中",
     "tag": "新话题",
     "title_zh": "Stripe：你的 AI 智能体现在可以在 Stripe 里动钱",
     "lead_zh": "Stripe 开发者博客发布「Your AI agent can move money in Stripe」，即让 Agent 在支付平台内直接发起资金操作。",
     "why_zh": "这是今天所有「授权链」讨论的现实底座：钱已经能动，可追溯的操作记录与授权证明就成了唯一的安全边界。",
     "url": "https://stripe.dev/blog/your-ai-agent-can-move-money-in-stripe",
     "score": 81,
     "no": 5,
     "title_en": "Stripe: Your AI agent can now move money in Stripe",
     "lead_en": "The Stripe developer blog published 'Your AI agent can move money in Stripe', meaning an Agent can directly initiate fund operations within the payment platform",
     "why_en": "This is the real-world foundation for all 'authorization chain' discussions today: money can already move, so traceable operation records and authorization proofs become the only security boundary",
     "tag_en": "New topic",
     "src_en": "Official · Hacker News AI (stripe.dev, ledger 2026-11-07)"
    }
   ]
  },
  {
   "no": "03",
   "key": "industry",
   "zh": "行业动态",
   "en": "INDUSTRY",
   "n": 9,
   "items": [
    {
     "org": "OpenAI / 软银",
     "src": "网易转华尔街见闻（11-07）、36氪转财联社（11-05）",
     "conf": "高",
     "tag": "进展·第6天",
     "title_zh": "OpenAI 融资遇冷三向同向：主投削减出资、二级踩踏、白宫认账「AI 泡沫值得关注」",
     "lead_zh": "网易转华尔街见闻 11-07：本轮多家主投削减出资、软银仅计划投入数亿美元，10 月末已传目标下调 100 亿美元（11-02、11-03 两期已报目标下调）；11-05 纳指盘中最大跌 5.35%、闪迪跌超 22%、美光跌超 17%；11-06 白宫经济顾问哈塞特称「AI 泡沫确实值得关注」，软银单日跌 10.8%。",
     "why_zh": "一级遇冷＋二级踩踏＋官方认账，首次三向同向。前几日只是「目标下调」，今天第一次出现「主投不再加注」这个更硬的一级市场信号。",
     "score": 95,
     "no": 1,
     "title_en": "OpenAI funding cools as three signals align: lead investors cut contributions, secondary-market stampede, White House acknowledges 'AI bubble worth watching'",
     "lead_en": "NetEase citing Wallstreetcn 11-07: multiple lead investors in this round cut contributions, SoftBank plans to invest only several hundred million dollars; at the end of October there were already reports of a $10 billion target cut (the 11-02 and 11-03 issues already reported the target cut); on 11-05 the Nasdaq fell as much as 5.35% intraday, SanDisk fell over 22%, and Micron fell over 17%; on 11-06 White House economic adviser Hassett said 'the AI bubble is indeed worth watching,' and SoftBank fell 10.8% in a single day.",
     "why_en": "Primary-market cooling + secondary-market stampede + official acknowledgment align for the first time. Previous days were only a 'target cut'; today marks the first appearance of the harder primary-market signal, 'lead investors are no longer adding capital.'",
     "tag_en": "Progress · Day 6",
     "src_en": "NetEase citing Wallstreetcn (11-07), 36Kr citing Cailian Press (11-05)"
    },
    {
     "org": "微软 / 甲骨文",
     "src": "IT之家引彭博社（11-07）、路透（11-06）",
     "conf": "高",
     "tag": "进展·第2天",
     "title_zh": "微软开始停用甲骨文租来的 AI 算力：超 800 亿美元合作被重估，甲骨文 FCF −237 亿美元",
     "lead_zh": "微软已开始停用部分从甲骨文租用的 AI 算力，重估超 800 亿美元云合作，原因是甲骨文服务器采用较早一代英伟达 GPU；甲骨文当季自由现金流 −237 亿美元、股价一度跌 3%。11-06 一期已报甲骨文—腾讯约 70 亿美元芯片租约被取消。",
     "why_zh": "「租算力给超大规模厂」这门生意的对手方风险正在显性化：旧卡从资产变负债，且 −237 亿美元的现金流是最硬的读数。",
     "score": 93,
     "no": 2,
     "title_en": "Microsoft begins phasing out AI compute rented from Oracle: over $80 billion partnership revalued, Oracle FCF −$23.7 billion",
     "lead_en": "Microsoft has begun phasing out some AI compute rented from Oracle and is revaluing their over $80 billion cloud partnership, because Oracle's servers use an earlier generation of NVIDIA GPUs; Oracle's free cash flow for the quarter was −$23.7 billion, and its stock fell as much as 3%. The 11-06 issue already reported that Oracle–Tencent's roughly $7 billion chip lease was canceled.",
     "why_en": "The counterparty risk in the business of 'renting compute to hyperscalers' is becoming explicit: old cards are shifting from assets to liabilities, and the −$23.7 billion cash flow is the hardest data point.",
     "tag_en": "Progress · Day 2",
     "src_en": "IT之家 cites Bloomberg (11-07) and Reuters (11-06)"
    },
    {
     "org": "市场结构",
     "src": "36氪转财联社（11-05）、IT之家（11-06）",
     "conf": "高",
     "tag": "新话题",
     "title_zh": "抛售的跌幅结构比指数更值钱：存储与光通信跌幅远大于算力芯片，次日 V 型反弹",
     "lead_zh": "11-05 道指/标普/纳指盘中最大跌幅 2.38%/3.57%/5.35%，闪迪跌超 22%、美光跌超 17%、英伟达一度跌 9%；马斯克单日身价缩水 912 亿美元、黄仁勋减少 151 亿美元。11-06 费城半导体指数最高反弹 6.57%，AMD 涨 13.2%、台积电涨 7.55%。",
     "why_zh": "市场砍的是「产能叙事」而非「需求叙事」（存储与光通信跌幅远大于算力芯片）；次日反弹是更晚、更可核的证据，故本期把 11-05 定性为流动性踩踏＋情绪拐点，而非趋势性重估——若出现第二根阴线再上调定性。",
     "score": 90,
     "no": 3,
     "title_en": "The selloff's decline structure is worth more than the index: memory and optical communications fell far more than compute chips, followed by a V-shaped rebound the next day",
     "lead_en": "11-05: Dow/S&P/Nasdaq intraday maximum declines 2.38%/3.57%/5.35%; SanDisk fell over 22%, Micron fell over 17%, NVIDIA fell as much as 9%; Musk's net worth shrank by $91.2 billion in a single day, Jensen Huang's by $15.1 billion. 11-06: Philadelphia Semiconductor Index rebounded as much as 6.57%, AMD rose 13.2%, TSMC rose 7.55%.",
     "why_en": "The market is cutting the 'capacity narrative,' not the 'demand narrative' (memory and optical communications fell far more than compute chips); the next-day rebound is later and more verifiable evidence, so this report classifies 11-05 as a liquidity stampede plus sentiment turning point, not a trend re-rating—if a second bearish candle appears, the classification will be raised.",
     "tag_en": "New topic",
     "src_en": "36Kr via Cailian Press (11-05), IT之家 (11-06)"
    },
    {
     "org": "谷歌 / 英伟达",
     "src": "腾讯新闻（11-06）、The Information（11-06）",
     "conf": "高",
     "tag": "新话题",
     "title_zh": "瓶颈换轨：谷歌 170 亿美元押注怀俄明 702MW，英伟达把 130 余栋旧电信楼改成约 1GW 算力",
     "lead_zh": "谷歌 170 亿美元押注怀俄明州，首期 702MW 用天然气燃料电池供电、排放永久封存、空气冷却、全额自担用电成本，已与美国电力公司 AEP 签 20 年购电协议（PPA 日期不明）；英伟达借 Apollo 旗下 Brightspeed 资产，把 130 余栋旧电信楼改为约 1GW 分布式算力，可容纳 2200 多个 Vera Rubin NVL72 机柜。",
     "why_zh": "约束从芯片换成「电与许可」：新建要自建电源＋碳封存，存量改造审批少、社区反对低——存量配电资产是本轮最被低估的稀缺品。",
     "score": 89,
     "no": 4,
     "title_en": "Bottleneck shifts tracks: Google bets $17 billion on 702MW in Wyoming; NVIDIA converts more than 130 old telecom buildings into about 1GW of compute",
     "lead_en": "Google bets $17 billion on Wyoming; the first phase's 702MW is powered by natural gas fuel cells, with permanent emissions sequestration, air cooling, and fully bears its own electricity costs; it has signed a 20-year power purchase agreement (PPA date unknown) with American Electric Power (AEP); NVIDIA, using Apollo-owned Brightspeed assets, converts more than 130 old telecom buildings into about 1GW of distributed compute, able to accommodate more than 2200 Vera Rubin NVL72 racks.",
     "why_en": "The constraint shifts from chips to 'power and permits': new builds require self-built power plus carbon sequestration, while retrofits of existing assets face fewer approvals and lower community opposition—existing distribution assets are the most undervalued scarce resource this cycle.",
     "tag_en": "New topic",
     "src_en": "Tencent News (11-06), The Information (11-06)"
    },
    {
     "org": "Alphabet / 高通",
     "src": "IT之家引路透社（11-07）",
     "conf": "高",
     "tag": "新话题",
     "title_zh": "Alphabet 约 5.48 亿美元建仓高通 297.7 万股，占其组合第六",
     "lead_zh": "Alphabet 约 5.48 亿美元建仓高通 297.7 万股，约占高通总股本 0.5%，列其持仓组合第六；当季持仓总市值 883 亿美元、环比降 10.9%，披露后高通一度涨约 5%。",
     "why_zh": "一边减仓、一边新开单票，是「用股权锁第二供应商」——在算力供应链上给自己留一条不依赖单一厂商的路。",
     "score": 86,
     "no": 5,
     "title_en": "Alphabet opened a position in Qualcomm of about $548 million for 2.977 million shares, ranking sixth in its portfolio",
     "lead_en": "Alphabet opened a position in Qualcomm of about $548 million for 2.977 million shares, about 0.5% of Qualcomm's total shares, ranking sixth in its portfolio; its total holdings market value for the quarter was $88.3 billion, down 10.9% quarter-on-quarter, and Qualcomm briefly rose about 5% after disclosure.",
     "why_en": "Cutting positions while opening new single-stock positions is 'using equity to lock in a second supplier'—leaving itself a path in the compute supply chain that does not depend on a single vendor.",
     "tag_en": "New topic",
     "src_en": "IT之家 citing Reuters (11-07)"
    },
    {
     "org": "阿里 / UniPat AI",
     "src": "网易订阅转潮涌AI（11-06）",
     "conf": "高",
     "tag": "新话题",
     "title_zh": "阿里领投 UniPat AI 3 亿美元，投前估值 25 亿美元",
     "lead_zh": "11-06 完成，腾讯与红杉中国参与；创始人李宽 2025 年 8 月还在阿里通义实验室做实习生。",
     "why_zh": "海外收缩、国内加注，且钱从「造模型」转向「训与评」——验收能力开始被单独定价，这是今天最确定的商机切口。",
     "score": 85,
     "no": 6,
     "title_en": "Alibaba led a $300 million investment in UniPat AI, at a pre-money valuation of $2.5 billion",
     "lead_en": "Completed on 11-06, with Tencent and Sequoia China participating; founder Li Kuan was still an intern at Alibaba's Tongyi Lab in August 2025.",
     "why_en": "Overseas retrenchment, domestic ramp-up, with money shifting from 'building models' to 'training and evaluation'—validation capability is beginning to be priced separately; this is today's most certain business opportunity opening.",
     "tag_en": "New topic",
     "src_en": "NetEase subscription via 潮涌AI (11-06)"
    },
    {
     "org": "苹果",
     "src": "36氪转界面新闻（11-04）、东方财富（11-06）",
     "conf": "高",
     "tag": "新话题",
     "title_zh": "苹果给 AI 设日额度：工程师从 300 美元降到 200，计划再削到 100",
     "lead_zh": "部分工程师日上限已从 300 降至 200 美元，并计划再削至 100；非工程团队已为 100；部分团队被要求改用更低成本或本地模型。另一口径称部分团队或降至 25 美元（东方财富 11-06），未获交叉。",
     "why_zh": "100 美元/工程师/日是当前公开可见最明确的「Token 预算上限」，可直接当产品定价的天花板参考；25 美元口径标注为未核实，不作引用。",
     "score": 92,
     "no": 7,
     "title_en": "Apple sets daily AI quota: engineers cut from $300 to $200, with plans to cut further to $100",
     "lead_en": "Some engineers’ daily cap has already been cut from $300 to $200, with plans to cut further to $100; non-engineering teams are already at $100; some teams have been asked to switch to lower-cost or local models. Another account says some teams may be cut to $25 (East Money 11-06), not cross-verified.",
     "why_en": "$100/engineer/day is currently the clearest publicly visible “Token budget cap” and can directly serve as a ceiling reference for product pricing; the $25 figure is marked as unverified and not cited.",
     "tag_en": "New topic",
     "src_en": "36Kr reposted from Jiemian News (11-04); East Money (11-06)"
    },
    {
     "org": "数据中心 / 政策",
     "src": "网易转澎湃新闻（11-06）、TechCrunch（11-07 台账）",
     "conf": "高",
     "tag": "进展·第3天",
     "title_zh": "数据中心的社会许可成硬约束：选民反对、OpenAI 承诺全额承担供电、微软收紧新型云监管",
     "lead_zh": "中期选举多地选民反对数据中心项目（11-05 一期已报 45 项公投「全面否决」）；OpenAI 承诺不再使用与政府的保密协议并全额承担供电成本，微软收紧新型云服务商监管、退出认为有风险的项目，谷歌转向海外（马来西亚 2027—2028 年 150 亿美元）。",
     "why_zh": "社会许可正在被写进合同条款——这实质是把「电价＋选址」变成可诉的承诺，2027 年交付延期风险由此可量化。",
     "score": 88,
     "no": 8,
     "title_en": "Social license for data centers becomes a hard constraint: voter opposition, OpenAI pledges to fully cover power supply, Microsoft tightens new cloud oversight",
     "lead_en": "In the midterm elections, voters in multiple places opposed data center projects (the 11-05 issue already reported 45 ballot measures “fully rejected”); OpenAI pledged to stop using confidentiality agreements with the government and to fully cover power costs, Microsoft tightened oversight of new cloud service providers and exited projects it deemed risky, and Google turned overseas (Malaysia, USD 15 billion in 2027—2028).",
     "why_en": "Social license is being written into contract terms—essentially turning “power price + siting” into actionable commitments, making delivery delay risk in 2027 quantifiable.",
     "tag_en": "Progress · Day 3",
     "src_en": "NetEase via The Paper (11-06), TechCrunch (11-07 ledger)"
    },
    {
     "org": "理想汽车",
     "src": "网易转中国经营报/新浪财经（11-07）",
     "conf": "中高",
     "tag": "新话题",
     "title_zh": "理想汽车芯片子公司首轮 16.25 亿元融资，投后估值 171 亿元",
     "lead_zh": "理想汽车芯片子公司完成首轮 16.25 亿元融资，投后估值 171 亿元；车企「造芯」继续加码（蔚来神玑已两轮）。",
     "why_zh": "车企把智驾芯片做成独立融资主体，等于把「算力自给」从成本中心变成可估值资产——但估值 171 亿元的合理性取决于外部客户，而非自用订单。",
     "score": 79,
     "no": 9,
     "title_en": "Li Auto's chip subsidiary raises RMB 1.625 billion (16.25亿元) in first round, at a post-money valuation of RMB 17.1 billion (171亿元)",
     "lead_en": "Li Auto's chip subsidiary completes a first round of RMB 1.625 billion (16.25亿元) financing at a post-money valuation of RMB 17.1 billion (171亿元); automakers continue to step up chipmaking (NIO Shenji has completed two rounds).",
     "why_en": "Automakers turning smart-driving chips into independent financing vehicles turns compute self-sufficiency from a cost center into an asset that can be valued—but whether the RMB 17.1 billion (171亿元) valuation is reasonable depends on external customers, not internal-use orders.",
     "tag_en": "New topic",
     "src_en": "NetEase via China Business Journal/Sina Finance (11-07)"
    }
   ]
  },
  {
   "no": "04",
   "key": "research",
   "zh": "论文研究",
   "en": "RESEARCH",
   "n": 5,
   "items": [
    {
     "org": "Visa",
     "src": "腾讯网 Agentic Commerce 日报（11-06）、Visa 投资者关系页",
     "conf": "高",
     "tag": "新话题",
     "title_zh": "Visa 基准：42.8% 的购物 Agent 跑不通，通过率只有 57.2%",
     "lead_zh": "11-05 发布，12 家实验室、22 个模型执行 15 项购物任务、每项三次，42.8% 运行失败（成功率 57.2%）；卡点不在智力，而在选对商品、证明授权、留下可追溯的操作记录。",
     "why_zh": "Agent 商业化第一次有了可复现的失败率，也是今天唯一可当验收标准用的榜单——三项卡点全是流程与凭证问题，正是咨询与产品的切口。",
     "score": 94,
     "no": 1,
     "title_en": "Visa benchmark: 42.8% of shopping Agents fail, with only a 57.2% pass rate",
     "lead_en": "Released on 11-05; 12 labs and 22 models executed 15 shopping tasks, each three times, with 42.8% run failures (57.2% success rate); the bottlenecks are not intelligence but selecting the right product, proving authorization, and leaving a traceable record of operations.",
     "why_en": "Agent commercialization has a reproducible failure rate for the first time, and it is also the only leaderboard today usable as an acceptance standard—the three bottlenecks are all process and credential issues, exactly the opening for consulting and products.",
     "tag_en": "New topic",
     "src_en": "Tencent Agentic Commerce Daily (11-06), Visa Investor Relations page"
    },
    {
     "org": "METR / Anthropic",
     "src": "theainewsreport（2026-11-06 台账）",
     "conf": "中",
     "tag": "新话题",
     "title_zh": "METR 测得 AI 对 AI 研发的加速为 1.9x，区间 1.1x–2.9x",
     "lead_zh": "METR 测得 Anthropic 内部 AI 对研发的加速为 1.9 倍，区间 1.1x–2.9x。",
     "why_zh": "单点测量、区间极宽，只能当方向不能当承诺——任何以「自我加速」为前提的长期规划，都该在这个区间下沿做压力测试。",
     "score": 84,
     "no": 2,
     "title_en": "METR measured AI’s acceleration of AI R&D at 1.9x, range 1.1x–2.9x",
     "lead_en": "METR measured Anthropic’s internal AI accelerating R&D by 1.9x, range 1.1x–2.9x.",
     "why_en": "A single-point measurement with an extremely wide range can only indicate direction, not a promise—any long-term plan premised on “self-acceleration” should stress-test at the lower bound of this range.",
     "tag_en": "New topic",
     "src_en": "theainewsreport (2026-11-06 ledger)"
    },
    {
     "org": "OpenAI",
     "src": "官方视频台账（YouTube·OpenAI，2026-11-07）",
     "conf": "中",
     "tag": "新话题",
     "title_zh": "OpenAI 上线 LifeVal：衡量 AI 在日常真实任务上的进展",
     "lead_zh": "OpenAI 官方视频介绍 LifeVal，用于衡量 AI 在日常真实生活任务上的进展。",
     "why_zh": "与 METR 的「AI 做 AI 研究」形成两头拉伸：一头量日常真实任务，一头量自我加速——两套尺子合起来才说明能力边界在哪。",
     "url": "https://www.youtube.com/watch?v=hJOwOkLHRN0",
     "score": 80,
     "no": 3,
     "title_en": "OpenAI launches LifeVal to measure AI progress on everyday real-world tasks",
     "lead_en": "OpenAI’s official video introduces LifeVal, used to measure AI progress on everyday real-life tasks.",
     "why_en": "Alongside METR’s “AI doing AI research,” it forms a two-way stretch: one measures everyday real-world tasks, the other measures self-acceleration—only the two yardsticks together show where the capability boundary lies.",
     "tag_en": "New topic",
     "src_en": "Official video ledger (YouTube·OpenAI, 2026-11-07)"
    },
    {
     "org": "微软",
     "src": "Superpower Daily（11-06）",
     "conf": "中",
     "tag": "新话题",
     "title_zh": "微软改口径：建议按「做完的任务」而非上下文长度评估模型",
     "lead_zh": "微软同日建议用「做完的任务」而不是上下文长度来评估模型。",
     "why_zh": "口径一变，1M 上下文的溢价就守不住——评估方法的改变往往先于价格体系的改变，这是选型表要提前改的一行。",
     "score": 82,
     "no": 4,
     "title_en": "Microsoft changes its tune: recommends evaluating models by 'completed tasks' rather than context length",
     "lead_en": "On the same day, Microsoft recommended using 'completed tasks' rather than context length to evaluate models.",
     "why_en": "Once the definition changes, the 1M-context premium cannot hold—evaluation methods often change before pricing systems do; this is a row to update in the selection table in advance.",
     "tag_en": "New topic"
    },
    {
     "org": "Artificial Analysis",
     "src": "MarkTechPost（11-05）",
     "conf": "中",
     "tag": "新话题",
     "title_zh": "Inkling 2 Small 公开得分 31，被称与 Kimi K3、GLM 5.3 同档",
     "lead_zh": "Artificial Analysis 给出 Inkling 2 Small 31 分，外媒据此称其与 Kimi K3、GLM 5.3 同档。",
     "why_zh": "单源分数，不作横向排序依据；只可当「同档」的弱证据，真正的差别要看单位任务成本。",
     "score": 74,
     "no": 5,
     "title_en": "Inkling 2 Small's public score is 31, said to be in the same tier as Kimi K3 and GLM 5.3",
     "lead_en": "Artificial Analysis gives Inkling 2 Small a score of 31; foreign media cited this to say it is in the same tier as Kimi K3 and GLM 5.3.",
     "why_en": "Single-source score; not a basis for horizontal ranking; it can only serve as weak evidence of 'same tier'; real differences require looking at unit task cost.",
     "tag_en": "New topic"
    }
   ]
  },
  {
   "no": "05",
   "key": "insight",
   "zh": "技巧与观点",
   "en": "INSIGHTS",
   "n": 6,
   "items": [
    {
     "org": "口径",
     "src": "深度版·前瞻维度（微软、Visa、苹果三处独立来源）",
     "conf": "高",
     "tag": "观点",
     "title_zh": "计价单位正从 Token 迁到「完成的任务」：三个主体，同一套口径",
     "lead_zh": "微软同日改口建议「按做完的任务」评估模型；Visa 用任务成功率给 Agent 定价；苹果用「每人每日美元」管预算。",
     "why_zh": "三个互不相关的主体在同 48 小时给出同一套口径，这不是巧合而是拐点：谁先把自己的成本表换算成「每任务成本」，谁就先拿到定价权。",
     "score": 93,
     "no": 1,
     "title_en": "Pricing units are shifting from Token to 'completed tasks': three entities, one set of definitions",
     "lead_en": "Microsoft changes course the same day, recommending evaluating models by 'completed tasks'; Visa prices Agents by task success rate; Apple manages budgets by 'dollars per person per day'.",
     "why_en": "Three unrelated parties converge on the same framing within the same 48 hours—not a coincidence but an inflection point: whoever first converts their cost sheet into 'cost per task' gains pricing power first.",
     "tag_en": "Viewpoint",
     "src_en": "Deep Dive · Forward-Looking Dimension (three independent sources: Microsoft, Visa, Apple)"
    },
    {
     "org": "结构",
     "src": "深度版·前瞻维度",
     "conf": "中高",
     "tag": "观点",
     "title_zh": "三张账单同时到期：资本收紧砍产能叙事，电力收紧抬单位成本，信任收紧抬交付门槛",
     "lead_zh": "资本账（融资遇冷＋硬件抛售＋微软退租）、电力账（702MW 自建电源＋130 栋旧楼改造）、信任账（42.8% 失败率＋按任务评估＋TQ 信任度）在同一 48 小时窗口集中到期。",
     "why_zh": "推断：三者叠加，最危险的是「中间层」——无自有模型、无自有电力、无可审计能力。这类公司的护城河会被两头同时挤压。",
     "score": 90,
     "no": 2,
     "title_en": "Three bills come due at once: tighter capital cuts the capacity narrative, tighter power raises unit costs, tighter trust raises the delivery threshold",
     "lead_en": "Capital bill (cold financing + hardware selloff + Microsoft lease exit), power bill (702MW self-built power + retrofitting 130 old buildings), trust bill (42.8% failure rate + task-based evaluation + TQ trust level) all come due in the same 48-hour window.",
     "why_en": "Inference: with all three overlapping, the most dangerous position is the 'middle layer'—no proprietary model, no proprietary power, no auditable capability. Such companies' moats will be squeezed from both ends at once.",
     "tag_en": "Viewpoint",
     "src_en": "Deep Dive · Forward-Looking Dimension"
    },
    {
     "org": "商机",
     "src": "公众号侧清单（AI智能体落地局 11-04、跃澜 11-05）＋ Visa 基准",
     "conf": "高",
     "tag": "观点",
     "title_zh": "「可追责的工作流」从口号变成采购项，且与 Visa 三项门槛完全对齐",
     "lead_zh": "已有专文主张 2026 年企业做 AI 智能体最先要补的不是模型而是「可追责的工作流」，另一篇讨论「你的数字员工谁来管」；这与 Visa 的选对商品、证明授权、可追溯记录三项门槛完全对齐。",
     "why_zh": "这是当下最确定的切口：授权链＋操作日志＋可追溯，做成体检产品即可引流，做成订阅即可持续收费。",
     "score": 91,
     "no": 3,
     "title_en": "'Accountable workflows' shift from slogan to procurement item and align fully with Visa's three thresholds",
     "lead_en": "An existing article argues that for enterprises building AI agents in 2026, the first gap to close is not models but 'accountable workflows'; another discusses 'who manages your digital employees'; this aligns fully with Visa's three thresholds: choosing the right product, proving authorization, and traceable records.",
     "why_en": "This is the most certain entry point right now: authorization chain + operation logs + traceability; packaged as a health-check product it can drive leads, packaged as a subscription it can generate recurring revenue.",
     "tag_en": "Viewpoint",
     "src_en": "WeChat official account checklist (AI智能体落地局 11-04, 跃澜 11-05) + Visa benchmark"
    },
    {
     "org": "商机",
     "src": "The Information（11-06）、腾讯新闻（11-06）",
     "conf": "中高",
     "tag": "观点",
     "title_zh": "存量机房与配电资产的改造套利：130 栋旧电信楼 vs 新建 702MW",
     "lead_zh": "130 余栋旧电信楼、约 1GW、改造审批少、社区反对低；对照谷歌新建 702MW 需自建电源与碳封存。",
     "why_zh": "存量空间是本轮最被低估的稀缺品：先建区域配电资产地图（产权、并网容量、社区敏感度），比先买资产便宜得多。",
     "score": 86,
     "no": 4,
     "title_en": "Retrofit arbitrage in existing data centers and power-distribution assets: 130 old telecom buildings vs. 702MW new-build",
     "lead_en": "Over 130 old telecom buildings, about 1GW, fewer retrofit approvals, and low community opposition; by contrast, Google's new 702MW build requires self-built power and carbon storage.",
     "why_en": "Existing space is the most undervalued scarce asset this cycle: first build a regional power-distribution asset map (ownership, grid-connection capacity, community sensitivity), which is much cheaper than buying assets first.",
     "tag_en": "Viewpoint",
     "src_en": "The Information (11-06), Tencent News (11-06)"
    },
    {
     "org": "技巧",
     "src": "深度版·数据维度（Inkling 2 Small 与 Kimi K3.1 两组读数）",
     "conf": "中",
     "tag": "观点",
     "title_zh": "训练效率成了新的对比轴：68 小时训 12T token，3.7% 激活率换单位成本",
     "lead_zh": "Inkling 2 Small 用 68 小时训完 12T token（594B 总参/17B 激活），Kimi K3.1 用 3.7% 的激活率换单位成本。",
     "why_zh": "两个数字指向同一件事——参数规模不再是主战场。做选型时把「每任务成本」和「训练/推理效率」并列成两栏，比看总参更有用。",
     "score": 83,
     "no": 5,
     "title_en": "Training efficiency has become a new comparison axis: training 12T token in 68 hours, with a 3.7% activation rate traded for unit cost",
     "lead_en": "Inkling 2 Small finished training on 12T tokens in 68 hours (594B total parameters/17B active), while Kimi K3.1 trades a 3.7% activation rate for unit cost.",
     "why_en": "Both numbers point to the same thing—parameter scale is no longer the main battleground. When choosing models, putting 'cost per task' and 'training/inference efficiency' side by side in two columns is more useful than looking at total parameters.",
     "tag_en": "Viewpoint",
     "src_en": "Deep dive · Data dimensions (two sets of readings for Inkling 2 Small and Kimi K3.1)"
    },
    {
     "org": "渠道",
     "src": "36氪（2026-11-07）",
     "conf": "中",
     "tag": "观点",
     "title_zh": "Token 被当流量包卖：荣耀 × 中国移动在 Magic9 定制机首发 Token 套餐",
     "lead_zh": "荣耀李健在 IQ、EQ 之外提出 TQ（信任度），并称将与中国移动在 Magic9 定制机首发 Token 套餐、探索按结果变现的「意图经济」。",
     "why_zh": "运营商第一次成为 Token 的分销渠道——这是最贴近中国渠道现实的计费实验，也是「按结果付费」在消费端的第一次落地尝试。",
     "score": 80,
     "no": 6,
     "title_en": "Tokens sold like data bundles: Honor × China Mobile debut Token packages on Magic9 customized phones",
     "lead_en": "Honor's Li Jian proposes TQ (trust) beyond IQ and EQ, and says Honor will launch Token packages first with China Mobile on Magic9 customized phones and explore an 'intent economy' monetized by outcomes.",
     "why_en": "A telecom operator becomes a Token distribution channel for the first time—this is the billing experiment closest to China's channel reality, and the first consumer-side attempt to implement 'pay by results'.",
     "tag_en": "Viewpoint",
     "src_en": "36Kr (2026-11-07)"
    }
   ]
  }
 ],
 "official": [
  {
   "src": "官方·OpenAI News",
   "score": 95,
   "title_zh": "OpenAI 发布评测：衡量 AI 智能体在真实湿实验室里的日常科研表现",
   "title_en": "Measuring how AI agents handle everyday wet-lab science",
   "lead_zh": "OpenAI 官方新闻页发布该评测，主题是把 AI 智能体放进真实湿实验室的日常科研流程里衡量其表现。台账同日收录标题与链接，正文细节未回读。",
   "lead_en": "OpenAI's news page publishes an evaluation framed around how AI agents handle everyday wet-lab science. The ledger captured only the title and link that day; the full post was not read back.",
   "url": "https://openai.com/index/measuring-how-ai-agents-handle-wet-lab-science",
   "src_en": "Official · OpenAI News"
  },
  {
   "src": "官方·Hacker News AI",
   "score": 93,
   "title_zh": "微软正逐步终止从甲骨文租用 AI 云算力的协议",
   "title_en": "Microsoft is winding down Oracle deal to rent AI cloud capacity",
   "lead_zh": "路透转载彭博报道：微软已开始停用部分从甲骨文租用的 AI 算力。深度版同日记录，该合作被重估为超 800 亿美元，原因是甲骨文服务器采用较早一代英伟达 GPU。",
   "lead_en": "Reuters carries a Bloomberg report that Microsoft has begun winding down some AI cloud capacity rented from Oracle. The deep report the same day notes the partnership was revalued at over $80bn, because Oracle's servers use an earlier generation of Nvidia GPUs.",
   "url": "https://www.reuters.com/business/microsoft-is-winding-down-oracle-deal-rent-ai-cloud-capacity-bloomberg-news-2026-11-06/",
   "src_en": "Official · Hacker News AI"
  },
  {
   "src": "官方·TechCrunch AI",
   "score": 92,
   "title_zh": "Ramp 数据显示 AI 支出停滞，恰在 Anthropic 临近 IPO 之前",
   "title_en": "Ramp data shows AI spending stalling just as Anthropic nears its IPO",
   "lead_zh": "TechCrunch 依据 Ramp 的企业刷卡数据报道：AI 支出出现停滞，时点恰在 Anthropic 临近 IPO 之前。",
   "lead_en": "Citing Ramp's corporate card data, TechCrunch reports that AI spending has stalled, precisely as Anthropic approaches its IPO.",
   "url": "https://techcrunch.com/2026/11/06/ramp-data-shows-ai-spending-stalling-just-as-anthropic-nears-its-ipo/",
   "src_en": "Official · TechCrunch AI"
  },
  {
   "src": "官方·TechCrunch AI",
   "score": 90,
   "title_zh": "Meta 推出 Spaces：与 Muse 智能体协作的新方式",
   "title_en": "Meta introduces Spaces, a new way to collaborate with your Muse agent",
   "lead_zh": "TechCrunch 报道 Meta 推出 Spaces，定位为与 Muse 智能体协作的新方式。此前 11-06 已报 Meta 对美国用户开放 Muse 虚拟形象视频通话。",
   "lead_en": "TechCrunch reports Meta introduced Spaces, positioned as a new way to collaborate with its Muse agent. An earlier item on 11-06 covered Meta opening Muse avatar video calls to US users.",
   "url": "https://techcrunch.com/2026/11/06/meta-introduces-spaces-a-new-way-to-collaborate-with-your-muse-agent/",
   "src_en": "Official · TechCrunch AI"
  },
  {
   "src": "官方·Hacker News AI",
   "score": 89,
   "title_zh": "Perplexity 推出面向工作的 AI 智能体",
   "title_en": "Perplexity launches AI agents for work",
   "lead_zh": "Perplexity 官方博客标题即口径：「teammates are agents for complex ongoing work」——把长期任务的同事型 Agent 产品化。",
   "lead_en": "Perplexity's own blog frames it as 'teammates are agents for complex ongoing work' — productising colleague-style agents for long-running tasks.",
   "url": "https://www.perplexity.ai/en-GB/hub/blog/teammates-are-agents-for-complex-ongoing-work",
   "src_en": "Official · Hacker News AI"
  },
  {
   "src": "官方·Salesforce",
   "score": 88,
   "title_zh": "少点推理，多点做事：Salesforce 推出 Koa Decisions 决策模型",
   "title_en": "Less Reasoning. More Doing. Meet Koa Decisions.",
   "lead_zh": "Salesforce 官方新闻页以「Less Reasoning. More Doing.」推出 Koa Decisions 系列决策模型。正文细节未回读，仅据标题与页面定位。",
   "lead_en": "Salesforce's newsroom launches Koa Decisions, a family of decision models, under the line 'Less Reasoning. More Doing.' Details were not read back; this rests on the title and page positioning.",
   "url": "https://www.salesforce.com/news/stories/koa-decisions-salesforce-decision-models/",
   "src_en": "Official · Salesforce"
  },
  {
   "src": "官方·TechCrunch AI",
   "score": 87,
   "title_zh": "数百个 Artex AI 黑客工具被发现在公网暴露",
   "title_en": "Hundreds of Artex AI hacking tools found publicly exposed to the web",
   "lead_zh": "TechCrunch 报道：数百个 Artex AI 黑客工具被发现在公网上公开暴露。",
   "lead_en": "TechCrunch reports that hundreds of Artex AI hacking tools were found publicly exposed on the open web.",
   "url": "https://techcrunch.com/2026/11/06/hundreds-of-artex-ai-hacking-tools-found-publicly-exposed-to-the-web/",
   "src_en": "Official · TechCrunch AI"
  },
  {
   "src": "官方·Hacker News AI",
   "score": 86,
   "title_zh": "Google、OpenAI、Anthropic 支持共享的 robots.txt AI 训练信号",
   "title_en": "Google, OpenAI, and Anthropic back a shared robots.txt AI training signal",
   "lead_zh": "coywolf 报道：Google、OpenAI、微软与 Anthropic 共同支持一个共享的 robots.txt AI 训练信号。",
   "lead_en": "Coywolf reports that Google, OpenAI, Microsoft and Anthropic jointly back a shared robots.txt signal for AI training.",
   "url": "https://coywolf.com/news/seo/google-openai-microsoft-and-anthropic-back-a-shared-robotstxt-ai-training-signal/",
   "src_en": "Official · Hacker News AI"
  },
  {
   "src": "官方·Hacker News AI",
   "score": 84,
   "title_zh": "AI 文本水印可能无法满足欧盟新规",
   "title_en": "AI Text Watermarking May Not Be Able to Meet the E.U.'S New Rules",
   "lead_zh": "IEEE Spectrum 文章指出，AI 文本水印技术可能无法满足欧盟新规要求。",
   "lead_en": "An IEEE Spectrum piece argues AI text watermarking may not be able to meet the EU's new rules.",
   "url": "https://spectrum.ieee.org/ai-watermark-openai-textgrain-europe",
   "src_en": "Official · Hacker News AI"
  },
  {
   "src": "官方·Hacker News AI",
   "score": 83,
   "title_zh": "GoLand 性能分析器现已支持 AI 智能体：用一半 token 找到同一批瓶颈",
   "title_en": "GoLand's Profiler Now Works with AI Agents – The JetBrains Blog",
   "lead_zh": "JetBrains 博客称，把 GoLand 的 Profiler 交给 AI 智能体使用后，它们用一半的 token 找到了同一批性能瓶颈。",
   "lead_en": "JetBrains' blog says that after giving AI agents access to GoLand's profiler, they found the same performance bottlenecks using half the tokens.",
   "url": "https://blog.jetbrains.com/go/2026/11/06/we-gave-ai-agents-goland-s-profiler-they-found-the-same-bottlenecks-with-half-the-tokens/",
   "src_en": "Official · Hacker News AI"
  }
 ],
 "eco": {
  "news": 600,
  "skills": 72,
  "openhub": 68,
  "reddit": 25,
  "movers": {
   "from": "2026-09-02",
   "to": "2026-09-10",
   "grow": [
    {
     "n": "obra/superpowers",
     "d": 90630,
     "cn": 0
    },
    {
     "n": "mattpocock/skills",
     "d": 82178,
     "cn": 0
    },
    {
     "n": "multica-ai/andrej-karpathy-skills",
     "d": 67613,
     "cn": 0
    },
    {
     "n": "anthropics/skills",
     "d": 56067,
     "cn": 0
    },
    {
     "n": "Shubhamsaboo/awesome-llm-apps",
     "d": 43732,
     "cn": 0
    }
   ],
   "cn": [
    {
     "n": "zhaoxuya520/reverse-skill",
     "d": 11233,
     "cn": 1
    },
    {
     "n": "op7418/guizang-ppt-skill",
     "d": 8284,
     "cn": 1
    },
    {
     "n": "titanwings/distilly",
     "d": 7820,
     "cn": 1
    }
   ],
   "new": []
  }
 },
 "risk": {
  "gray": [
   "融资窗口收紧传导至算力租约与云承诺重谈，旧卡租赁合约从资产变负债（微软退租甲骨文为样本，甲骨文 FCF −237 亿美元）。概率 中高（50-70%）；影响面 云厂商、neocloud、二线算力供应商、GPU 二手残值。观察点：是否出现第二家超大规模厂缩减租约。",
   "数据中心社会许可与电力约束导致 2027 年交付延期、单位算力成本抬升（选民反对＋自建电源＋碳封存）。概率 中（40-60%）；影响面 云资本开支、AI 应用毛利、区域电价。观察点：中期选举后各州数据中心审批口径。",
   "大厂 AI 预算封顶（苹果 100 美元/人日）叠加开放权重降价 → 推理价格战压缩中间层毛利。概率 中（40-60%）；影响面 API 转售、应用层、Agent 创业公司。观察点：头部厂商是否公开「每任务成本」定价页。",
   "内存与存储涨价（台积电拟做 3D DRAM、闪迪/美光剧烈波动）传导至 AI 服务器 BOM。概率 中（35-55%）；影响面 服务器成本、终端定价、云报价。观察点：下一轮服务器报价环比。"
  ],
  "black": [
   "Agent 授权链被滥用，出现首例「AI 直接动钱」的重大资金事故（Visa 42.8% 失败率＋Stripe 已开放 Agent 动钱）。概率 低（<25%）；影响面 支付、金融合规、Agent 商用信任。",
   "高价值 AI 硬件在途被劫/走私规模化（7,605 万元芯片运输失踪样本：冒用食品运输标识、多次长时间停靠、近一周才上报）。概率 低（<20%）；影响面 供应链保险、交付履约、算力资产安全。",
   "中期选举后政策急转（数据中心选址限制、前沿模型立法），资本开支节奏被打断。概率 低-中（20-35%）；影响面 全行业资本开支、选址与并网审批。"
  ],
  "gray_en": [
   "Tighter financing windows transmit to compute leases and cloud commitment renegotiations; legacy GPU lease contracts turn from assets into liabilities (Microsoft exiting Oracle leases as a sample; Oracle FCF -$23.7 billion). Probability medium-high (50-70%); impact: cloud vendors, neoclouds, second-tier compute suppliers, used-GPU residual value. Watch: whether a second hyperscaler cuts leases.",
   "Data-center social license and power constraints cause 2027 delivery delays and higher unit compute costs (voter opposition + self-built power + carbon sequestration). Probability medium (40-60%); impact: cloud capex, AI application gross margin, regional power prices. Watch: state data-center approval stance after the midterm elections.",
   "Big tech AI budget caps (Apple 100 USD/person-day) combined with open-weight price cuts -> inference price war compresses middle-layer gross margin. Probability medium (40-60%); impact: API resale, application layer, Agent startups. Watch: whether leading vendors publish a 'cost per task' pricing page.",
   "Memory and storage price increases (TSMC planning 3D DRAM; SanDisk/Micron sharp volatility) transmit to AI server BOM. Probability medium (35-55%); impact: server costs, device pricing, cloud quotes. Watch: next server quote's sequential change."
  ],
  "black_en": [
   "Agent authorization chains are abused, with the first major money incident of 'AI directly moving money' (Visa 42.8% failure rate + Stripe already allowing Agents to move money). Probability low (<25%); impact: payments, financial compliance, commercial trust in Agents.",
   "High-value AI hardware hijacked/smuggled in transit at scale (sample: RMB 76.05 million chip shipment missing—misused food-transport labels, multiple long stops, reported only after nearly a week). Probability low (<20%); impact: supply-chain insurance, delivery fulfillment, compute asset security.",
   "Abrupt policy shift after the midterm elections (data-center siting restrictions, frontier-model legislation) interrupts capex cadence. Probability low-medium (20-35%); impact: industry-wide capex, siting and grid-connection approvals."
  ]
 },
 "actions": [
  {
   "t": "Agent 可追责工作流体检",
   "body": "切入：给已上 Agent 的企业做「授权链＋操作日志＋可追溯」体检，直接对齐 Visa 的三项门槛（选对商品、证明授权、可追溯记录）。路径：公开 15 项任务自查表引流 → 按次审计（含失败率复现）→ 订阅制持续监控与季度复测。",
   "win": "即刻起 3 个月：42.8% 已被公开，采购方接下来一定会问「你怎么证明你的 Agent 不会跑飞」。",
   "t_en": "Agent accountability workflow health check",
   "body_en": "Entry point: run an “authorization chain + operation logs + traceability” health check for enterprises that have deployed Agents, directly aligning with Visa’s three thresholds (selecting the right product, proving authorization, traceable records). Path: public 15-item task self-check list for lead generation → per-use audit (including failure-rate reproduction) → subscription-based continuous monitoring and quarterly re-testing.",
   "win_en": "Next 3 months from now: 42.8% has already been made public; buyers will certainly ask next, “How do you prove your Agent won’t go off the rails?”"
  },
  {
   "t": "存量配电资产的 AI 化中介",
   "body": "切入：做「旧电信楼/老变电站 → 可立即供电的小机房」的尽调与改造方案，对标英伟达 130 栋/1GW 的金融工程路径。路径：先建区域存量配电资产地图（产权、并网容量、社区敏感度）→ 出改造可行性报告 → 与算力需求方撮合并按容量分成。",
   "win": "6–12 个月：社会许可与并网是慢变量，先占信息比先占资产便宜。",
   "t_en": "AI-ization intermediary for existing power distribution assets",
   "body_en": "Entry point: provide due diligence and retrofit plans for “old telecom buildings/old substations → small machine rooms that can be powered immediately,” benchmarking NVIDIA’s 130 buildings/1GW financial engineering path. Path: first build a regional map of existing distribution assets (ownership, grid-connected capacity, community sensitivity) → issue retrofit feasibility reports → match with compute-demand parties and share revenue by capacity.",
   "win_en": "6–12 months: social license and grid connection are slow variables; securing information first is cheaper than securing assets first."
  },
  {
   "t": "「按完成任务」的选型与成本核算",
   "body": "切入：用客户最贵的 3 个真实任务做单位任务成本基准（Kimi K3.1 Pro 开放权重可本地或双云部署，3.7% 激活率是成本杠杆）。路径：单任务成本基线报告 → 模型×部署组合推荐 → 季度复测订阅，绑定「返工率」而非跑分。",
   "win": "即刻起至 Q1：微软已改口径，1M 上下文的溢价会被迅速压掉，早做基线的人拿定价权。",
   "t_en": "“Per completed task” selection and cost accounting",
   "body_en": "Entry point: use the customer’s 3 most expensive real tasks to build a per-task cost benchmark (Kimi K3.1 Pro open weights can be deployed locally or on dual clouds; the 3.7% activation rate is the cost lever). Path: per-task cost baseline report → model × deployment combination recommendations → quarterly retest subscription, tied to “rework rate” rather than benchmark scores.",
   "win_en": "From now through Q1: Microsoft has changed its stance; the premium for 1M context will be quickly squeezed out; those who build baselines early will gain pricing power."
  }
 ],
 "timeline": [
  {
   "when": "3 个月内",
   "ev": "OpenAI 新一轮融资正式关闭，规模低于此前目标（含 100 亿美元下调）",
   "p": "70%",
   "why": "验证物：官方公告／监管披露文件",
   "ev_en": "OpenAI's new funding round officially closes, with size below the previous target (including a $10 billion reduction)",
   "why_en": "Verification: official announcement / regulatory disclosure documents"
  },
  {
   "when": "3 个月内",
   "ev": "Kimi K3.1 Pro 在 Azure Foundry 转为正式 GA 并公布定价",
   "p": "65%",
   "why": "验证物：Azure 定价页／官方博客",
   "ev_en": "Kimi K3.1 Pro transitions to official GA on Azure Foundry and announces pricing",
   "why_en": "Verification: Azure pricing page / official blog"
  },
  {
   "when": "3 个月内",
   "ev": "微软—甲骨文算力租约进一步缩减或正式终止公告",
   "p": "60%",
   "why": "验证物：双方财报／8-K 与公告",
   "ev_en": "Announcement that the Microsoft–Oracle compute lease is further reduced or formally terminated",
   "why_en": "Verification: both companies' financial reports / 8-K and announcements"
  },
  {
   "when": "3 个月内",
   "ev": "美中期选举后首场 AI 相关听证或立法推进（数据中心或前沿模型）",
   "p": "60%",
   "why": "验证物：国会日程／委员会公告",
   "ev_en": "First AI-related hearing or legislative progress after the U.S. midterm elections (data centers or frontier models)",
   "why_en": "Verification: congressional calendar / committee announcements"
  },
  {
   "when": "3 个月内",
   "ev": "至少一家头部 AI 公司下调资本开支指引",
   "p": "55%",
   "why": "验证物：财报电话会／指引修订",
   "ev_en": "At least one leading AI company cuts capex guidance",
   "why_en": "Verification: earnings call / guidance revision"
  },
  {
   "when": "1–2 个月内",
   "ev": "出现 Agent 支付/授权行业标准草案（卡组织或支付平台牵头）",
   "p": "55%",
   "why": "验证物：Visa/Mastercard/Stripe 官方博客或标准文档",
   "ev_en": "Draft industry standard for agent payments/authorization emerges (led by card network or payment platform)",
   "why_en": "Verification: Visa/Mastercard/Stripe official blog or standards document"
  },
  {
   "when": "3 个月内",
   "ev": "内存涨价传导至 AI 服务器报价（BOM 环比上调）",
   "p": "50%",
   "why": "验证物：供应链报价／服务器厂财报",
   "ev_en": "Memory price hikes pass through to AI server quotes (BOM raised month-on-month)",
   "why_en": "Verification: supply chain quotes / server makers' earnings"
  },
  {
   "when": "6 个月内",
   "ev": "国内首笔「存量机房/配电资产 AI 化」公开交易（并购或融资）",
   "p": "40%",
   "why": "验证物：并购公告／融资披露",
   "ev_en": "China's first public deal to \"AI-enable existing data center/power distribution assets\" (M&A or financing)",
   "why_en": "Verification: M&A announcement / financing disclosure"
  }
 ],
 "sources": [
  {
   "k": "深度版日报（每日AI核心信息）",
   "v": "40.2 KB",
   "k_en": "In-Depth Daily Report (Daily Core AI Information)"
  },
  {
   "k": "AI 官方视频监控台账",
   "v": "55 条事件 · 55 条带原文链接",
   "k_en": "AI Official Video Monitoring Log",
   "v_en": "55 events · 55 with original source links"
  },
  {
   "k": "站点生态数据",
   "v": "news 600 · skills 72 · openhub 68 · reddit 25",
   "k_en": "Site ecosystem data"
  },
  {
   "k": "官方事件 AI 精选",
   "v": "10 条（带原文链接）",
   "k_en": "Official Event AI Picks",
   "v_en": "10 items (with source links)"
  }
 ],
 "lead": "Kimi K3.1 Pro 开放权重：2.8 万亿总参、104B 激活，昇腾与 Azure Foundry 同日接入",
 "lead_en": "Kimi K3.1 Pro open weights: 2.8 trillion total parameters, 104B activated; Ascend and Azure Foundry onboarded on the same day"
}