The Information2026-11-08 20:26 UTC
92政策
英伟达董事会宣布启动独立调查，此前有报道称 CEO 黄仁勋曾是一起外国情报行动的目标。黄仁勋在公司内部备忘录中承认自己曾有婚外情。耶鲁大学的 Jeffrey Sonnenfeld 认为黄仁勋保住职位的概率“大约五五开”。
意味着什么: 现在治理问题比个人问题更重要：投资者在看董事会是否发现公司机密或 CEO 席位面临风险。
- The Information — Nvidia Board Is Launching Investigation After Reporting That CEO Targeted By Chinese Informant First report of the board probe; independent directors + outside counsel.
- TechCrunch — Nvidia will investigate CEO Jensen Huang's relationship with possible Chinese spy Huang admitted the affair in a company memo.
- Wired — Jensen Huang Hired a PR Firm Known for Handling Sex Scandals Last Year Heller PR firm on retainer since October 2025.
- TechCrunch — Is Jensen Huang's job on the line? Yale's Jeffrey Sonnenfeld estimates it's “about a 50-50” Governance expert's odds on Huang staying CEO.
- 量子位 — 黄仁勋承认出轨发全员信道歉！英伟达启动紧急调查 Huang's apology letter to all staff; board's emergency probe.
The Decoder2026-11-07 17:37 UTC
90融资
Anthropic 下周将启动 IPO 路演，目标在感恩节前融资 800–1000 亿美元，估值 1.25–1.75 万亿美元——远低于此前期望的 2 万亿美元。此前科技股已连续两周剧烈波动。
意味着什么: 这是首家纯 AI 实验室上市，将给整个 AI 热潮定价：开局疲软会给行业融资降温，开局强劲则重新打开水龙头。
- The Decoder — Anthropic pushes ahead with pre-Thanksgiving IPO despite tech selloff, accepts lower valuation Valuation $1.25–1.75T, raise $80–100B.
- The Information — Anthropic to Proceed With IPO Roadshow Next Week Roadshow timing; follows two weeks of volatility.
TechCrunch2026-11-06 22:53 UTC
88融资
OpenAI 已完成最新一轮融资的第一部分，募资 110 亿美元，估值 1.25 万亿美元，下周将向投资者通报第三季度业绩。
意味着什么: 一周内出现两个 1.25 万亿美元估值——Anthropic 与 OpenAI 如今定价相同，市场即将判定两者是否都值这个价。
- TechCrunch — OpenAI reportedly raises an additional $11 billion at a $1.25 trillion valuation Round not yet closed; expected in coming weeks.
- The Information — OpenAI Expected to Brief Investors Next Week After Raising $11 Billion Investor call next week; Q3 results.
The Decoder2026-11-08 09:50 UTC
86模型
在阿里巴巴的 Co-Design Bench（取自真实用户请求，都是曾让助手翻车的任务）上，表现最好的 GPT-6 Astra 也只能完全解决 26.4% 的任务，而在旧基准上同类模型得分超过 99%。
意味着什么: 基准差距本身就是新闻：演示里近乎完美的助手，在真实任务中近四分之三仍需人工收尾。预算里要留出检查时间，不只是 token 费用。
- The Decoder — AI coding agents still need human cleanup on nearly three out of four tasks in a real-world test Alibaba's Co-Design Bench; GPT-6 Astra tops at 26.4%.
The Decoder2026-11-09 15:44 UTC
79政策
腾讯首席 AI 科学家姚顺雨警告，AI 很快将能研发生物武器，并主张走中间路线：封闭模型拦截过度，而开源模型的安全过滤器容易被剥离——因此腾讯直接把危险知识从即将发布的 Hy4 模型的训练数据中过滤掉。
意味着什么: 安全辩论正在从“开源还是封闭”转向“哪些东西根本不该进训练数据”。
- The Decoder — Tencent's AI chief says AI bioweapons are coming, but closing off models isn't the answer Yao Shunyu's middle-ground argument; Hy4 training-data filtering.