概念词典 · 宏观
Mag7 七大科技巨头
美股最具市值与定价权的七家科技公司：苹果、微软、Google、亚马逊、Meta、英伟达、特斯拉，AI 算力投入的需求侧主力。
别名： Mag7／ Mag 7／ Magnificent Seven／ 七大科技巨头／ 七姐妹
是什么
Magnificent Seven（Mag7） 是华尔街对美股最具影响力的七家科技公司的统称：
| 公司 | 在 AI 供应链里的角色 |
|---|---|
| $AAPL Apple | 设备终端 + Private Cloud Compute |
| $MSFT Microsoft | OpenAI 大股东 + Azure AI 主战场 |
| $GOOGL Google | 自研 TPU + Gemini |
| $AMZN Amazon | AWS Trainium / Inferentia 自研芯片 |
| $META Meta | 自研 MTIA ASIC + Llama 开源 |
| $NVDA NVIDIA | GPU 唯一霸主 |
| $TSLA Tesla | Dojo 训练芯片 + Optimus 机器人 |
为什么 Serenity 反复提它
Mag7 的资本支出 (capex) 直接决定 AI 上游产业链的景气度。他们今年宣布建多少数据中心、买多少 GPU、签多少光模块订单——下个季度就是上游公司的收入。
具体到 Serenity 的研究框架：
- $MSFT、$META、$GOOGL、$AMZN 这四家是 AI capex 四大金主，他们的 capex 指引修订是先行指标
- $NVDA 的 GTC 演讲是 CPO/HBM/光子学路线图的官方权威发布
- $TSLA Optimus 量产时间表决定机器人产业链节奏
”Mag7 vs 非 Mag7” 二分法
很多隐形冠军的判断核心问题就是：这家公司是不是被 Mag7 之一深度绑定。$IREN 被他点名警示的原因之一就是还没有 Mag7 客户合同。
提到此概念的推文 (88)
按时间倒序 · 滚到底自动加载更多
-
OpenAI构建独立生态激怒七巨头，被视为真正的颠覆者。
这是一个逆向观点，但如果 Sam Altman / OpenAI 能激怒包括 $AAPL 在内的所有“七巨头(Mag7)”，说明他们做对了什么。他们正在构建自己的生态系统，而不是被卷入另一个生态。自 $GOOGL、苹果、微软等公司以来，已经很久没有出现过真正的颠覆者了。
英文原文
Contrarian take, but Sam Altman / OpenAI is doing something right if he's pissing off every Mag7 like $AAPL. By building their own ecosystem instead of being sucked into another. It's been awhile since $GOOGL, Apple, Microsoft, and others had some genuine disruptors.
-
总结近期半导体与 AI 供应链的多项变化
这是我对最近半导体发展的一个 TLDR： 1. $TSM 在强推 CoPoS - VisEra / 其他公司可能会比预期更早起飞。 2. $AAPL 选择 $INTC 做半导体生产，这是个重大变化，因为他们通常会选 TSM。美国制造继续起飞，Intel 继续起飞。 3. $NVDA Vera Rubin 最近据说对散热架构做了很新的调整。 “台湾的热管理供应商正在成为 AI 硬件生态里增长最快的板块之一” - 上个月的说法。 “Vera Rubin server 架构预计会推动数据中心散热和系统设计发生根本变化” 我之后会专门看热管理生态，也许该开始研究一下？ 4. 2D NAND 短缺在 Samsung、Micron 和其他玩家退出后继续恶化。 Macronix、Windbond 都在起飞，这对 GigaDevice 和其他利基玩家也有影响。 5. “大科技公司据称愿意出资帮助 SK Hynix 建 fab 和 EUV” 说明 memory 短缺严重到 Mag7 都愿意出钱，那 $MU、SK Hynix、Samsung 都在起飞。 6. $TSM 2026 年 4 月净营收 126 亿美元，收入同比增长 30%，半导体继续起飞。 7. Anthropic 需要算力 -> SpaceX。 这意味着算力需求极其强，对 $NBIS 等都是“BRRR”级别利好。 但很有意思的是，他们绕开了 Neoclouds，选了 SpaceX。 8. “SKC 将在今年年底前加速为美国客户量产玻璃基板” “提前于原计划，已宣布今年年底前推进” 像 $LPK 这样的玻璃基板量产玩家，以及 SKC 这类相关公司都在起飞。 玻璃时间线被提前了，真的 heavy brrr glass。 9. AI 服务器需求和 GaN 争夺加剧，电源芯片短缺也在加深。 也许是时候看看电源芯片瓶颈了？ 10. Adata 说 DRAM 和 NAND 闪存合约价在 2026 年第二季度都会上涨 40% 以上。 这对 $MU、SK Hynix、Samsung、$SNDK 等都是正面。
英文原文
Just a TLDR of recent semi developments: 1. $TSM pushing hard CoPoS - VisEra/others might go brrr earlier than expected. 2. $AAPL goes with $INTC for semi production, which is a major shift cause they normally go with TSM. Made in America go like Intel go brrr. 3. $NVDA Vera Rubin reportedly makes changes to cooling architectures very recently. "Taiwan's thermal management suppliers are emerging as one of the fastest-growing segments in the AI hardware ecosystem" - From Last Month. "Vera Rubin server architecture is expected to drive a fundamental shift in data center cooling and system design" Will cover thermal ecosystem later, maybe it's time to take a look? 4. 2D NAND shortage spirals after Samsung, Micron, and rivals exit market Macronix, Windbond go brrr. implications for GigaDevice and other niche players. 5. "Big Tech reportedly offers to fund SK Hynix fabs and EUV" - Memory that badly bottlenecked that mag7 wants to pay for it, so $MU, SK Hynix, Samsung go brr. 6. $TSM 2026 net revenue $12.6B for April 2026. Revenue up 30%, Semis keep going brr. 7. Anthropic needs compute -> SpaceX. So implications for compute demand is extreme here which is BRRR $NBIS and others. But it's very interesting they sidestepped Neoclouds and went with SpaceX. 8. "SKC to Accelerate Mass Production of Glass Substrates for U.S. Clients by the End of the Year" "the end of the year, ahead of its original plan, it has been announced" Glass Core substrates players like $LPK for mass production and other related players like SKC go brrr. Glass timelines moved up. heavy brrr glass. 9. "Power chip shortages deepen as AI server demand and GaN battles escalate" Maybe time to look into the power chip bottleneck anon? 10. "Adata said DRAM and NAND flash contract prices will each climb more than 40% in the second quarter of 2026" Another positive for $MU, SK Hynix, Samsung, $SNDK, and others.
-
说 META 没踩到最低点，但长期不需要精确择时。
@mi20483980476 不错……我没能把 $META 的最低点踩得最漂亮，但长期看也不需要。 不是每天都会出现 mag7 在短时间里暴跌 20% 的情况。 https://t.co/kLtzZeuFQl
英文原文
@mi20483980476 Nice… I didn’t manage to time the very bottom for $META but don’t need to in the long run. Not every day does mag7 growing at record paces sell off -20% in short timeframes. https://t.co/kLtzZeuFQl
-
AI数据中心供应链严重依赖亚洲，中东冲突加剧流动性风险，稀土应成美国国安优先事项
Hi Ron，这个话题分为两个不同的部分： 1. 超大规模云服务商（$ORCL、$META、$AMZN）：他们的支出超过了盈利水平（$GOOGL 是罕见的例外，而 $AAPL 相对而言支出并不多）。 因此市场担心这并没有转化为实质性的收入。 尤其是像 Oracle 这样负债累累的公司，降息是使其稍微可持续的重要因素（如果你看债务利息，数目非常庞大），而降息既能推动产能扩张，也能带来实质性的成本节省。 由于伊朗战争和原油价格上涨，原定2026年的降息预期已经落空。 雪上加霜的是，人们担心他们的 AI 数据中心建设供应链会中断。 如果你看看他们的 AI 数据中心部件产地： - 来自台湾 - 来自韩国 - 来自日本 这是一个全球化的供应链。 虽然美国看起来没问题，因为它在石油、液化天然气、氦气和其他半导体供应链所需的资源方面是绝缘的。 但我们的亚洲合作伙伴并不处于长期冲突中。 所以当你看到亚洲那些规模较小、专业化的公司时，它们正在苦苦挣扎。 这种影响会层层放大，一直传导到终端的亚马逊网络服务 AI 数据中心（成本上升、供应短缺）。 2. AI 整体而言：投资者对此非常乐观。 问题在于实现路径，而美国在表面上看起来被中国卡住了。 - 美国投入过多，而中国和其他国家正在蒸馏我们的最新模型（需要 KYC 端点管控），坐享其成。 - 随着 AI 项目的规模扩大，它正变得越来越依赖中国公司。 我认为这是因为美国缺乏稀土，这应该成为我们的头号国家安全优先事项。 这是制造新一代 AI 硬件、机器人供应链和太空探索所需的。 这些资源都由中国和俄罗斯控制。即使是领先的"西方"公司，我已发现它们仍然主要依赖中国/俄罗斯。 这是我们确保前沿项目安全最大的漏洞之一。 如果你看光子学（AI 硬件如何通过光加速），我们没有西方的供应链来满足需求。主要材料来源是中国的 $AXTI 和中国的 Vital 两家公司，然后流转到日本或其他公司，最后进入美国。 如果你看我们的人形机器人项目（我们如何将 AI 带入现实世界），$TSLA Optimus 的所有躯体都是在中国制造的。 美国供应链缺乏制造这些组件所需的稀土，因为成本太高。 3. 中东地区的冲突 近期下跌的一个重要原因是流动性问题。我们在阿联酋、中东的合作伙伴主要在为私募市场提供资金（想想 OpenAI），他们大量的支出流向 AI 市场。 他们也是从 Meta 到微软等科技七巨头最大的投资者之一。 如果他们的油田继续受到干扰，他们可能不得不从美国市场撤出流动性。 这既影响这些公司在 AI 上的支出能力，也会导致美国市场的抛售下跌。 如果有takeaway（要点）的话：请把稀土列为美国头号国家安全优先事项。特朗普总统已经投资了 $USAR、$MP 和几个单独的公司。 但还有许多更多至关重要的、尚未盈利的公司，它们对于制造 AI、自动驾驶汽车和机器人都是必需的。我们需要打破对中国/俄罗斯的依赖。
英文原文
Hi Ron, so there's two different parts to it: 1. Hyperscalers ( $ORCL, $META, $AMZN): They're spending more than they have with profit ( $GOOGL is the rare exception, and $AAPL isn't really spending much relatively). So markets are worried less worried it's translating into material revenue. Especially with companies like Oracle taking on immense amounts of debt, rate cuts were a large driving factor in why it's slightly more sustainable (if you look at debt interest, it's massive), and cutting rates drives both forward expansion and material amounts of savings. Due to the War in Iran, rising crude, former projections with cutting rates in 2026 are now gone. This is compounded that there are worries that their AI buildout supply chain gets disrupted. If you look at where their AI datacenter parts are made: - it comes from Taiwan - it comes from Korea - it comes from Japan It's global. While the US might look like it's fine, since it's insulated to Oil, LNG, helium, and others needed for the semiconductor supply chain. Our Asian partners are not in a long drawn out conflict. So when you look at companies in Asia that are more small and niche they are struggling. And this gets compounded tens of times until it reaches all the way down to the end Amazon Web Services AI datacenter (increased costs, lack of supply). 2. AI as a whole: Investors are very bullish on it. It's just how we get there, and America looks strangled by China optically. - America is spending way too much on it, with China and others distilling our latest models (need KYC endpoints), and reaping all the benefits. - As we scale our AI programs, it's becoming increasingly reliant on Chinese companies. I'd argue it's because America lacks the rare earths, which should be our #1 national security priority. This is needed to make new generations of AI hardware, robotic supply chains, and Space. They're all controlled by China and Russia. Even the leading "Western" companies, I've identified they're mainly relying on China/Russia still. And this is one of the biggest vulnerabilities in securing our frontier programs. If you look at photonics (how AI hardware is sped up by light), we no Western supply chain to fulfill our needs. It's $AXTI (in China), and Vital (in China) as the two main sources of materials needed to make them, that gets passed down to Japan, or other companies before they end up in US. If you look at our humanoid program (how we move AI to the real world), all the bodies of $TSLA Optimus are made in China. American supply chains lack the rare earths needed to make the components because it gets too expensive. 3. Disruption in the Middle East A large part of the recent drop is mainly due to liquidity. Our partners in the UAE, Middle East are largely funding private markets (think OpenAI) and a lot of their spend going to AI markets. They're also one of the biggest investors in Mag7 from Meta to Microsoft. If their oil fields continue getting disrupted they may have to pull out liquidity from US markets. This impacts both the amount of money these companies can spend on AI. As well as causing a drop in American markets from selling. If there's any takeway: Please make Rare Earths Amercia's #1 national security priority. President Trump already invested in $USAR, $MP, and a few individual names. But there are many more extremely critical, yet unprofitable companies that are needed to make AI, to self driving cars, to robotics. And we need to break our reliance on China / Russia.
-
认为美股科技股卖压过头，SpaceX IPO 会制造流动性真空，短期别碰期权。
$MSFT、$GOOGL 和 $META 这轮美国科技股抛售，看起来已经过度了？ Mag7 大致跌幅： Microsoft 年初至今跌 -21.9% Google 年初至今跌 -7.69% $AMZN 年初至今跌 -7.36% Meta 年初至今跌 -8.54% 更糟的是，如果 SpaceX 以 1.7 万亿美元估值在快速纳入指数后 IPO。 大型 IPO 通常会制造巨大的市场流动性真空（对 SpaceX 来说当然是好事）。 不过，我不认为市场真的相信伊朗战争会很快结束。 -> 大盘股出现巨大流动性抽水， -> SpaceX IPO 还会带来更多抽水， -> 尽管基本面在改善 = 短期别碰期权。 长期来看，我预计 Mag7 会恢复。
英文原文
The US tech selloff from $MSFT, $GOOGL and $META looks really overdone now? Mag7 roughly down: Microsoft down -21.9% YTD Google down -7.69% YTD $AMZN down -7.36 YTD Meta down -8.54% YTD What makes things worse is SpaceX IPOing at $1.7T if they get fast-index inclusion. Mega-IPOs typically serve as a massive market liquidity vacuum (good for SpaceX though) However, I don’t think markets are truly convinced War in Iran is ending anytime soon. -> Massive liquidity drain from large caps, -> more upcoming soon with SpaceX IPO -> despite improving fundamentals = don’t touch short term options. Long term I expect Mag7 to recover.
-
认为 Mag7 卖压已经过头，资本会重新流回美国科技。
感觉从 $META 593 美元到 $MSFT 373 美元这一段 Mag7 抛售…… 是不是有点过头了？ 现在看起来，超大规模云厂的支出 / ROI / 宏观环境的风险收益都挺有吸引力。 我觉得很快会看到资本重新流回美国科技板块。 https://t.co/KBCi7gMfGT
英文原文
Feel like Mag7 selloff from $META at $593 to $MSFT at $373… Has gone a bit too far? At this point, risk/reward feels on hyperscaler spend vs. ROI vs. macro climates, feels compelling. And we should see capital rotations back into the American tech sector soon. https://t.co/KBCi7gMfGT
-
认为地缘政治不会导致科技巨头削减AI资本支出，成本分析优于需求预测。
@nengcunren2527 是的，我同意很多因素在于需求端，但我不认为“七巨头”(Mag7) 或 $GOOGL 会因为伊朗局势紧张而削减资本支出(capex)，从而在人工智能(AI)竞赛中输给了中国。 局势过于错综复杂，在战争升级的短短几天内，还无法对需求进行建模。但成本分析(cost analysis)则更具可操作性。
英文原文
@nengcunren2527 Yep agreed a lot of it is on demand side but I don’t think mag7 or $GOOGL is going to cut capex and lose the AI race to China over Iran tensions. It’s way too multifaceted and can’t model in demand yet from just a few days into war escalation. But cost analysis is more do-able.
-
谷歌CTO暗示AI支出激增，澄清市场误判，看好AI基建加速。
@4intheflames $GOOGL 人工智能首席技术官刚刚明确暗示未来8年将增加支出。 https://t.co/zSf77JAHBQ （该推文引用了 @aleabitoreddit 的推文，引用内容仅供理解语境）： “谷歌数据中心建设规模可能超过1万亿美元” 这对谷歌TPU生态系统1.5万亿美元以上的资本支出意味着什么： 对$LITE、联发科、$AVGO、$TSM、SK海力士、三星等均为广泛利好。 在接受福布斯关于人工智能资本支出的采访时，$GOOGL首席技术官表示： “今年（资本支出）在1750亿至1850亿美元之间，假设不会下降，人们可以想象这在10年内可能延伸到某个巨大的数字。” 福布斯评论道：“谷歌的数据中心野心与OpenAI有着巨大差异：谷歌是一台赚钱机器。” 昨天，市场抛售了OpenAI和$ORCL相关的半导体股票，导致一些“星际之门”相关项目暂停。 市场将OpenAI的成长阵痛与由最富有的超大规模云服务商资助的更广泛的人工智能基础设施建设混为一谈。 从$GOOGL和印钞机般的Mag7角度来看： 人工智能基础设施建设在未来十年看来只会加速。
英文原文
@4intheflames $GOOGL CTO of AI literally just signaled increase spend over the next 8 years. https://t.co/zSf77JAHBQ
-
谷歌AI基建支出庞大且持续加速，利好相关半导体供应链。
“谷歌数据中心建设规模可能突破1万亿美元” 这对谷歌TPU生态系统1.5万亿美元以上的资本支出(capex)意味着什么： 对$LITE、联发科(Mediatek)、$AVGO、$TSM、SK海力士、三星等普遍利好。 在接受福布斯关于AI资本支出的采访时，$GOOGL首席技术官(CTO)表示： “今年在1750亿至1850亿美元之间，假设不会下降，人们可以想象这在10年内可能延伸到某个巨大的数字” 福布斯评论道：“谷歌的数据中心野心与OpenAI有着巨大差异：谷歌是一台赚钱机器” 昨日，市场抛售了OpenAI和$ORCL相关的半导体股，暂停了一些Stargate相关项目。 市场将OpenAI的成长阵痛与由最富有的超大规模云服务商(hyperscalers)资助的更广泛的AI基础设施建设混为一谈。 从$GOOGL和印钞机般的Mag7角度来看： AI基础设施建设在未来十年看来只会加速。
英文原文
“Google’s Data Center Buildout Could Top 1 Trillion” The implications for $1.5T+ in capex spend for the Google TPU ecosystem: From $LITE, Mediatek, $AVGO, $TSM, SK Hynix, Samsung and others are widely positive. In an interview with Forbes on AI capex spend, $GOOGL CTO stated: “We’re at $175 to $185 billion this year, one could imagine, assuming it's not going to go down, that this could extend to some big number over 10 years” Forbes comments: “There’s a big difference between Google’s data center ambitions and OpenAI’s: Google is a money-making machine” Yesterday, markets sold off semi names from OpenAI + $ORCL putting a pause on some Stargate related projects. The market conflated OpenAI's growing pains with the broader AI infrastructure buildout funded by the richest hyperscalers. From $GOOGL and a money-printing Mag7 perspective: The AI buildout only looks to accelerate over the next decade.
-
AI从训练转向推理，电网成新瓶颈，电力股或迎起飞。
好问题！答案是宏观/芯片。2023-2024年AI起步时，美联储将利率大幅上调至5%+。公用事业股通常类似债券替代品，因为基建背负大量债务。当时最佳选择是$NVDA，因为硅/芯片短缺。2023年市场不太关心电网电力，因为服务器尚未联网。但现在Mag7囤积的数百万芯片终于开始运行。虽然$TSM仍是瓶颈，但美国电网可能是最大瓶颈之一。此外，我们从训练（约占AI电力预期的10-15%）转向推理，后者由数十亿人24/7使用。因此，在去年降息、今年进一步降息以及前所未有的推理扩张后，现在可能是电力/电网起飞的时候。
英文原文
Great question! Answer to that is macro/chips. 2023, 2024 when AI first started taking off, fed was jacking rates up to 5%+. Utilities are often kinda like bond proxies since they carry a lot of debt for infra buildout. Best play then was $NVDA since silicon/chips were in shortage. Market didnt really care about grid power back in 2023 because servers weren' connected. But now there's millions of chips hoarded by mag7 that are finally getting turned on. So while $TSM is still a bottleneck, US power grid is likely one of the biggest ones. Also we've shifted from training (which was like 10-15% of AI power expectancy) into inference, which just runs 24/7 used by billions of people. So after rate cuts last year, more this year, and just unprecedented inference expansion, probably now is the time for power/grid to take off.
-
独立电力公司具定价权，AI需求推动电网扩张。
@adenois 几点补充，$CEG、$VVST 等是独立公司，因此拥有更强的定价权。它们在 $XLU 的集中度中占比较大，且电力容量价格正在飙升。随着 AI 的发展，需求确实大于供应，Magnificent Seven（科技七巨头）正在为这一历史性的电网扩张提供主要资金。
英文原文
@adenois few things, $CEG, $VVST and others are independent so they have more pricing power. they're also a large % of $XLU concentration + power capacity prices are skyrocking. And with AI, yes demand > supply and mag7 is largely funding this new historic grid expansion.
-
AI 驱动下，三星和海力士营业利润增速惊人，有望在 2027 年追平或超越美股巨头。
全球最盈利公司排名（Mag 7 vs. 世界） 2025->2026->2027 年营业利润(Operating Income)预测。 #1: $NVDA (美国, 4.4T 市值) 🇺🇸 ~1350 亿 -> 1865 亿 -> 2401 亿美元 #2 三星电子 (韩国, 8200 亿市值) 🇰🇷 ~302 亿 -> ~1700 亿 -> ~2267 亿美元 #3 $MSFT (美国, 2.9T 市值) ~1285 亿 -> 1530 亿 -> 1815 亿美元 #4 $GOOGL (美国, 3.7T 市值) ~1290 亿 -> 1420 亿 -> 1730 亿美元 #4 海力士 (韩国, 4100 亿市值) ~327 亿 -> ~1240 亿 -> ~1610 亿美元 #5 $APPL (美国, 3.76T 市值) 1331 亿 -> 1460 亿 -> 1605 亿美元 #6 $AMZN (美国, 2.13T 市值) 800 亿 -> 1050 亿 -> 1365 亿美元 #7 $Meta (美国, 1.62T 市值) 833 亿 -> 970 亿 -> 1215 亿美元 #8 $TSLA (美国, 1.31T 市值) 44 亿 -> 80 亿 -> 240 亿美元 韩国 8200 亿市值的三星电子预计将在 2027 年在营业利润上追平 $NVDA。 同时，海力士预计将在 2027 年在营业利润上超越 $APPL 和 $AMZN。 主要结论是，由于人工智能(AI)的加速部署，美国超大规模云服务商和韩国股票的增长令人惊叹。
英文原文
Global ranking of the most profitable companies in the world (Mag 7 vs. World) Projections for 2025->2026->2027 (Operating Income). #1: $NVDA (USA, 4.4T MC) 🇺🇸 ~$135.0B -> $186.5B -> $240.1 Billion #2: Samsung Electronics (Korea, $820B MC) 🇰🇷 ~$30.2B -> ~$170B -> ~$226.7 Billion #3 $MSFT (USA, $2.9T MC) ~$128.5B -> 153.0B -> $181.5 Billion #4 $GOOGL (USA, $3.7T MC) ~$129.0B -> $142.0B -> $173.0B #4 Sk Hynix (Korea, $410B MC) ~$32.7B -> ~$124B -> ~$161.0 Billion #5 $APPL (USA, $3.76T MC) $133.1B -> $146.0B -> $160.5B #6 $AMZN (USA, $2.13T MC) $80.0B -> $105.0B -> $136.5B #7 $Meta (USA, $1.62T MC) $83.3 -> $97.0B -> $121.5B #8 $TSLA (USA, $1.31T MC) $4.4B -> $8.0B -> $24.0B Samsung Electronics, a $820B company in Korea is projected to catch up to $NVDA in 2027 in operating income. Meanwhile Sk Hynix is projected to overtake both $APPL and $AMZN in operating income in 2027. The main takeaway is that the growth of both US hyperscalers and South Korean equities is astounding due to Artificial Intelligence ramp.
-
七巨头盈利强劲支撑科技股，但加密市场低迷警示短期流动性风险。
我认为“七巨头”(Mag7)的盈利过于强劲，不足以引发全面修正。$META 指引营收增长约 30%。$GOOGL 庞大的 1800 亿美元资本支出由其 2025 年超 1320 亿美元的营业利润资助。这些资金流入 AI、存储和半导体领域，为其提供了底部支撑（例如，存储占 AI 集群物料清单(BOM)的很大比例）。即使是其他板块的中盘股如 $RDDT，在 35% 的净利率转化为净利润后，远期市盈率(Forward P/E)估值也接近 26-30 倍，营收同比增长 70% 且远期增长超 50%。增长目前看来仍可持续，且软件等其他板块大幅折价。但比特币/以太坊尚未恢复通常是短期流动性问题的最佳“煤矿金丝雀”(canary in the coal mine)。我预计中期选举前会出现更广泛的复苏。
英文原文
IMO Mag7 earnings are too strong for a full-blown correction. $META projecting guided ~30% revenue growth. $GOOGL's enormous $180B capex spend is funded by it's $132B+ operating income from 2025. All those funds trickle down into AI, memory, semis and puts a floor on them (eg. memory is a huge percentage of BOM for AI clusters). Even mid cap in other segments like $RDDT, is close to ~26-30 forward P/E estimates after 35% of revenue as net income -> 70% Y/Y revenue growth and 50%+ forward. Growth still looks sustainable right now, and other segments like software are heavily discounted. But the fact Bitcoin/Ethereum is not recovering yet is usually the best canary in the coal mine for short term liquidity issues. I'd expect a broader recovery going forward to midterms.
-
建议放弃逆向股，重仓三星和SK海力士等AI存储巨头。
老实说，光看这份预估净利润图表…… 让人不禁疑惑，为什么人们还要试图全仓押注像 $PYPL、$FIG 或 $DUOL 这样的逆向反转交易？ 我觉得，坐在像三星或 SK 海力士这样盈利且超增长的公司里“躺平”，才是最容易做的事？ 它们相比“七巨头(Mag7)”规模相对较小，但预计将在 2027 年成为全球最盈利的公司。 有时候，最好的做法就是参与这些显而易见的交易，开启“简单模式”的生活。
英文原文
Honestly just looking at the quoted net income chart… Makes you wonder why people try and full port it into contrarian turnaround plays like $PYPL, $FIG, or $DUOL I feel like sitting back and chilling in profitable, hyper-growth companies like Samsung or SK Hynix is just the easiest thing to do? They’re relatively small compared to Mag7, but projected to become the most profitable companies in the world in 2027. Sometimes the best thing to do is just join in on the no-brainer trades and live life on easy mode.
-
白宫与科技巨头共同背书，推动英特尔成功。
@Crypto_fox7 没错。并不是每天都有像 $INTC 这样得到白宫背书的公司。政府和所有像 $NVDA 和 $APPL 这样的“七巨头”公司都在推动英特尔取得成功。
英文原文
@Crypto_fox7 Yep. Not every day do you get a company backstopped by the White House like $INTC. Both the government and all the Mag7 companies like $NVDA and $APPL are just pushing for Intel to succeed.
-
讽刺WMT增长堪比英伟达，应加入七巨头。
@wilson_eth 确实，市场需要把 $WMT 加入“七巨头(Mag 7)”，毕竟这家微波炉转售商的增长显然与 $NVDA 相当。
英文原文
@wilson_eth True, markets need to add $WMT to the Mag 7 since this microwave reseller is growing on par with $NVDA apparently.
-
2026年1月1日美股评级：推荐SMCI、INTC等复苏及AI基建标的，回避高估量子及零售股。
欢迎来到2026年。1月1日评级： 强烈买入： $TTD $SMCI $AIRO $INTC $HIMS $AXTI $TSM $NBIS $CIFR 三星电子 (KRX: 005930) $HUT $IREN $WULF $GLXY $TSSI $META $ETOR $CRCL 买入： $KRKNF $ONDS $GEMI $NVDA $MU $AMKR SK海力士 $SNAP $RDDT $AAOI $COHR $FISV $FLY $DJT $LITE $AMZN $MRVL $AVGO $OSS $BULL $ORCL $CRDO $ALAB 回避： $RGTI $QBTS $RGTI $BMNR $ETH $PLTR $WMT _ 简而言之的想法： TTD - 估值完全重置，年初至今下跌67%，叠加年底税务抛售。进入2026年是极佳的复苏标的。 SMCI - 仅因推迟一个季度以符合新Blackwell规格而交易得像一家困境公司。前瞻收入同比增长50%，市销率(P/S)接近0.5。税务收割后的极佳复苏标的。 AIRO - 资产负债表约1/6为现金。随着政府加速投资，无人机领域备受追捧。另一只被抛售的IPO新股。进入2026年，尤其在热门细分领域，是极佳的复苏标的。市销率约3.8倍，相比ONDS的25-30倍市销率，当然其教育等业务对利润率计算有很大干扰。 INTC - 它实际上已成为美国政府的半导体臂膀。超大规模云厂商很可能被激励（强烈施压）在有机会时优先使用Intel而非TSM、三星等。我不会赌美国政府会输。 HIMS - 进入2026年前大幅抛售。从70多美元下跌。销售/流量下降，但Zava收购/增长应在2026年带来巨大顺风。尤其是有数亿美元回购，前两个月是强劲的复苏标的。 AXTI - 之前发过相关论点。CEO称“40%的磷化铟(InP)供应链”，InP将是2026-2027年超大规模AI建设中的巨大瓶颈，直到2028年有足够时间通过工程手段绕过它。 TSM - 我引用过很多次。利润率提升。需求最大化。未来几年极好的复利增长者。 三星电子 - 受益于代工/存储。在所有顺风助力业务方面简直是金蛋。 NBIS - 极强买入，$7-90亿年度经常性收入(ARR)，它实际上是5家不同公司，每年三位数增长。管理层引用20-30%息税前利润(EBIT)率，这只是等待游戏。 CIFR, HUT, IREN, WULF, GLXY - 整个数据中心板块在Oracle/OpenAI担忧后极度超卖。OpenAI最近融资$400亿，$AMZN再投$100亿等。因此关于资本支出(Capex)的担忧已大幅降低风险。这也是多方面的，例如比特币下跌影响$CIFR资产负债表，$GLXY在加密领域。但总体而言，新云(neoclouds)板块有巨大的复苏/扩张潜力。 TSSI - 类似SMCI。递延收入=核弹级利空。税务收割后应恢复，且大量收入将被确认。 META - 财报后因一次性税务问题遭遇巨大算法抛售。他们还削减了现实实验室等部门的资本支出/运营支出，这应为2026年的每股收益(EPS)带来巨大顺风。 ETOR - 市值$28亿，坐拥$12亿现金，仍保持双位数同比增长。$1.5亿回购应是不错的顺风，且年初至今表现带来的税务收割效应应会消退。 CRCL - 稳定币论点在2026年应非常稳固。 买入 KRKNF - Anduril合作伙伴+规模。2026年可能转板，国防支出带来大量顺风。 ONDS - 收入增长极具爆炸性，到处都有新的$1000万合同。大量现金余额资助研发。市销率很高，但该领域的投机性领导者如RKLB有估值溢价。 GEMI - 通常我不喜欢交易所，但Gemini从$30+ IPO跌至$10以下。极佳的复苏标的。 NVDA - 巨大的积压订单。大家都知道Nvidia的牛市逻辑。 MU - 存储火热。 SK海力士 - 存储火热。 AMKR - 受益于“美国制造”芯片生产扩张。 SNAP - 存储运营支出削减，存储货币化增加收入，Perplexity贡献$4亿。季度收入$15亿。如果他们将所有这些转化为$10亿+自由现金流(FCF)/年，即使收入完全停止增长，也会完全重估Snap。 RDDT - 说实话，未来10+年不会消失。它是社交媒体界的Robinhood，通过新的收入货币化方式极速增长，且极其盈利。 AAOI - Amazon, MSFT ASIC规模化的互连标的。 COHR - 受益于下一代ASIC的光子学部署。 FISV - 财报后抛售过多，税务收割后是强劲的复苏标的。 FLY - SpaceX IPO带动太空领域火热。税务收割结束后应表现良好，且即将有Northrop的中期催化剂。 DJT - 我从没想过会把它放在这里，但这只是因为他们的TAE合并。 LITE - Google TPU部署带来巨大的物料清单(BOM)，估值有吸引力。Google TPU修正预估后轻微抛售，但它基本存在于每个超大规模ASIC部署中。 AMZN - 七大科技巨头中估值不过高的一个。 MRVL - 分析师误导导致抛售，进入2026年强烈买入。尤其是MSFT Maia收入翻倍时，Marvell当前收入也将随之增长。 AVGO - 像NVDA一样，随着AI基础设施部署加速，强劲的长期持有标的。 OSS - 我曾发帖推测他们是Anduril的供应商之一。但无论如何，边缘计算在2026年将很火热，其1.8亿市值呈现有吸引力的上行空间。 BULL - 类似Robinhood，拥有巨大用户群，但他们只需找出货币化方法。 Oracle - 我认为抛售过多。几个月前我将其列入回避，但从$330跌至$190后，尤其在OpenAI再融资$400亿后，再次具有吸引力。 CRDO - 极高利润率，数据中心部署所需的必要连接。 ALAB - 极高利润率，数据中心部署所需的必要连接。 回避： 有很多在“高估名单”上的东西，比如$RKLB，我喜欢但除了这些之外我不会说回避它们。 RGTI, QBTS, RGTI - 量子名称仍然高估，且可能在未来几年无法交付自由现金流。 BMNR, ETH - 如果你看过我的ETH帖子，我不太看多，因为每天ETH燃烧量只是个位数到低双位数，这简直是笑话。 PLTR - 最被高估的AI名称之一。 WMT - 这怎么是40倍市盈率？这是沃尔玛？ __ （这些基于今日价格） 简而言之： Circle, Etoro, AIRO, Klarna, Figma等IPO名称在下跌+税务收割后进入2026年呈现有吸引力的上行空间。 大量如SMCI, HIMS等过去3个月下跌约40%的名称，在税务收割+一月效应后是极好的波段/复苏交易标的。 许多如FiserV或The Trade Desk等暴跌的名称在税务收割后也是良好的复苏交易标的。 许多数据中心股票如nebius, iren, cifr, wulf, galaxy也是极好的复苏交易标的。 许多其他领域如存储、瓶颈、光子学等在2026年只是极好的长期持有标的，尽管各自都触及历史新高。 仍有相当多高估的名称，从量子到某些太空股票（如planet或rocketlab），特定AI名称如Palantir到零售股票如沃尔玛，我可能会暂时回避，直到有轻微回调。 这只是一个简而言之，如果我仅做短期交易（非长期），但欢迎提问。
英文原文
Welcome to 2026. Jan 1st ratings: Strong Buy: $TTD $SMCI $AIRO $INTC $HIMS $AXTI $TSM $NBIS $CIFR Samsung Electronics (KRX: 005930) $HUT $IREN $WULF $GLXY $TSSI $META $ETOR $CRCL Buy: $KRKNF $ONDS $GEMI $NVDA $MU $AMKR SK Hynix $SNAP $RDDT $AAOI $COHR $FISV $FLY $DJT $LITE $AMZN $MRVL $AVGO $OSS $BULL $ORCL $CRDO $ALAB Avoid: $RGTI $QBTS $RGTI $BMNR $ETH $PLTR $WMT _ TLDR thoughts: TTD - Complete valuation reset dropping 67% YTD, compounded by EOY tax sell-off. Great recovery play going into 2026. SMCI - Trades like distressed company just because they delayed revenue by 1 quarter for new blackwell specs. Forward revenue is increasing 50% Y/Y, P/S close to .5 now. Great recovery play from tax harvesting. AIRO - Roughly ~1/6th balance sheet was cash. Everyone seems to be into drones, especially with accelerated gov inevstments. Another IPO name that got sold off. Great recovery play going into 2026 with esp. hot segment. Roughly ~3.8x P/S compared to ONDS trading at 25-30 P/S, but obviously there's quite a lot of other businesses like their education sector which messed up margin calculations quite a bit. INTC - It's literally become the semi arm of the US government. Hyperscalers will likely be incentived (strongly pressured) to use Intel whatever chance it gets over TSM, Samsung, etc. I would not bet against the US government. HIMS - Huge selloff going into 2026. Down from $70's. Sales/Traffic is down, but Zava acquisition/growth should add a huge tailwind going into 2026. Esp. with few hundred mill buybacks, strong recovery play first two monts in. AXTI - Posted thesis on this earlier. CEO - "40% of Inp supply chain", InP will be a huge, huge bottleneck for hyperscaler AI buildout 2026-2027 until there's enough time to engineer around it in 2028. TSM - I've covered this quote a lot. Increasing margins. Maxed out demand. Just extremely good compounder next few years. Samsung Electronics - benefits from foundry/memory. just golden egg regarding all the tailwinds helping the buisness. NBIS - Extremely strong buy, $7-9B ARR, it's literally 5 different companies growing triple digits Y/Y. management quoted 20-30% EBIT margins, it's just a waiting agme. CIFR, HUT, IREN, WULF, GLXY - Whole datacenter space is extremely sold off after Oracle/OpenAI fears. OpenAI recently raised $40B, another $10B from $AMZN, and more. So a lot of fears regarding capex spend has been de-risked. It's multifaceted too, eg. Bitcoin drop, affects $CIFR balance sheet, $GLXY in crypto space. But generally huge recovery play/ramp for neoclouds sector. TSSI - Similar to SMCI. deferred revenue = nuke. Should recover after tax harvesting + lot of revenue gets recognized META - Huge algorithmic selloff post earnings due to one-time tax. They also cut capex/opex spend of their reality labs and other departments and this should be a huge tailwind for EPS going into 2026. ETOR - Literally sitting on $1.2B with a $2.8B marketcap and growing double digits Y/Y still. $150M buyback should be a nice tailwind, and tax harvesting from YTD performance should subside. CRCL - Same as stablecoin thesis should be really solid going into 2026 Buy KRKNF - Anduril partner+ scale. Probable uplisting in 2026, lot of tailwinds from defense spending. ONDS -pretty explosive revenue growth, new $10m contracts left and right. large cash balance to fund r&d. Pretty high p/s but there's valuation premiums for speculative leaders in the space like rklb. GEMI - So i typically dont like exchanges, but gemini got nuked from $30+ IPO sub $10. pretty solid recovery play. NVDA - Huge backlog lol. Everyone knows bull case for nvidia MU - Memory is hot SK Hynix - Memory is hot AMKR - benefits from "made in america" chip expansion in prod. SNAP - Opex Cut from memory, increase revenue from memory monetization, $400m from perplixity. $1.5B revenue/quarter. They could literally stop growing revenue complelty if they convert all of that to $1B+ FCF/year, it would re-rate snap completly. RDDT - This is not going anywhere for the next 10+ years tbh, it's like robinhood of social media, growing extremely fast from new ways to monetize revenue, and just extrmeely profitable. AAOI - interconnect play for amzn, msft asic scale up. COHR - benefits from photonics rollout for next gen asics. FISV - Nuked a bit too much post ER, strong recovery play esp. post tax-harves.t FLY - Space is hot from SpaceX IPO. Should do well given tax harvesting is over, and they have medium lift coming up with northrop. DJT - I never thought i'd put this here lol, but this is just because of their TAE merger. LITE - Large BOM from Google TPU rollout, attractive valuation. Slight selloff after Google TPU revised est. but it's basically in every single hyperscaler asic deployment. AMZN - one of the mag7 that's not overvalued MRVL - Selloff from analyst misinformation, strong buy going into 2026. Especially with msft maia revenue doubling Marvell's current revenue when it ramps up AVGO - Like NVDA just strong long, as AI infrastructure deployment ramps up OSS - I made a post speculating that they're one of andruils' suppliers. but regardless, edge computing will be hot 2026 and its 180m mc presents attracctive upside. BULL - similar to robinhood where they have a huge userbase, but they just need to figure out monetization Oracle - Sold off a bit too much imo. I put this on avoid months ago but after the from from $330 to $190, it's more attractive again esp. after openai raised another $40B CRDO -extremely high margin, necessary connectivity for dc rollout ALAB - extremely high margin, necessary connectivity for dc rollout Avoid: There's a lot of stuff on the "overvalued list" like $RKLB that i like but I wouldn't quite say avoid it either aside from these. RGTI , QBTS, RGTI - Quantum names are still overvalued and likely won't deliver fcf in the next few ytears. BMNR, ETH - if you saw my eth post, not exactly bullish since the amount of ETH burn is just single-low double digits every day, which is a joke. PLTR - one of the most overvalued ai names WMT - How is this 40 p/e? This is Walmart? __ (these are based on today's prices) TLDR: IPO names like Circle, Etoro, AIRO, Klarna, Figma, present attractive upsides post drop + tax harvesting going into 2026. Tons of names like SMCI, HIMS that dropped 40% or so past 3 months, are amazing swing/recovery trades post-tax harvest + Jan effect. Lot of the names that doom dropped like FiserV or The Trade Desk present good recovery trades too post-tax harvest. Many datacenter stocks like nebius, iren, cifr, wulf, galaxy, are amazing recovery trades too. Lot of other segments like memory, bottlenecks, photonics, and others are just great longs in 2026, despite each hitting ATHs. There's still quite a lot of overvalued names from Quantum, to certain Space stocks (eg. planet or rocketlab), specific AI names like Palantir to retail stocks like Walmart that I would probably avoid for the time being until there's a slight correction. This was a TLDR just if I'm short term trading-only (not long term) but feel free to ask questions.
-
新云板块处于验证期，当前下跌是建立护城河前的黄金买入机会。
新云(Neoclouds)正处于“证明实力”阶段。 $NBIS 的年经常性收入(ARR)增长率高达700%+，达到70-90亿美元；$IREN 的数据也超过34亿美元。 许多公司拥有多年的收入可见性，并得到七大科技巨头(Mag7)超大规模云服务商的背书。 然而，市场表示： “这不是一个无限刷钱的漏洞”。 即股价上涨 -> 可转换票据/稀释 -> ARR增加 -> 循环往复。 它们现在都有资金，$NBIS 持有48亿美元以上的现金储备，$IREN 持有来自 $MSFT 预付款/票据的数十亿美元以完成建设并转化为自由现金流(FCF)。 尽管有《加速推进电气化法案(SPEED Act)》通过、OpenAI融资（降低交易对手风险）以及降息等顺风因素，但近期的下跌趋势似乎是年底税务收割、做空、主动式自动取款机(ATM)增发以及主要是： 等待证明这些公司能否从 $MSFT 与 IREN 的交易到 Nebius 的杠杆内部收益率(IRR)预测中，实现规模化的利润率。这似乎归结为执行力和等待下一次财报。 许多散户投资者在此期间似乎已经投降，但机构持股比例仅上升（$NBIS 从30%多升至50%以上）。 但非对称性就在这里： 如果 Nebius 管理层能实现20-30%的息税前利润(EBIT)利润率并达到70-90亿美元的 ARR 目标，那么在保守分析师给出200美元以上目标价(PT)的情况下，85美元的价格显得极其、极其低估。特别是考虑到 Clickhouse 的扩张/IPO 以及 Avride 与 Uber 合作的自动驾驶出租车规模化。 我在主题上也特别看好 $IREN 在微软交易上的杠杆 IRR 预测，以及来自 $CIFR / $WULF 与 $GOOGL 交易的新云数据中心(Colo)玩家。 简而言之：如果新云能在其预测的利润率、规模上执行到位，并在这2-3年的窗口期内建立自己的护城河，那么整个板块的抛售似乎是一个黄金机会。
英文原文
Neoclouds are in the "Prove It" phase. You have absurd 700%+ ARR growth rates to $7-9B from $NBIS and $3.4B+ figures from $IREN. Many have multiple year revenue visibility and backstop from Mag7 Hyperscalers. However, markets have said: "This is not an infinite money glitch". Where a stock goes up -> convertible note/dilution -> ARR increases -> repeat. They all have funding now, $NBIS sitting on a $4.8B+ cash stack, $IREN sitting on billions from $MSFT prepayment/notes to finish their buildout and turn that into FCF. Despite tailwinds from the initial SPEED act passing, OpenAI fundraising (for less counterpaty risk), and rate cuts, the recent downtrend seems to combine EoY tax harvesting with short selling, active ATMs, and mainly: Waiting for proof that these companies can deliver margins at scale from levered IRR projections on $MSFT's IREN deal to Nebius. It seems to comes down to execution and waiting for their next earnings report. Many retail investors seem to have capitulated during this time but institutional ownership has only gone up (30's from $NBIS to 50's+ now) But here's where the asymmetry comes in: If Nebius management scales to 20-30% EBIT margins with their $7-9B ARR target, this seems incredibly, incredibly off at $85 when conservative analysts are throwing out $200+ PTs. Especially considering possible Clickhouse ramp/IPO and Avride robotaxi scaling with Uber. I'm especially bullish thematically too with $IREN levered IRR projections on the Microsoft deal and Neocloud colo players from $CIFR / $WULF $GOOGL deals. TLDR: If Neoclouds can execute with their projected margins, scale, and create their own moats during this 2-3 year window, then the whole sector selloff seems like a golden opportunity.
-
分析InP材料供应瓶颈，通过历史案例说明关键材料博弈论，预判超大规模云商将抢购InP衬底产能。
当我发WSB帖子时，我半开玩笑地提到$AXTI。但对于极端瓶颈和InP供应冲击，我可不是开玩笑的。这是一个关于关键材料博弈论的呼唤。 如果我们回顾历史上许多瓶颈案例： 1. 镝(Dysprosium)：约2300%涨幅（约$100/kg -> $2400/kg，2010-11年） 2. 氖气(Neon Gas)：约1000%-2000%涨幅（2022年） 3. 铑(Rhodium)：约$3,000/盎司 -> 约$29,000/盎司（约860%） 当前铟(In)的价格涨幅与其他材料相比如何？ 89%：（约$440/kg -> 约$832/kg）。 用于半导体的关键材料（如用于光刻激光器的氖气），因为芯片制造商如英特尔和三星在乌克兰冲突期间进行了储备，价格暴涨超过2000%。他们吸收了成本，因为氖气只占芯片价值的一小部分。 与氖气一样，磷化铟(InP)是一种关键的"赋能"材料。晶圆成本相对于AI集群成本（$100的晶圆对比$30,000的GPU）来说很小。超大规模云商宁愿吸收巨大的价格上涨，也不愿延迟部署。InP目前还没有出现这种情况，但我们很可能会看到。 传统分析师和许多AI模型不知道如何建模关键瓶颈，因为历史上很少有这样的案例：一种极其廉价的商品（如InP，TAM只有几亿美元）突然成为价值数万亿美元的AI建设中最关键的材料。 目前，超大规模云商可能已经向下游供应商Coherent/$LITE下了订单，但可能没有意识到激光器生产和产能受到上游材料短缺的制约而排满。 InP衬底（$AXTI + Sumitomo + JX生产）和磷化铟原料（$AXTI、Vital、$DOWA）都存在供应储备，需求超过供应数倍。 $NVDA最初通过锁定EML大量产能加剧了供应短缺。但短缺早在超大规模云商扩产之前就已经存在，大部分需求将在2026年底至2027年Trainium或Maia出现时涌入。 但如何绕过原始瓶颈并确保AI部署？ 直接从（$AXTI、Sumitomo）购买衬底产能，或者更进一层，购买InP原料并将其交给衬底生产商，以确保谷歌的TPU项目不会停滞。 与其他关键材料（如氖气用于光刻激光器）类似，磷化铟和InP衬底对整个AI建设更为关键： 1. Nvidia：InfiniBand / NVLink（光学 - 800G/1.6T EML光模块）- 对InP高度依赖 2. Google：Jupiter / Apollo（光学 - OCS），高度依赖 3. Microsoft：Azure Maia：高（800G DR4/FR4） 4. Meta：F16 / Artemis（800G）高 这还不包括其他Mag7公司如$AMZN或使用光子技术扩展ASIC的其他公司。 目前，分析师基于TAM对InP公司进行建模。例如，2024年InP晶圆市场总规模为$183M-$205M（Straits Research, Mordor Intelligence，2024年） - 6英寸（150mm）细分市场TAM：$65M - 预计2032年InP总TAM：$580M-$700M 现在看激光级磷化铟原料的TAM： - 多晶InP原料TAM：约$3亿 - 高纯度铟（6N/7N）仅：约$4亿 衬底数据：Mordor Intelligence（InP晶圆市场2025-2030）和Straits Research。 原料数据：Market Report Analytics（磷化铟多晶行业分析2025）。 这是一种极其便宜的电信产品，现在成为整个供应链中最珍贵的材料之一。 价格上涨3000%只会占$GOOGL TPU部署或$MSFT MAIA部署BOM值的低个位数。而仅微软2027年的扩产就会占用全部InP产能的两位数百分比。 就像2022年的英特尔或半导体行业一样，超大规模云商可能会像当时囤积氖气一样，直接购买InP衬底+InP作为"保险"。 如果"TAM"实际上只有几亿美元，但每个超大规模云商如果不确保这种材料就会面临部署延迟，那么这就变成了一场博弈论竞标战。 在一个因$100衬底而停滞的价值$200亿的TPU部署中，超大规模云商会毫不犹豫地支付$10,000（比当前价格高出100倍）。 这就是为什么当你看到InP衬底公司有$5000万订单积压（来自2024年），同样的$5000万产能对未来来说在极端情况下可能价值$5亿、$50亿甚至更多。 这是一个"尾部风险"投资，如果超大规模云商被这种材料卡住，回报是极高的，但并非没有重大的地缘政治风险和稀释。 我不知道会发生什么，但我预计2026年InP衬底和InP原料都会出现巨大瓶颈。 那么，如果像$AXTI这样的公司拥有整个InP供应链40%的份额（来源：AXT CEO），而估值只有$7亿，你会给它什么估值？
英文原文
When I made my WSB post, I'm half joking about $AXTI. But not about the extreme bottleneck and InP supply shock. This is a call on critical materials game theory. If we see many historically at bottlenecks: 1. Dysprosium: ~2,300% increase (~$100/kg -> $2400/kg 2010-11) 2. Neon Gas: ~1000%-2,000% increase (2022) 3. Rhodium: ~$3,000/oz -> ~$29,000/oz (~860%) The current Indium spike compared to others? 89% : (~$440 / kg -> ~$832 / kg). Critical materials used for semiconductors like Neon Gas (semi-grade lithography lasers), spiked over 2000% because chipmakers like Intel and Samsung had stockpiled during Ukraine conflicts. They absorbed the cost because neon is a small fraction of the chip's value. Like Neon, InP is a critical "enabler" material. The cost of the wafer is small relative to the cost of the AI cluster ($100 wafer vs $30,000 GPU). Hyperscalers would happily absorb huge price increases rather than delay deployment. We haven't seen that yet with InP, but we likely will. Traditional analysts and many AI Models do not understand how to model critical bottlenecks because there's not many examples in history where an extremely cheap commodity like InP with few hundred million TAM suddenly became the most critical material for the multi-trillion+ AI buildout. Currently, hyperscalers likely placed backlogs on downstream providers Coherent / $LITE, but likely didn't realize that laser production and capacity are backlogged because of materials shortage upstream. There is a supply storage both for InP substrates ( $AXTI + Sumitomo + JX production) as well as the Indium Phosphide feedstock ( $AXTI, Vital, $DOWA) where demand exceeds supply by multiple factors. $NVDA originally contributed to the supply shortage by locking in a large percent of capacity of EML. But shortage is already pre-hyperscaler ramp, where majority of the demand will come in late 2026 into 2027 when Trainium or Maia show up. But how do you get around original chokepoint and securing your AI deployment? Buying out substrate capacity directly from ( $AXTI, Sumitomo) or going one step lower and buying InP feedstock and passing them to substrate producers so Google's TPU program doesn't stall. And similar to how other critical materials were needed for lithography lasers. Indium Phopshide and InP substates are even more critical to the entire AI buildout: 1. Nvidia: InfiniBand / NVLink (optical - 800G/1.6T EML Transceivers) - Extreme dependency on InP 2. Google: Jupiter / Apollo (optical - OCS), Extreme dependency 3. Microsoft: Azure Maia: High (800G DR4/FR4) 4. Meta: F16 / Artemis (800G) High and this is not considering other Mag7 companies like $AMZN or other companies scaling out ASICs with photonics. Currently, analysts are modeling InP companies based on TAM. For example, the total InP Wafer Market from 2024 is $183M – $205M Straits Research, Mordor Intelligence (2024) - 6-inch (150mm) Segment TAM: $65m - Projected Total InP TAM (2032): $580M – $700M Now looking at TAM for Laser-Grade Indium Phosphide Feedstock: - Polycrystalline InP Feedstock TAM: ~$300 Million - High-Purity Indium (6N/7N) Only: ~$400 Million Substrate Data: Mordor Intelligence (InP Wafer Market 2025-2030) and Straits Research. Feedstock Data: Market Report Analytics (Indium Phosphide Polycrystalline Industry Analysis 2025). This was an extremely cheap telecom product and now it's one of the most precious materials in the entire supply chain. Increasing prices 3000% would only be low single digits of BOM value to $GOOGL TPU deployments or $MSFT MAIA deployments. And the project Microsoft ramp in 2027 alone would take up double digits of all InP capacity. Like Intel or semis back in 2022, hyperscalers will likely buy insurance directly with InP substrates + InP, similar to how semis stockpiled Neon Gas. If the "TAM" really was a few hundred million, but every single hyperscaler would face deployment delays if they don't secure this material. Then this becomes a game theory bidding war. In a $20B TPU deployment is stalled because of a $100 substrate, a hyperscaler would easily pay $10,000 (100 times current prices) for it without hesitation. That's why when you look at InP substrate companies and see $50m backlog (from 2024), that same $50m worth of capacity for future ones could be valued at $500m, $5B or more in an extreme scenario. This is a "tail-risk" investment, extremely reward if hyperscalers get bottlenecked by this material, but not without significant geopolitical risk and dilution. I don't know what's going to happen, but I do expect there to be an immense bottleneck in 2026 both in InP substrates and InP feedstock. So, if companies like $AXTI owns 40% of the entire InP supply chain (source: AXT CEO) and is valued at $700m, what value would you place on it?
-
认为 Neocloud 板块在长期合同去风险后仍被错杀
同意。 像 $CIFR 这样的公司，现在交易得仿佛 $AMZN、$GOOGL 的合同从未发生过。 在整个 Neocloud 板块里，我们看到的是一场崩盘，尽管 Mag7 合同已经让全板块去风险，并提供了长期收入可见性。 $ORCL、$CRWV 以及一些个股因为 OpenAI 敞口而下跌是有道理的，但它把板块其他公司也一起拖下去了。
英文原文
Agreed. Companies like $CIFR are trading like the $AMZN, $GOOGL deals never happened. In the whole Neocloud sector, we’re seeing a crash despite de-risking and long term revenue visibility across the board with Mag7 contracts. $ORCL, $CRWV, and some individual names made sense given OpenAI exposure but it just dragged the rest of the sector alongside it.
-
惊讶 NBIS 和 IREN 回到 Mag7 合同前更低水平
真没想到我还能再看到 $NBIS 到 80 美元，或 $IREN 到 35 美元。 它们现在的交易价格，比最初宣布 Mag7 合同的时候还低很多。
英文原文
Never thought I’d see $NBIS at $80 or $IREN at $35 again. They’re both trading a lot lower than when they first announced Mag7 contracts. https://t.co/boqDYeyXzL
-
AVGO财报被误读，AI逻辑未变，大跌提供买入良机。
我对 $AVGO 财报的解读是，市场主要误解了其积压订单(backlog)数据。关于 $NVDA GPU 生态系统和 Mag7 ASIC 生态系统的 AI 交易整体逻辑依然成立，但市场却将其视为泡沫破裂+严重不及预期。我们正经历从“AI是下一个大趋势”到“AI是泡沫”，再到“AI是下一个大趋势”，最后又回到“AI是泡沫”的循环。波动性和13%以上的跌幅使得多只股票成为良好的买入机会。
英文原文
So my takeaway from $AVGO earnings is that it's mainly misunderstood backlog numbers. Overall thesis on AI trade from $NVDA GPU ecosystem and Mag7 ASIC ecosystem is in-tact but the market is treating it like a bubble popping + terrible miss. We're just going from AI is the next big thing -> Ai is a bubble -> AI is the next big thing -> AI is a bubble. Volatility and -13%+ drops makes several names a good buy.
-
博通财报后AI板块抛售是误解造成的买入机会，新云板块中OpenAI依赖股除外
博通[$AVGO]业绩及其对AI板块的影响，如$LIITE和$NBIS： 博通的业绩"双重超预期"，营收$180.2亿(+28% Y/Y)，EPS $1.95，超出共识预期。 但AVGO下跌-11.64%，并拖累了整个AI板块。 这是买入机会吗？ 是的。 博通被视为超大规模云厂商ASIC代理增长的代表，因为亚马逊$AMZN Trainium、微软$MSFT Maia、尤其是谷歌$GOOGL TPU V7 Ironwood都通过它进行规模化部署。 而像$ALAB(-13.2%)、$CRDO(-5.11%)、$LITE(-12.23%)、$TSM(-3.71%)、$COHR(-9.25%)等公司都是TPU/ASIC建设以及博通作为公司的直接受益者。 博通下跌有三个原因，市场下跌有一个原因： 就博通而言，有一些小问题，如税率变化影响EPS模型，或因更多定制AI芯片而非更高利润率软件导致的"利润率压缩"，但这只是会计处理框架问题。（类似于$META在一次性税收后最初的下跌） 对博通和整体市场而言，是积压订单预期问题。以上所有引用的内容与ASIC市场预期增长相比都是小问题。 博通披露未来18个月$730亿的AI积压订单。而有关Anthropic和META购买价值数十亿$GOOGL TPU的传言，人们隐含预期是$800亿+。 然而，这次抛售是由算法和短期AI泡沫情绪驱动的价格错位，而非基本面破裂。 这条积压订单引用是确认订单的最低合同底线。谷歌$GOOGL、亚马逊$AMZN等公司可能会继续增加ASIC订单，而市场未能辨别这一细微差别。 分析师预期营收转化会更加前置，Q4之后积压订单应该会减少，这给出了2026年更高的可能范围$550-600亿+，而非$730亿预期中的$500亿。 TLDR：关于超大规模云厂商ASIC加速以与$NVDA依赖竞争这一论点没有改变。$AVGO和其他如$COHR、Sk Hynix、$MU、$VRT和$LITE都将受益。 这不是关于营收积压订单的最佳消息，但由于交付周期/订单周期和最低底线而被误解。 如果非要说什么，较低的超级云厂商ASIC需求对$NVDA及其生态系统是有益的，但我们也看到$CRWV、$SMCI、$NBIS和$NVDA GPU/DC计算生态系统今天都从盲目抛售中下跌超过5%，尽管存在负相关性。 这又是典型的"AI泡沫"周期因误解而再次来袭。AI股票普遍下跌10-12%的恐慌是一个绝佳的买入机会。 （该推文引用了@aleabitoreddit的推文，引用内容仅供理解语境）: 甲骨文[$ORCL]业绩及其对新云板块如$NBIS和$IREN的影响： 甲骨文报告EPS超预期且积压订单创纪录，但盘后下跌12%。 甲骨文较9月11日高点下跌39.8%，并拖累整个板块。 原因如下： 这次抛售不仅仅是对边际营收miss的反应，而是算法做空和投资者对AI资本支出周期可持续性以及该板块主要租户信用资质的不看好： OpenAI。 甲骨文宣布2026年资本支出增加$150亿至近$500亿，这与报道的与OpenAI $3000亿合作伙伴关系密不可分。 最初，OpenAI是前沿LLM，对甲骨文、Coreweave等公司有前景良好的资本支出承诺，促成了该板块的初始重新定价。 然而，随着超过$1万亿的义务以及Anthropic、Gemini、XAI等竞争对手的增加，市场严重质疑甲骨文、Coreweave等公司是否为无法从运营现金流履行其义务的租户在建设。 我们看到市场有效地发出信号：甲骨文正在为OpenAI创造一种不可持续的债务驱动型"供应商融资"，而OpenAI无法履行其承诺。 因此下跌是理性的：这次抛售是由信用风险和资本密集度的理性重新定价所驱动。 OpenAI融资担忧是合理的：OpenAI缺乏资金履行合同的假设得到其营收($130亿)与义务($600亿/年)之间明显不匹配的支撑。 信用担忧是真实的：甲骨文CDS利差扩大显示"信用事件"降级或违约的概率上升。 此外，我们看到这在新云板块引发传染效应，$NBIS从$140跌至$90s，$IREN从$80跌至$40s，$CIFR从$24s跌至$17s。 但这对$WULF、$NBIS、$IREN等新云公司来说是买入机会吗？ 是的。 这对$ORCL来说是好的买入机会吗？ 不是。 前瞻展望： $ORCL（很大一部分）、$CRWV（25%积压订单）是两个主要依赖OpenAI的公司，这一叙事可能因OpenAI的融资活动而瞬间翻转（+30%+变化）。 如果OpenAI在2026年以高估值超额认购IPO，且其新GPT模型击败Gemini/Claude，我们可以看到这种改变。 然而，许多其他公司与OpenAI无关。新云板块的原始论点是Mag7资本支出从其现金牛业务（Azure、AWS、GCP）向下流向：$NBIS、$IREN、$CIFR、$WULF等。 但随着最大玩家($ORCL、$CRWV)下跌，这些公司算法性地拖累了整个板块。 如果你看各家公司，$CIFR和$WULF由$GOOGL兜底，$IREN/$NBIS由$MSFT资助。 这些是与超大规模云厂商/Mag7的锁定合同积压订单，而非OpenAI。 这种因误解风险而导致的不理性抛售为新云板块提供了绝佳的买入机会，但不是与OpenAI相关的公司如$ORCL和$CRWV。
英文原文
Broadcom [ $AVGO ] earnings results and its effect on the AI sector like $LITE and $NBIS: Broadcom's ER was "double beat" with $18.02B revenue (+28% Y/Y) and $1.95 EPS, beating consensus. But AVGO dropped -11.64% and brought down the AI sector. Is this a buying opportunity? Yes. Broadcom is seen as a hyperscaler ASIC proxy growth as companies like $AMZN Trainium, $MSFT Maia, and most importantly $GOOGL TPU V7 Ironwood are scaled through it. And by proxy companies like $ALAB (-13.2%), $CRDO (-5.11%), $LITE (-12.23%), $TSM (-3.71%), $COHR (-9.25%), and are direct beneficiaries of the TPU/Asic buildout and Broadcom as a company. There's three reasons why Broadcom fell and one why the market fell: For Broadcom, there's minor things such as tax rate changing EPS models or "margin compression" from accounting from just more custom AI chips than higher-margin software, but this is just accounting framing. (Similar to how $META dropped initially on one-time tax post-ER) For both Broadcom general market, it was backlog expectations. Everything cited above is all minor compared to expected growth of ASIC markets. Broadcom cited $73B in AI backlog for the next 18 months. And rumors of Antrophic and META buying billions of $GOOGL TPUs, people were implicitly expecting $80B+. However, the selloff represents a dislocation in price driven by algorithms and short-term AI Bubble sentiment rather than a fundamental breakage. This backlog quote was the MINIMUM CONTRACTUAL FLOOR of confirmed orders. Companies like $GOOGL, $AMZN, will likely continue ramping up ASIC orders and the market failed to discern this nuance. Analysts are expecting revenue conversion to be more front loaded, and that there should be less backlog beyond Q4 given the cycles, which gives a higher likely range of $55-60B+ for 2026 rather than $50B expected of the $73B. TLDR: The thesis regarding hyperscaler ASIC ramp to compete vs $NVDA dependency has not changed. $AVGO and other players like $COHR, Sk Hynix, $MU, $VRT, and $LITE all stand to benefit. It's not the best news regarding the revenue backlog, but it's misunderstood due to lead-time/order cycles and minimum floors. If anything, lower hyperscaler ASIC demand is beneficial to $NVDA and their ecosystem, but we've also seen $CRWV, $SMCI, $NBIS and $NVDA GPU/DC compute ecosystem drop over 5%+ today from an indiscriminate sell-off despite inverse correlation. This is just the typical "AI Bubble" cycle hitting again from misunderstanding. The widespread panic of AI stocks dropping 10-12% is a great buying opportunity.
-
解析Mag7 ASIC供应链，看好$AAOI因$MSFT散热互连修复及$AMZN订单带来的增长。
不客气！如果我们看Mag7的ASIC，$GOOGL TPU v7 Ironwood的供应商在$META交易后曾有一波上涨。 $AMZN Trainium即将像$GOOGL那样开始放量，近期超大规模云厂商$AAOI的订单很可能来自他们。 $AAOI的超额收益逻辑在于：$MSFT Maia 200的订单更多是研发支出，且他们推迟了路线图（可能由于散热和互连问题，例如连接10万颗芯片而不熔化），标准线缆无法用于Maia 200。 但这对$AAOI是利好，因为$MSFT很可能使用他们来解决这一问题，且所有这些ASIC支出将发生在2026-2027年，加上远期收入指引的大幅超预期。 但$AMZN提供了约40亿美元的安全垫以降低风险。
英文原文
NP! If we look at mag7 ASICs, $GOOGL TPU v7 Ironwood suppliers had a run after the $META deal. $AMZN Trainium is going to ramp up soon like $GOOGL and the recent hyperscaler $AAOI order was likely from them. The alpha for $AAOI is that $MSFT Maia 200 orders were more of R&D spend and they delayed their roadmap (likely due to thermal and interconnect issues eg. 100,000 chips to connect without melting) and standard cabling wasn't working for Maia 200. But this is bullish for $AAOI since $MSFT likely uses them for this fix and all of this ASIC spend will be occur in 2026-2027 + huge beat on forward revenue projections. But $AMZN is providing the nice $4B floor for de-risking.
-
分析AAOI依赖Mag7的垂直整合优势，已建仓并看好LITE。
我原本在研究一篇来自 Semianalysis 的文章中关于 $AMZN Trainium 基准测试以及亚马逊供应链的内容。鉴于一些朋友（如 @yianisz）之前曾对此发帖讨论，我决定深入挖掘。亚马逊在扩大 Trainium/Inferentia 集群时提供了约 40 亿美元的支出作为后盾，加上 $MSFT 在 ASIC 集群上的投入，随着行业向 800G 升级，我们很可能会看到更多超大规模云厂商客户。对于一家市值 25 亿美元的公司来说，如此依赖 Mag7 的情况实属罕见 lol。下行风险与积极因素并存：由于激光器在德克萨斯州内置生产，该公司实现了垂直整合（更高利润率）。我看好未来巨大的量增需求，但：关键在于执行，他们必须在不搞砸的情况下建成 800G/1.6T 光收发器。我已经全仓持有 $AAOI，出于风险管理考虑，仓位规模相对较小适中（毕竟它仍是 25 亿市值的小盘股）。我依然更看好 $LITE，但鉴于其本月已上涨 46%，很难进行补仓。
英文原文
I was actually looking into $AMZN Trainium benchmarks from a semianalysis post and at Amazon's supply chain. And some friends like @yianisz post about it earlier, so I decided to look deeper at it. But it has ~$4B spend from $AMZN as a backstop as they ramp up Trainium/Inferentia clusters, spend from $MSFT ASIC clusters, and we'll likely see more hyperscaler clients as the industry moves to 800g ramp. Mag7 dependency is incredibly rare coming from a $2.5B marketcap company lol. Downside + positive: Is vertically integrated (higher-margins), since it's lasers are builtin Texas. I see just incredible volume demand, moving forward but: the whole name of the game is execution since they actually have to build out the 800G/1.6T transceivers without messing up. I'm fully invested already in $AAOI already with relatively small-moderate position sizing for risk management (it's still a small cap with $2.5B MC). I still like $LITE a lot more, but a bit hard to cost average up on that when it's up 46% this month.
-
看好$AAOI，因其在AI光互联中的核心地位及美国本土制造优势，认为其被严重低估。
我买入 $ALAB、$NBIS、$TSM 和 $LITE，是因为“七巨头”(Mag7) 的收入正流向这些公司。 $LITE 的独特之处在于其在 GOOGL TPU v7、AMZN Trainium v3/4 和 NVDA Blackwell 中的角色。 但我发现了一个新标的。 一家市值低于 30 亿美元的小盘股，符合我的投资逻辑： 名称 - $AAOI 一家小型光子学玩家，也是我持仓的两家光子学公司之一： 1. Lumentum ($LITE) 在每一款芯片的部署/爬坡中都占据独特地位，因为其光电路交换(Optical Circuit Switching) 技术被用于 Blackwell、Trainium 和 Ironwood，作为一种“横向扩展”(scale-across) 型技术。 无论超大规模云厂商选择 ASIC 还是 GPU，$LITE 都能胜出，因为它处于核心位置。 2. Applied Optoelectronics ($AAOI) 更多涉及“纵向扩展”(scale-out) 连接，针对 Trainium、Maia 等定制 ASIC 集群，通过 400G 和 800G 光缆及光收发器实现。 鉴于 $AAOI 在 AWS（Trainium 的大客户）和 MSFT Maia ASIC 中的角色，它也能无论何种情况都胜出。 行业正经历由向 800G 速度迁移驱动的“超级周期”，而 AAOI 正处于中心位置。 此外，AAOI 具有独特的地缘政治角度： 美国优先。 与许多无晶圆厂(fabless)并将生产外包到台湾的公司不同，Applied Optoelectronics 在德克萨斯州自行制造激光器。 美国超大规模云厂商（特别是 Amazon 和 Microsoft）正在积极减少对关键基础设施中国供应链的依赖，这有助于 AAOI 实现本土激光器制造。 我们还看到一家“主要超大规模云厂商”对其 800G 数据中心光收发器下达了另一笔巨额订单。 但 AAOI 的交易表现像一家困境公司；然而，其亚马逊权证协议隐含的收入为 2025 年 Q4 和 2026 财年的 800G 爬坡创造了非对称的风险/回报特征。鉴于其在 AI 建设中的角色及现有的超大规模云厂商合同，AAOI 似乎结构性被低估。 市场终于开始重估 $LITE，但感觉 $AAOI 才刚刚开始，鉴于其小市值以及作为超大规模云厂商 ASIC 集群关键玩家和“美国制造”的独特角度。 市场目前正在对光子学玩家进行大幅重估并赋予高溢价，但 AAOI 今年仅上涨 2.20%，似乎蓄势待发。
英文原文
I entered $ALAB, $NBIS, $TSM, and $LITE because of Mag7 funneling revenue numbers into them. Lite uniquely because of its role in GOOGL TPU v7, AMZN Trainium v3/4, and NVDA Blackwell. But there's a new one I found out about. A small cap <$3B player that fits the thesis: Name - $AAOI A small cap photonics player, and one of the two photonics players I'm invested in: 1. Lumentum is uniquely positioned in every single supply chip deployment/ramp, as the Optical Circuit Switching technology is used in Blackwell, Trainium, and Ironwood as a "scale-across" type technology. $LITE wins no matter what. Hyperscaler ASIC vs. GPU as it's in the center of it all. 2. Applied Optoelectronics is more of the "scale-out" connectivity, for custom ASIC clusters like Trainium, Maia through 400G and 800G optical cables and transceivers. $AAOI wins no matter what as well given its role with AWS as a whale client for Trainium, and with MSFT Maia ASICs. The industry is going through a "supercycle" driven by the migration to 800G speeds and AAOI is in the center of it. On top of that, AAOI plays unique geopolitical angle, America first. Unlike many other companies that are fabless and export production elsewhere to Taiwan, Applied Optoelectronics makes their own lasers in Texas. US hyperscalers (specifically Amazon and Microsoft) are aggressively reducing reliance on Chinese supply chains for critical infrastructure and that helps AAOI’s ability to manufacture lasers at home. And we've seen another huge volume order from a "major hyperscaler" on its 800G data center transceivers. But AAOI trades like a distressed company; however, the implied revenue of its Amazon warrant agreement creates an asymmetric risk/reward profile on its 800G ramp in Q4 2025 and FY2026. AAOI seems structurally undervalued, given its role in the AI buildout and existing hyperscaler contracts. The markets are finally catching up to LITE, but it feels $AAOI is yet to begin, given its small market cap size but unique angle of a critical player to hyperscaler ASIC clusters and Made in America. The market is currently re-rating heavily with photonics players and assigning a heavy premium, yet AAOI is only up 2.20% this year and seems like it's just about to begin.
-
OpenAI优势减弱，AI行业多极化，基本面未受损公司现买入机会
OpenAI 此前是明确的领导者。现在，Gemini 3.0 在图像生成方面超越了它，Claude Opus 4.5 在编码方面超越了它。以此类推。看起来在获得先发优势后，ChatGPT 似乎无法被颠覆。但现在从 Similarweb 数据我们可以看到，Gemini 的受欢迎程度相对于 ChatGPT 正在增长。而且我相信由于使用内部张量处理单元(TPU)进行推理，它们的成本效益也更高。OpenAI 试图通过要求 Mag7 和政府为其合同承诺提供担保，使整个行业依赖它，以便大家同船沉没，但这最终失败了。整个 AI 行业不仅仅是 OpenAI，随着我们看到更多的代理(Agentic)和机器人应用，我们很可能会看到 Anthropic、Google、XAI 和其他供应商的广泛大型语言模型(LLM)和日益增长的算力使用。但目前我们看到这种传染效应从 OpenAI 蔓延到该领域的几家公司 $ORCL、$CRWV，可能还有 $AMD 以及其他几跳远的公司，它们正在为承诺的资本支出构建产能。但这正在拉垮那些与 OpenAI 隔离（就股价而言）的其他公司，因此如果基本面没有受到实质性影响，这就是一个买入机会。
英文原文
OpenAI was the clear leader before. Now, Gemini 3.0 surpasses it in image generation. Claude Opus 4.5 surpasses it in coding. Can go down the line. It looked like ChatGPT couldn’t be disrupted after they got first movers advantage. Now from similarweb data we can see popularity with Gemini growing vs ChatGPT. And im sure they’re a lot more cost effective as well using in house TPUs for inference. OpenAI tried making the whole industry dependent on it, by asking Mag7 + government to backstop their contract promises so they all go down the ship together but that ended up flopping. AI industry as a whole is not OpenAI, we’ll likely see widespread LLM + growing compute usage from Anthropic, Google, XAi and other vendors as we see more agentic + robotics applications next. But right now we’re seeing contagion spread from OpenAI to several companies in the sector $ORCL, $CRWV, maybe $AMD and others few hops away that were building out capacity for their promise capex spend. But it’s taking down other companies isolated from OpenAI (in terms of stock price), which is why it’s a buying opportunity if fundamentals aren’t too materially impacted.
-
对比NBIS与CRWV债务风险，指出市场忽视个体差异盲目联动。
是的，这是个很好的例子。如果你查看 $NBIS 和 $IREN 的可转债发行，由于 $MSFT（Magnificent 7）是它们共同的积压订单，两者的利率都在1-2.5%左右。许多新云厂商受到 OpenAI 的影响，甚至像 $APLD 这样通过 $CRWV（主要租户）间接关联两跳的公司，也不得不以9.25%的利率出售垃圾债券。$NBIS 每年支付约7300万美元利息，而 $CRWV 支付超过13亿美元。但市场似乎因 OpenAI 的传染效应而整体联动该板块，未能区分那些被隔离的个别公司。利息债务、风险和客户锚点的差异是巨大的。
英文原文
Yep that’s a good example. If you look at convertible offerings for $NBIS and $IREN both their interest rates were 1-2.5% roughly because $MSFT (mag7) is their shared backlog. A lot of Neoclouds are affected by OpenAi though and even by two hops like $APLD from $CRWV(main tenant), which had to sell junk bonds at 9.25%. $NBIS is paying ~$73m/year from interest, while $CRWV is paying upwards of $1.3B. But the market seems to move the whole sector together from OpenAI contagion without discerning individual companies that are isolated. The difference in interest debt, risk, and customer anchors is massive.
-
板块错杀提供买入机会，NBIS/IREN无交易对手风险且利用率高。
超额收益(alpha)在于了解哪些类型的合同能让人免受当前恐惧的影响，以及行业内哪些个别组件存在定价错误。$MSFT、$META 等公司的算力容量合同是“照付不议”(take or pay)的。 $NBIS、$IREN 等公司由于与“七巨头”(Mag7)（拥有无限资产负债表）签订5年期合同，实际上消除了**交易对手风险**(counterparty risk)。 然而，它们下跌的原因不同（$NBIS 因2500万股的自动行使机制(ATM)发行；$IREN 因市场担忧其为将剩余3GW的AI云容量管道变现而进行稀释，而非用于数据中心托管(colocation)）。 然而，主要的普遍担忧是 $ORCL、$CRWV 面临来自 OpenAI 破产+无力支付的严重交易对手风险，但由于它们是两大主要玩家，这引发了整个板块的算法抛售。 但这种板块抛售为一些未受影响（除信贷收紧外）/被误解的公司提供了良好的买入机会。 此外，AI是一个增长的市场，目前可能存在算力紧张（例如 anthropic、gemini 和其他模型），但 $NBIS、$IREN 在超大规模云服务商合同上的利用率基本为100%。 显然，如果存在过度建设，可能会导致利润率压缩，但我们目前尚未看到这种情况。5年后会发生什么，我不知道。
英文原文
The alpha is knowing what types of contracts lead to isolation from current fears and where there's mispricing on individual components in the sector. Compute capacity contracts are take or pay for $MSFT, $META, and others. Companies like $NBIS, $IREN and others are effectively de-risked due to **no counterparty risk** from Mag7 (infinite balance sheets) and 5 Year contracts. However, their drops were different reasons ( $NBIS, 25M share ATM), $IREN fears over dilution for monetizing their rest of the 3 GW capacity pipeline for AI Cloud instead of colo. However, the main overarching fear was that $ORCL, $CRWV faces severe counterparty risk from OpenAI insolvency + inability to pay, but this causes a sector algorithmic selloff because they're the two largest players. But this sector sell-off is a good buying opportunity for some of the unaffected (minus credit tightening) / misunderstood companies. Also AI is a growing market, and there's likely compute strain for the time being (eg. anthropic, gemini, and other models) but it's basically 100% utilization for hyperscaler contracts on $NBIS, $IREN. Obviously if there's overbuild, there's probably margin compression but we're not seeing that right now. What happens after 5 years I don't know.
-
甲骨文因OpenAI信用风险大跌，新云板块错杀，建议买入非OpenAI依赖标的。
甲骨文($ORCL)的财报结果及其对$NBIS和$IREN等“新云(Neocloud)”板块的影响： 甲骨文财报EPS超预期且积压订单创纪录，但盘后仍下跌12%。 甲骨文自9月11日高点以来已下跌39.8%，并拖累了整个板块。 原因如下： 抛售不仅是对营收轻微不及预期的反应，更是算法做空和投资者对AI资本支出(Capex)周期可持续性，以及该板块主要租户OpenAI偿债能力的担忧。 甲骨文宣布2026年资本支出增加150亿美元至近500亿美元，这与据报道的与OpenAI的3000亿美元合作伙伴关系密不可分。 最初，OpenAI作为前沿大语言模型(LLM)，向甲骨文、Coreweave等承诺了诱人的资本支出，推动了板块的初步重估。 然而，随着超过1万亿美元的债务义务以及来自Anthropic、Gemini、XAI等的竞争加剧，市场严重怀疑甲骨文、Coreweave等是否为一家无法通过经营性现金流履行义务的租户建设基础设施。 市场有效发出了信号：甲骨文正在为无法履行承诺的OpenAI提供不可持续的债务融资“供应商融资(Vendor Financing)”。 因此，下跌是理性的：抛售由信用风险和资本密集度的理性重估驱动。 OpenAI资金担忧是合理的：OpenAI缺乏资金履行合同的假设，由其收入(130亿美元)与义务(600亿美元/年)之间的巨大错配所支持。 信用担忧是真实的：甲骨文信用违约互换(CDS)利差的扩大表明“信用事件”降级或违约的概率上升。 此外，我们看到这在“新云”板块中引发了传染效应：$NBIS从140美元跌至90美元区间，$IREN从80美元跌至40美元区间，$CIFR从20多美元跌至17美元区间。 但这对于$WULF、$NBIS、$IREN等新云公司是否是买入机会？ 是的。 这对$ORCL是否是好的买入机会？ 不是。 前瞻展望： $ORCL(大部分)和$CRWV(25%积压订单)是两大主要依赖OpenAI的玩家，这一叙事可能因OpenAI的融资活动而瞬间反转(+30%以上波动)。 如果OpenAI在2026年以高估值进行超额认购的IPO，且其新GPT模型击败Gemini/Claude，情况可能改变。 然而，许多其他玩家与OpenAI隔离。新云板块的原始逻辑是Mag7从其现金牛业务(Azure, AWS, GCP)向下漏斗资金至：$NBIS, $IREN, $CIFR, $WULF等。 但随着最大玩家($ORCL, $CRWV)下跌，算法交易拖累了整个板块。 如果单独看公司，$CIFR和$WULF由$GOOGL背书，$IREN/$NBIS由$MSFT资助。 这些是来自超大规模云厂商/Mag7的锁定合同积压，而非OpenAI。 这种因误解风险而导致的非理性抛售，为新云板块提供了绝佳的买入机会，但不包括与OpenAI绑定的$ORCL和$CRWV。
英文原文
Oracle [ $ORCL ] earning results and its effect on the neocloud sector like $NBIS & $IREN: Oracle reported earnings with a beat on EPS and a record backlog but, dropped 12% after hours. Oracle is down 39.8% from September 11th highs and brought down the sector with it. Here's why: The sell-off was not merely a reaction to marginal revenue miss, but both an algorithmic short and investor selloff on the sustainability of the AI capex cycle and the creditworthiness of the sector's primary tenant: OpenAI. Oracle's announcement of a $15 billion increase in 2026 capital spending to nearly $50 billion was inextricably linked to a reported $300 billion partnership with OpenAI. Originally, OpenAI was the frontier LLM, with promising capex promises to Oracle, Coreweave and others, contributing to the initial repricing of the sector. However, with over $1t+ in obligations and increasing competition from Anthropic, Gemini, XAI, and others, the markets have serious doubts on whether Oracle, Coreweave, and others are building for a tenant that cannot currently fund its obligations from operating cash flow. WE're seeing the market effectively signaling that the market Oracle is creating an unsustainable debt-funded "vendor financing" for OpenAI, which cannot fulfill its promises. So, the drop was rational: The sell-off was driven by a rational repricing of credit risk and capital intensity. OpenAI Funding Fear is Valid: The hypothesis that OpenAI lacks the funds to honor its contracts is supported by a glaring mismatch between its revenue ($13B) and its obligations ($60B/year). Credit Fears are Real: The widening of Oracle's CDS spreads sees a rising probability of a "credit event" downgrade or default. Furthermore, we're seeing this trigger a contagion effect across the "Neocloud" sector from $NBIS dropping from $140 to $90s, $IREN dropping $80 to $40s, $CIFR dropping from $24s to $17s. But is this a buying opportunity for Neoclouds like $WULF, $NBIS, $IREN, and others? Yes. Is this a good buying opportunity for $ORCL? No. Forward Outlook: $ORCL (large portion) , $CRWV (25% backlog) are the two players largely dependent on OpenAI and this narrative can flip in an instant (+30%+ change) depending on capital raising activity from OpenAI. If OpenAI files for an oversubscribed IPO in 2026 at high valuations and it's new GPT models beats out Gemini/Claude, we can see this change. However, many other players are isolated from OpenAI. The original thesis of the Neocloud sector was Mag7 capex funndel from their cash cows segments (Azure, AWS, GCP) down into: $NBIS, $IREN, $CIFR, $WULF, and others. But as the largest players ( $ORCL, $CRWV) fall, these algorithmically bring down the whole sector. If you look at the companies individually, companies like $CIFR and $WULF are being backstopped by $GOOGL, and $IREN / $NBIS are funded by $MSFT. These are locked in contract backlogs from Hyperscalers/Mag7, not OpenAI. This irrational selloff due to misunderstanding of risks presents an amazing buying opportunity for the Necoloud sector, but not companies tied to OpenAI like $ORCL and $CRWV.