全部推文
共 6406 条 · 第 92 / 214 页 · 滚到底自动加载更多
-
Sivers以2.5亿美元估值复制LITE/COHR模式，成为CPO/ELS时代的光源核心供应商，被严重低估。
估值约2.5亿美元的$SIVE|$SIVEF看起来是光通信半导体市场中结构性错误定价最严重的标的之一。 $LITE和$COHR目前拥有450亿美元以上的估值，主要因为它们控制了外部腔体激光器(EML)和垂直腔面发射激光器(VCSEL)的出货——这些器件流向-> InnoLight和Eoptolink，用于当前可插拔光收发器(supercycle)。 我会一直强调这个观点，直到市场理解： Sivers已经为下一个技术范式复制了这一高度盈利的Merchant Supplier（商业代工）模式： -> 共封装光学(CPO)和外部光源(ELS)。 同时还有： -> 激光器，某大型制造商如Jabil正在用于当前的1.6T升级周期。 Sivers预见到架构正在从EML转型，并成为连续波(CW)激光器的纯Play供应商。 当前周期（可插拔光模块）：LITE/COHR供应EML -> InnoLight/Eoptolink制造模块 -> $GOOGL、$META、$MSFT、$AMZN购买。 下一个周期（CPO/ELS）：Sivers供应CW DFB激光器 -> $POET、Ayar Labs -> 流向超大规模数据中心商如$AMZN、$META、$MSFT等。 但与建设大型晶圆厂、承担良率/产能爬坡风险不同： 他们转型为无晶圆厂模式，通过Win Semi外包生产大批量CW激光器，有效降低了规模扩张风险。 那为什么在光电子supercycle期间估值只有2.5亿美元？ 我的观点： 1. 没人知道这家公司。 2. 许多基金的投资限制阻止它们购买瑞典的小盘股。 然而，当你开始关注超大规模数据中心光供应链中那些不知名的上游标的时： 光源领域最顶层的公司是$SIVE。 一家2.5亿美元的小公司，置身于$MTSI、$LITE、$COHR、住友电工以及百亿至数百亿美元市值的现有巨头之间。 再说一遍…… 一家2.5亿美元的公司： 为Jabil的1.6T LRO光收发器模块提供核心光源。 Jabil是巨头。 同时还为Ayar、$POET、Enablence/$ONET以及许多其他名字（尚未公开披露）提供多元化供应。 $SIVE已成功打入超大规模数据中心商的供应链，成为未来光电子架构的核心光源供应商。 相对于未来市场空间扩张和产能爬坡来看，估值被严重低估了。
英文原文
< $SIVE | $SIVEF > at a ~$250M valuation looks like one of the most severe structural mispricing in the optical semi market. $LITE and $COHR, have $45B+ valuations today: Largely because of their control over EMLs and VCSELs that they ship to -> InnoLight and Eoptolink. For current pluggable transceiver supercycles. I will keep hammering this home until markets understand: Sivers has replicated this exact, highly lucrative merchant-supplier model for the next paradigm: -> CPO and ELS. And also: -> Lasers that a massive manufacturer like Jabil is using for the current 1.6T upgrade cycle. Sivers saw architectures shifting away from EML and became the pure play supplier for CW. Current Cycle (Pluggables): LITE/COHR supply EML -> InnoLight/Eoptolink build modules -> $GOOGL, $META, $MSFT, $AMZN buy them. Next Cycle (CPO / ELS): Sivers supplies CW DFB -> $POET, Ayar Labs -> flows to hyperscalers like $AMZN, $META, $MSFT and others. However, instead of building up mega fabs with yield/capacity ramp risk: They transitioned to an outsourced, fabless model for high-volume CW Lasers with Win Semi and effectively de-risked scaling. So why does it have a $250M valuation during the photonics supercycle? My opinion: 1. Nobody knows about it yet. 2. Many fund mandates prevents them from buying small caps in Sweden However, when you start looking at obscure upstream names in hyperscaler light supply chain: At the top for the light source sits $SIVE. A small $250M company among $MTSI, $LITE, $COHR, Sumitomo and the $10B-$40B+ incumbents. Again... A $250M company: Powers Jabil's 1.6T LRO optical transceivers modules. Jabil is massive. As well as Ayar, $POET, Enablence/ $ONET and many other names now (that are not disclosed) for diversification. $SIVE found their way to scale as the lightsource of hyperscaler supply chains for future photonic architectures. And looks incredibly undervalued, relative to forward TAM expansion and ramp.
-
说作为美国人，自己倾向避开中国股票，但它们的 AI/机器人估值对散户太有吸引力。
我个人因为作为美国人的利益冲突，通常会尽量避开中国股票。 但它们的 AI 或机器人公司的估值，对散户资本来说实在是太有吸引力了。 和美国对应公司相比尤其如此。 我担心这会导致美国散户资金流向别的国家。
英文原文
I personally prefer to stay away from Chinese equities just due to conflict of interest as an American. But the valuations of their AI or robotics companies are just so much more compelling for retail capital. Compared to their US counterparties. I worry this will lead to US retail capital flowing to other countries.
-
说人们买苹果设备是为了工作和学校，但对方看的是三星衬底良率如何支撑 capex。
@jukan05 大多数人买苹果设备是为了日常工作或者上学。 @jukan05 买它，是因为三星 8.6G 衬底的良率足以支撑他们的 CapEx。
英文原文
@jukan05 Most people are buying Apple devices for daily work or school. @jukan05 is buying it because Samsung 8.6G substrate yields justifies their capex.
-
认为 SpaceX 以 3500 亿估值去年还能吸引散户，但 5 倍后再 IPO 就会变得完全不同。
散户去年、2025 年初，本来会很乐意把大量资金投进 3500 亿美元估值的 SpaceX……那时候它还被认为只是溢价。 可如果你随手把这个数字乘 5，再拿去 IPO 呢？ 那就是资金涌入，然后正和增长。 现在这种先在私募里把估值一路抬高，再 IPO 的模式，真的很不对……
英文原文
Retail would be happily pour in tons of money into SpaceX at $350B last year early 2025… which was still considered to be premium. Not as much so when you arbitrarily 5x the number, then IPO? It’s an influx of liquidity, then positive sum growth. The current model of privately inflating valuations, then IPO is so wrong…
-
说这还在早期，而不是成长后期。
@Alex__0x0 这并不代表它已经晚了。 它其实还早着呢。 只是他们在私下把估值抬得很高，然后才上市，这样散户就拿不到 2013 年那种 10-20 倍的上行空间了。
英文原文
@Alex__0x0 It’s not late in its growth phase. It’s still early. But they inflate valuations a lot privately before it goes public, before retail can get that 10-20x upside like IPOs back in 2013
-
夸张说 Anduril 会在 retail 进场前就以天价 IPO，并认为 Circle 才是最近更重要的美国前沿公司。
@Alex__0x0 我的意思是，Anduril 大概会在散户还没拿到之前，就以 500 万亿美元的估值 IPO。 我最近能想到的美国真正有吸引力的“前沿”公司，大概是 Circle，市值 320 亿。
英文原文
@Alex__0x0 My point is Andruil will IPO at $500 trillion before retail gets access to it. The main compelling US “frontier” company I can think of recently was Circle at $32.
-
质疑美国 IPO 模式，认为前沿独角兽估值过高，散户拿不到真正上行。
美国 IPO 模式到底怎么了？ 中国那边的领先前沿公司，比如 Minimax 或 Unitree，能以大约 50 到 60 亿美元的估值 IPO，挺有吸引力的。 而在美国，我们有 OpenAI 8000 亿美元、SpaceX 1.7 万亿美元、Stripe 可能 1400 亿美元。 Facebook 这样的公司去哪了？ 那种让散户真正能拿到上行，而不是只当退出流动性的公司，去哪了？
英文原文
What changed with the US IPO model? China out there IPOing their leading frontier companies like Minimax or Unitree at very compelling valuations at ~$5-$6B? In America, we have OpenAI at $800 billion; SpaceX at $1.7 trillion; possibly Stripe at $140 billion. Where are the Facebooks of the world, where retail actually can get upside instead of being exit liquidity?
-
转述一篇关于 TSM 与 INTC 的文章，提到台积电在光收发器制造上落后于英特尔。
@RenoHemonc 这是几年前的事了……出自一篇关于 $TSM 和 $INTC 的 SemiAnalysis 文章。 “TSMC 也参与了光收发器的制造，但它们的技术被认为落后于英特尔。” https://t.co/RYPfEFZ5rV
英文原文
@RenoHemonc This was a few years ago… from a semianalysis piece on $TSM vs $INTC “TSMC operates in manufacturing of both optical transceivers, but their technology is considered to be behind Intel’s” https://t.co/RYPfEFZ5rV
-
对某条内容的直接反应，语气很短很冲。
@jukan05 这事真没法委婉。那到底是什么玩意。
英文原文
@jukan05 No way to sugarcoat it. What was that
-
吐槽英特尔把可插拔光收发器硅光业务低价卖给 Jabil，几年后却迎来光子周期。
$INTC 把自己的可插拔光收发器硅光子业务给卖了…… 2023 年时还低价卖给了 Jabil。 结果两三年后，光子行业却进入了超级周期。 这可能算是十年来最蠢的决定之一？ https://t.co/ITvaFR20Xq
英文原文
$INTC dumping its pluggable optical transceiver silicon photonics segment… At dirt cheap in 2023 to Jabil. When there’s now a photonics supercycle 2-3 years later. Was possibly, one of the dumbest moves of the decade? https://t.co/ITvaFR20Xq
-
用“AI 养牛”调侃 AI 造成的自动化失业会继续外溢到更多行业。
我当时在推演那家做“牛用 AI”的 20 亿美元公司会带来哪些二阶效应。 然后我突然意识到一件很残酷的事： 牧羊犬也要失业了。 AI 带来的岗位替代会波及一切。 $ORCL、$AMZN、$META 等公司都在因为 AI 自动化裁员。 现在连你家附近的狗也要失业了。 我的宠物鱼是不是也要下岗了？
英文原文
I was mapping second-order effects of the $2B company that does AI for Cows. I just realized something brutal: Shepard Dogs Are Going to Be Unemployed. AI displacement is coming after everything. $ORCL, $AMZN, $META, and others are laying off workers from AI automation. And now your local dogs will be unemployed too. Is my pet fish going to be out of a job too?
-
认为 SMCI 光学业务很差，但按约 120 亿美元市值看，业务本身仍可能被低估。
@afgmantu 听我说……$SMCI 的光学业务烂得不行。 但它本身的业务按大约 120 亿美元市值看，似乎又被低估了。所以这里也许值得冒点风险。
英文原文
@afgmantu Hear me out... $SMCI optics are trash. But business itself looks undervalued at ~$12B MC. So might be worth taking the risk here.
-
感谢对方阅读自己的内容，并表达被关注的受宠若惊。
@Kaizen_Investor 不，应该是我谢谢你读我的内容。 这么多人在关注我的想法，真的让我受宠若惊！
英文原文
@Kaizen_Investor No, thank you for reading them. It’s humbling this many people are paying attention to my ideas!
-
认为股票长期是正和游戏，公开分享研究有助于信息更快扩散。
谢谢！我个人觉得，股票从长期来看是正和游戏。所以价值创造和新信息的传播，最终大家都会受益，没必要设付费墙。 以前的标准模式是：一份 2 万美元的分析师报告 -> 对冲基金先拿到 -> 股价上涨 -> 散户在市值已经到 200 亿美元以上后才知道。 现在有了 X，散户可以和机构抢先一步，第一时间拿到同样的信息发现。 （我并不反对分析师或想把成果变现的人；我只是更愿意打破这个老模式，把内容公开发出来。）
英文原文
Thanks! I personally think stocks are a positive sum game long term. So everyone benefits from value creation and novel information distribution and there’s no need to paywall. Before the standard model went from $20,000 analyst report -> hedge funds accumulate -> price goes up -> retail finds out after something is $20B+ Now with X, retail are able to frontrun institutions, by getting the same information discovery instantly. (No problem with the analysts or people who want to monetize their work, I prefer just to break the trend and post it publicly)
-
解释自己额外页面是用来放研究材料和零散想法的。
别觉得有压力去看，我不会把我的 thesis 帖子设付费墙。 那个页面只是多出来一块空间，用来放一些研究材料或者沿途漏掉的随机想法。我不想把主时间线刷得太满。 但如果其中某个点子或思路，能启发你自己的想法，那也挺好。
英文原文
Don’t feel pressured to, I don’t paywall any of my thesis posts. That page was extra space for research material that gets lost along the way or random thoughts. Didn’t want to flood my main timeline. But might be helpful if one idea or thought processes helps spawn some of your own
-
说自己经营金融科技公司，交易和供应链分析只是副业，也不想过度商业化。
我自己就在经营一家金融科技公司！所以我很乐意保持独立，只是分享一下我的想法，希望能帮到别人。 交易 / 供应链分析本来也只是我的副业爱好，碰巧我还挺擅长。 所以我从来没觉得有必要像现在很多账号那样，拼命商业化或者出卖自己。
英文原文
I run my on fintech company! So I’m happy to remain independent and just share my thoughts in case it helps others. Trading/supply chain analysis was just a side hobby anyway that I happened to be good at. So never felt the need to aggressively monetize or sellout like most of the accounts I see nowadays.
-
认为 AXTI 三个月涨 5 倍已经够传奇，并希望 SIVE 成为下一个类似标的。
@KonradJaku11615 我觉得 $AXTI 三个月涨 5 倍已经够传奇了吧？ 不过还是希望 $SIVE 成为下一个，真心感觉它是那种一年才会出现一次的机会。
英文原文
@KonradJaku11615 I feel like $AXTI going up 5x in 3 months was legendary enough? But hope $SIVE is the next, genuinely feels like a once a year type spot.
-
说自己昨天从日本飞到台湾，被邀请参加一些行业讨论和晚餐。
@JLIRL20 呃，我昨天是从日本飞到台湾的。受邀参加了一些行业讨论和晚餐。 https://t.co/rRjlGEIqHX
英文原文
@JLIRL20 Uhh flew in to Taiwan from Japan yesterday. Got invited for a few industry discussions and dinner. https://t.co/rRjlGEIqHX
-
感叹自己可能是 X 上最受订阅的金融账号，并表示风格不会变。
我刚意识到…… 我现在会不会是 X 上最被订阅的金融账号？ 最近真的有很多基金和公司开始找我提需求，这感觉有点不真实。 不过不管我只有 300 个关注，还是现在 10 万+，我都不打算改变任何东西。 我就是继续开心地和大家分享我的想法。
英文原文
I just realized… Am I the most subscribed to finance account on X right now? Kinda surreal so many funds and companies started reaching out for requests recently. That aside; not planning on changing anything when I was 300 followers and now 100k+. Just having fun sharing my thoughts with everyone.
-
吐槽对方测试新功能却不说明功能内容。
@nikitabier 老兄，别只说在测试新功能，却不告诉我们到底是什么！
英文原文
@nikitabier Bro u can’t just say testing a new feature but not tell us what it is!
-
用“牛”来调侃市场资金流向和避险主题。
黄金在跌。白银在跌。股票在跌。原油在涨。 那钱都流到哪里去了？ 我有答案： -> 牛。 牛群指数的表现已经跑赢了大盘。 硅谷和 Peter Thiel 已经往养牛相关初创公司里投了几十亿美元。 市场正在把“牛”当成对冲不确定性的最大避风港。 因为你不确定特朗普会怎么回应新局势， 但你知道牛会怎么做： 它们会“哞”一声，然后吃草。 市场也许在告诉我们，牛才是对冲不确定性的终极资产。
英文原文
Gold is crashing. Silver is crashing. Stocks are crashing. Crude Oil is rising. So where is the money flowing? I have the answer to that: -> Cows. The cattle index has outperformed the broader market. Silicon Valley and Peter Thiel have poured billions into cow-related startups. Markets are rotating to Cows as the biggest hedge against uncertainty. Because while you don't know how Trump will react to new situations? You know how cows do: They say Moo and Eat Grass. Markets might be telling us, that Cows are the greatest hedge against uncertainty.
-
回应 MU 的表现，认为其盈利很强，市场并非因为财报“超预期”而突然重估。
@alejandrobatiz 我从没说它不该更高。我本来就看多 $MU，不过更广泛的宏观环境正在拖累所有标的。 我只是说，如果像我这样的人都已经能把 $MU 的毛利率模型算进财报里，那市场就不该把它当成别人口中的“意外超预期”。
英文原文
@alejandrobatiz I never said it shouldn't be higher. I'm bullish on $MU, but there's broader macro dragging down every name. I'm just saying if people like myself were able to model in $MU gross margins for earnings, it shouldn't come to markets as a "surprise beat" that others are claiming.
-
回顾自己用便签式测算提前推演 MU 毛利率，结果与实际财报非常接近。
回到一月份： 我当时用便签式的粗算方法，拿美光的财报毛利率预测和官方指引做了对比。 我的原话是： “如果 $MU 指引 2026 年 Q2 的毛利率为 68.0%，而 NAND 价格从估算的 33-38% 直接上涨 100%， 那毛利率预测就可能被推到 73-75% 以上。” 华尔街当时的共识大概就在 67% 到 69% 左右。 而当 $MU 真正公布财报时？ 74.9% 的毛利率，正好落在我当时预判的 73-75%+ 区间。 这就是为什么信息会在财报前就被提前计价。 所以这根本不是一个“意外结果”，因为连我这种普通人一月份都已经能估出来 / 提前计入定价了。
英文原文
Back in January: I napkin-math modeled Micron's earnings margin projections vs. their official projections: My quote: "If $MU guided for 68.0% gross margins Q2 2026, and we see a 100% NAND hike from 33-38% est. That could bring gross margin projections over 73-75%." Wall Street consensus was sitting right around 67% to 69%. When $MU reported earnings? Their 74.9% margin landed in that predicted 73-75%+. This is how people price in information early, before the actual earnings. So it's not a "Surprise Result" when even people like myself could estimate it/price it in, back in January.
-
认为 MU 之所以今年仍跑赢大盘，是因为利润太强；若没有宏观冲击，本应更高。
@Indomitable912 是的，你的直觉是对的。$MU 今年到现在还涨了 34%+，而整个市场在下跌，原因就是他们利润实在太强。 如果不是伊朗战争、而市场又在定价降息，它现在可能已经涨到 60%+ 了。
英文原文
@Indomitable912 Yes your intuition is correct. The reason why $MU is still up 34%+ YTD while the entire market is tanking. Is exactly because they're making so much profit. If there weren't a War in Iran and markets were pricing in rate cuts, it would might be up 60%+ by now.
-
说明存储与光子股在宏观下跌中仍然坚挺，因为利润与增速过高，重定价通常发生在财报前。
同意。记忆芯片 / 光子相关标的在更广泛的宏观抛售里还能撑住，原因就是它们的净利润（存储）和收入增长（光学）都高得离谱。 如果没有伊朗战争，$MU 可能会高得多，但核心观点仍然成立：绝大多数重估发生在财报前，而不是财报当天，除非真的出现重大意外。
英文原文
Yep agreed. Reason why memory/photonics is holding up against broader macro selloffs is that their net income (for memory) and revenue growth (for optical) are just way too high. If there were no Iran war $MU would likely be tons higher, but the core point still stands that most repricing happens before earnings, not on the actual day, unless there's a genuine surprise.
-
总结 MU 的长期看多逻辑，并强调期权要在信息被市场完全定价前买。
是的，我确实认为 $MU 长期还会继续走高。 如果要给一个 TLDR： 1. 财报周不要去买被做空的期权，因为大概率财报信息已经被计价了。 2. 如果真要买期权，要在市场开始计价信息之前就下手（比如 DRAM / NAND 价格上涨的传闻阶段）。 3. 我不认为资本开支是这次抛售的主因，更多是期权资金流相关，然后市场需要一个故事去解释它。 4. 宏观逆风也很多，但总体来说，重估发生在财报前，靠着不同信息碎片慢慢拼出基本面变化。 财报只是对这一过程的确认而已。
英文原文
Yeah I do think $MU is going higher long term. If I had to give a TLDR: 1. Don't buy shorted options on earnings week, since earnings is more likely than not priced in. 2. If you're going to buy options, do it before markets price in information (eg. around rumors from DRAM/NAND price hikes) 3. I don't think capex was the reason for selloff, lot of it is option flow related and then there needs to be a story for it. 4. Lot of macro headwinds too, but in general, repricing happens leading up to earnings, piecing together different information that changes fundamentals. Earnings is just confirmation of that.
-
拆解 MU 财报前后的定价路径，认为市场在财报前已提前消化大部分利好，期权周押注并不划算。
我看到很多散户在 $MU 468 美元附近的财报上慌成一团。 一般来说，股票的重定价会在财报公布之前就发生。 例外只有一种情况：指引出现重大惊喜。 而美光这次的盈利超预期，并不是那种重大惊喜。 我的看法是： - 2026 年 1 月 25 日：公开消息传出，三星把 Q1 NAND 价格翻倍，DRAM 价格也比环比预估高出约 70%（估算当时只有 33-38%）。 机构可能提前几天就从 300 份付费供应商分析里拿到消息，所以 $MU 才会从 300 美元涨到 400 美元。 - 3 月初：又有消息称三星把 Q2 NAND 价格翻倍，这属于重大意外，比一些预期高出 20 倍，也比另一些预期高出 5 倍。 同时还有消息说三星 / SK 海力士准备在下个季度大幅提高 DRAM 价格。 但这又叠加了伊朗战争、以及更广泛的宏观恐慌导致的指数下跌。 尽管大多数标的都在大幅抛售，这些基本面变化仍然让 $MU 跑赢了那些今年跌了 20% 的公司。 在财报前，彭博终端上还有一份报告预测美光会超预期，这又把股价从 400 多美元推到 460 美元。 问题在于： 这次财报超预期的大部分内容，都是提前知道并且在整整一年里被慢慢计价进去的。 如果你读过我的帖子， - 我一直在跟踪 DRAM / NAND 价格，同时估算超大规模云厂商需求； - 也在追踪原油 / LNG / 氦气扰动对半导体供应链的影响，并把任何毛利率冲击建模进去。 这就是你该怎么给存储公司定价。 -> 财报只是证明你在财报前做出的估算是对的。 所以如果你在财报周买期权： 这真的是个很差的主意。 真正能赚到的利润，都在信息被市场知道并计价之前。
英文原文
I see a lot of retail panicking over $MU earnings at $468. In general, re-pricing in stocks happen before earnings are announced. The exception is "unless there's a major surprise with guidance". Micron earnings beat was not a major surprise. Here's my take: - Jan 25th 2026: Public news broke out Samsung was doubling NAND prices Q1, DRAM ~70% compared to Q/Q estimates of 33-38%. Institutions probably got word a few days early from 300 paid vendor analysis, which is why $MU went from $300 -> $400. - Back in early March: News broke out Samsung was doubling NAND prices for Q2, which was a major surprise, beating some estimates by 20 times, while beating others by 5 times. There was also news that Samsung/SK Hynix were planning large DRAM price increases for the next quarter. However, this was coupled with index dropping/broader macro fears from the Ukraine war. While there was a major sell-off in majority of names, these fundamental changes, likely led $MU to outperform names like $MU that dropped 20% YTD. Leading up to earnings, there was a Bloomberg terminal report projecting Micron to beat earnings, which sent the stock from $400's to $460. The thing is: Majority of this earnings beat was known in advanced and priced in throughout the entire year. If you ever read my posts, - I've been trying to track DRAM/NAND prices along with estimating hyperscaler demand: - Tracking to track crude/LNG/Helium disruption on semi supply chains and modeling in any margin impact Which is how you price in memory companies. -> Earnings are just confirmation that your estimates before were right. So if you're buying options on the earnings week: This is a terrible idea. All the profit to be made is before information is known and priced in by the broader market.
-
回顾父亲最看好的 PL，以及自己后来转向 RKLB 的经历和遗憾。
我爸最看多的股票？ $PL，3 美元时买的，现在 18 个月涨了 996.5% 以上。 我自己也为了好玩买过 1 股。 因为我当时并不觉得 Planet Labs 的营收能真正增长。 后来我转而重仓了 $RKLB。 现在 Planet Labs 已经涨了 10 倍以上，而我只剩下一堆懊悔。 顺带一提：他当年 35-40 美元时也很看好 $COIN，早期 $TSLA 他也看得很准，我一个都没听。 教训是： 有时候，你爸可能比你更会选股。尤其当你本来就跟他很像的时候。
英文原文
My dad’s most bullish stock? $PL at $3, which is up over 996.5% in 18 months. I bought 1 single share for fun. Because I didn’t think Planet Labs would grow revenue. I ended up long $RKLB instead. Planet Labs is now up over 10 times, and now I‘n left with a lot of regrets. FYI: They liked $COIN at $35-40 and early $TSLA as well, I didn’t listen to any of them. Lesson learned: Sometimes your parents are better investors than you are. Especially when you’re a mirror of them.
-
认为 SMCI 约 130 亿美元估值具备买点，即使短期波动也不太可能跌破 100 亿美元。
@gofallingknife 同意，我也觉得 $SMCI 130 亿美元的估值看起来像个买点。 短期围绕这个价位可能还会有波动，但从基本面看，很难想象它会跌破 100 亿美元。
英文原文
@gofallingknife Agreed, $13B valuation for $SMCI looks like a buying opportunity to me as well. Could be near term fluctuations around the level but hard to see it going below $10B given fundamentals
-
认为超大规模云厂商需要电力支撑推理，美国电网也在升级，但伊朗局势增加能源链的不确定性。
@Investmnt_Eagle @EquityAndy 是的，超大规模云厂商需要推理算力，也需要电力，而美国正在为此升级电网。 不幸的是，伊朗在能源贸易周围搞出一堆爆炸和冲击，这让短中期的不确定性多了很多。
英文原文
@Investmnt_Eagle @EquityAndy Yes, hyperscalers need power for inference, and US is modernizing the grid for that. Unfortunately Iran blowing a bunch of stuff up around the energy trade, leaves a lot of short-medium term uncertainty.