个股论点
围绕单一股票的投资逻辑与仓位表态 · 共 2378 条 · 滚到底自动加载更多
-
计划周末深入研究SHMD，看好其受益于玻璃基板生产启动。
@pennycheck 谢谢，这非常有趣。本周末我将深入研究 $SHMD。如果玻璃基板（Glass Substrate）生产周期启动，他们将受益匪浅。
英文原文
@pennycheck Thanks, that's incredibly interesting. I'll take a deeper dive into $SHMD over the weekend. They would benefit a ton from the start of glass substrate production cycles.
-
SHMD被视为玻璃基板领域的ASML，需深入研究。
@LucasFdez87 @pennycheck $SHMD 让我想起晶圆制造周期中的 $ASML，但玻璃基板领域也会出现类似情况。据悉，他们提供70%的基板生产设备，并很可能纳入英特尔、三星等公司的路线图。非常有趣，我需要深入研究一下。
英文原文
@LucasFdez87 @pennycheck $SHMD reminds me of $ASML for the Fab cycle, but there's going to be one for glass substrates. Apparently they provide 70% of the equipment for substrate production, and are likely part of Intel, Samsung, and other roadmaps. Very interesting, I need to look into it
-
NBIS波动极大，但看好其2026年跑赢大盘并扭转趋势。
$NBIS 目前基本上是一家由 5 家不同公司组成的实体，也是整个市场中波动率（Beta）最高的股票之一。 不过其波动性极大。其同行如 $CRWV 前阵子单日涨幅曾达 24%。 Nebius 在最初达成 $MSFT 交易时单日也曾上涨 40%。且在两周内经历了 100->140->100 的完整轮回。所以波动性极强，绝对不适合心脏脆弱者。 抛开这点，我对 Nebius 在 2026 年跑赢大盘并扭转当前趋势极具信心。这基本上等同于投资一家 FSD 级别的 L4 级自动驾驶出租车公司、一家几乎被所有财富 500 强使用的数据库，以及其他同比增长三位数并将更多利润反哺核心业务的业务板块。
英文原文
$NBIS is basically 5 different companies in 1 at this point and one of the most high-beta stocks in the entire market. Extremely volatile though. It's peers like $CRWV went up 24% in a day the other week off. Nebius also went up 40% in a day on the initial $MSFT deal. And round tripped 100->140 -> 100 all in the span of two weeks. So just extremely volatile and definitely not a stock for the faint of heart. That aside I'm extremely confident in Nebius outperforming in 2026 and reversing current trends. It's basically investing in a FSD-level 4 Robotaxi company, a DB that almost every fortune 500 company uses, and other moving parts growing triple digits Y/Y that derive more profit to the core business.
-
分析$PINS高毛利与营收增长，评估研发支出合理性。
我喜欢人们分享他们的高确信度观点。$PINS 很有趣，我完全忘了它的存在。这绝对是一个不错的候选标的，过去6个月下跌了28%。我看好的一点是，他们在季度营收超过10亿美元的情况下，毛利率约为79.8%，且营收同比仍增长16%。仅从高层概览来看：研发(R&D)支出：3.71亿美元；销售与营销(S&M)支出：2.97亿美元；一般及行政(G&A)费用：1.1亿美元。可能还需要花更多时间进行建模，看看那占营收35%的研发支出是否属于浪费性的冗余。但除此之外，2.97亿美元的营销支出是合理的。
英文原文
I love it when people share their high conviction. $PINS is interesting, I completely forgot that existed. Definitely a good candidate, down -28% last 6 months. What I like is that they have ~79.8% gross margins off $1B+ quarterly revenue, and still have 16% Y/Y revenue growth. Just looking at a high-level overview: R&D: $371M Sales & marketing: $297M G&A: $110M Will probably need to spend some more time to do modelling and see if that 35% of revenue on R&D is wasteful bloat. But otherwise the $297M marketing makes sense.
-
分析NVDA、SMCI等超跌股数据，认为其均值回归潜力大。
我也可以为 $NVDA 提出看空论点，即超大规模云服务商正在自研 ASIC。但在这种情况下，你要看数据而非叙事，比如英伟达的积压订单(backlog)，简直惊人。 $SMCI 远不及此，这也是它三个月下跌 40% 的原因，但它营收同比(Y/Y)仍增长 50%+，甚至将全年指引从 330 亿美元上调至 360 亿美元。 至于 $SNAP，如果你是价值投资者，它被严重低估。Perplexity 交易增加 4 亿美元价值，加上内存变现的净自由现金流(FCF)为 +6.3 亿美元。然而它年初至今仍下跌 25%。 $HIMS 完成了 Zava 收购并拓展欧洲。亚马逊等虽已推出竞品，但我认为 Hims 明年营收同比(Y/Y)仍将增长 20%+，且有回购。它们显然不像 SK 海力士、台积电、Lite-On 等神级股票那样以疯狂速度增长，但超跌+均值回归速度更快，上行空间可能更大。
英文原文
So I can argue the bear case for $NVDA too, that hyperscalers are building their own ASICs. But in cases like these, you look at numbers rather than narrative like, the backlog for Nvidia, it's wild. $SMCI is nowhere close, which is why it's down 40% in 3 months but it's still growing revenue 50%+ Y/Y and even raised FY guidance from $33B to FY $36B. As for $SNAP, it's materially undervalued if you're value investing. Perplexity deal adding $400m in value, then net FCF from memory monetization is +$630M. Yet it's trading 25% down from the start of the year. $HIMS has Zava acqusition and is expanding to Europe. Amazon and others already launched competitors but Hims I think is still growing 20%+ Y/Y into next year and has buybacks. They're obviously not god-tier stocks like Sk Hynix, TSM, Lite, and others just growing at insane rates but this is the point of being over-sold + revert to mean a lot faster. And upside could be higher.
-
谷歌股价波动源于情绪反转，预计Marvell也会如此。
@FinansMerak137 @SaItinerant 是的，我在 $GOOGL 跌至 140 美元时做多。基本面实际上没有任何变化，这纯粹是基于对搜索业务措辞解读的情绪反转。 我预计 Marvell 也会随机出现同样的情况。
英文原文
@FinansMerak137 @SaItinerant Yeah I was long $GOOGL on the drop to $140. Literally nothing changed about the business, it was just a sentiment flip just based on how you worded search. I'd probably expect the same with Marvell randomly.
-
驳斥分析师误判，强调Marvell在XPU设计上的营收潜力。
是的，很好的笔记。仅这一点就反驳了分析师声称 $MRVL 仅做连接业务，而将实际设计输给 Alchip 等竞争对手的说法。 机构分析师如何散布此类虚假信息，因称 Marvell 仅做连接业务而抹去 $100 亿+市值或 14-20% 的股价，引发机构研报的连锁反应，然后在上周无人关注时又说“只是开玩笑”，这太疯狂了。 “来自 XPU 设计的 2028 财年营收已相当可观”这一点也至关重要。
英文原文
Yeah great note. That alone addresses the analyst claims $MRVL was only doing connectivity while losing the actual design to a competitor like Alchip too. How institutional analysts can spread misinformation like this, wipe off $10B+ or 14-20% in stock price saying DG because of Marvell was only doing connectivity, send cascading narrator through institutional notes, then say “just kidding” when nobody is paying attention a week ago is wild. “There’s meaningful FY 2028 revenue already too from XPU designs” pretty important to read too.
-
批评机构误报，认为忽略噪音后MRVL是稳健长线标的。
$POET 通过 Celestial 间接受益于 $MRVL。此外还有大量与 Maia 绑定的其他光子学玩家，如 $AAOI。 但时间线简直荒谬，让我很恼火： 12月8日：Benchmark 下调评级，声称“高度确信”Marvell 输掉了 Trainium 3/4 给 Alchip。 12月9日：股价下跌约10%，CEO 上 CNBC 称“我们没丢失任何业务”。 12月23日：Benchmark 悄悄收回说法。Marvell “保留份额”，Alchip 仅增加“额外设计支持”。 Benchmark 做出此类声明后又在压低股价后悄悄收回，简直不可思议。 The Information 也如法炮制，但他们本就因煽动性报道闻名。 如果你只看数据并忽略新闻报道的噪音，$MRVL 看起来是一个极其稳健的长线标的。
英文原文
$POET indirectly benefits from $MRVL through Celestial. Then there's tons of other photonic players tethered to Maia like $AAOI. The timeline is so stupid though, it makes me annoyed: Dec 8: Benchmark downgrades, claims "high conviction" Marvell lost Trainium 3/4 to Alchip Dec 9: Stock drops ~10%, CEO goes on CNBC saying "we didn't lose any business" Dec 23: Benchmark quietly walks it back. Marvell "retains a position," Alchip added for "additional design support" Just found it insane how Benchmark could make claims like that then silently walk it back after tanking the stock price. Then The Information does the same but they're already known for their sensational journalism. If you just look at the numbers and ignore the noise from news reporting then $MRVL looks like an extremely solid long.
-
驳斥MRVL被取代谣言，指出物理限制使其无法短期切换，视抛视为买入良机。
终于深入研究了 $MRVL。 我的第一反应：市场和分析员/新闻到底在抽什么？ 来自 Maia 量产的收入几乎是 $MRVL 2025 财年总收入的两倍。 看数据，这令人震惊： 富邦 (Fubon) 预计 Marvell 仅在 2027 年微软 Maia 量产中就能获得 100-120 亿美元的收入。 作为对比，Marvell 2025 财年收入为 57.7 亿美元。 无论如何看，Marvell 看起来都是博通 (Broadcom) 的强劲竞争对手（在收购 Celestial 之后），也是一个极佳的长线标的。 但是... The Information / Benchmark 发布了新闻称： - 微软正在与博通谈判，以在未来几代产品中取代 Marvell。 - 亚马逊 Trainium 3 和 4 的设计流向了台湾竞争对手 Alchip Technologies。 导致股价大幅抛售。 深入研究后，这怎么可能呢？ “Marvell Technology 股价下跌，因为 Benchmark Equity Research 下调评级， citing 失去亚马逊 AI 芯片业务。” “Benchmark 认为 Marvell 输给了 Alchip，尽管 Marvell 最近预测数据中心增长强劲。” - Trainium 3 的生产已锁定 Marvell 的设计。如果不从头设计全新芯片，你无法“切换”到 Alchip。如果要共享范围，当然可以，但措辞简直糟糕透顶。 $MRVL 已经获得了 $AMZN 整个 2026 年的 Trainium 3 订单，亚马逊为什么要取消所有订单然后选择新供应商？ 可能有不同的范围、额外的设计服务等，但所有新闻的框架听起来像是 $MRVL 被取代了。 Benchmark 自己在造成损害后（通过更多“行业讨论”）大幅收回了说法，称 Marvell 并未完全失去 Trainium 3/4，而是亚马逊增加了 Alchip 作为额外设计支持，而 Marvell 仍保留地位。 但股价已经受损。 至于 The Information 关于 $MSFT Maia 输给博通的报道，这是煽动性新闻。 摩根大通 (JP Morgan) 后来直接出来基本说这是假的，“没有 ASIC 项目份额损失”。 在微软交易上，市场完全错过了“立即取代”和“未来几代”讨论之间的细微差别。 物理规律使得从 Maia 到博通的立即切换不可能，除非有惊人的多年延迟。 现在用博通取代 Marvell 的 Maia 300 需要废弃整个 2nm 掩模集，重新设计 I/O 环，并重启 3 年的验证周期。Marvell 特定的 SerDes IP 也嵌入在芯片的 I/O 环中（重定时器和交换机通信）。如果微软目标是在 2027 年量产 Maia 300，芯片必须在 2026 年中后期进入验证阶段。这意味着物理设计今天已经冻结或接近冻结。 今天（2025 年 12 月）与 $AVGO 的新 ASIC 设计将面临以下最短时间线： - 架构与 RTL（12 个月）：逻辑定义和 IP 选择。 - 物理设计到流片（9-12 个月） - 掩模生产与晶圆厂（3-5 个月）：TSMC 2nm 周期。 - 硅后验证（6-9 个月） 总时间：~30-38 个月。 即使加快几个月，这些“与 $AVGO 的讨论”最早也要到 2028 年才能大规模部署。为了赶上 2027 年的量产时间表，微软必须出货 Marvell 的设计。 此外，微软的机架基础设施使用 Marvell 的“Alaska”重定时器和 DSP 在电缆中。 微软必须： - 重新认证机架中每根电缆和背板的信号完整性。 - 拆下主板上的 Marvell 重定时器并替换为博通版本 还要更改他们的定制液冷机架“sidekick”，这是针对 $MRVL Maia 芯片特定版图分布指定的。$AVGO 芯片会有完全不同的物理版图和热密度，他们需要重新设计机架。 所以当市场计入立即取代风险并基于 TMZ 式的讨论八卦谣言交易时，他们完全是在开玩笑，谣言称微软想探索选项。 $MRVL CEO 甚至出面表示他们手中握有明年全年预测的“采购订单”。2026 年的预测可能偏低，因为 $AMZN 和 $MSFT 的量产都面临延迟。 但如果你看长期 Q4 2026 到 2027，仅从他们的订单和 Maia 300 来看：即使仅估计 ~200 亿美元收入带来 7.72 美元每股收益，30 倍市盈率（之前交易在 35-40 倍），如果是两年后等待，股价将从 85 美元涨到 231.60 美元/股。 所以 TLDR：整个“取代”理论被极其完全地误解了，只是噪音。通常我不会为此写整篇文章，但看到 The Information 和其他新闻一次又一次出现，通过煽动性措辞传播错误信息，这简直疯了。 中期切换在物理上是不可能的，市场误解了取代。所以最近的抛售看起来是一个坚实的买入机会。 至于未来的讨论，当然任何公司都希望供应商多元化。如果 Meta 想从博通多源采购，$MRVL 也可以对 $META MTAI 未来几代做同样的事。 如果仅按单个项目 Maia 300 建模并忽略媒体噪音，$MRVL 看起来非常有前景。
英文原文
Finally took a deeper look at $MRVL. My first reaction: What are the markets and analysts/news smoking? Revenue derived from Maia ramp is literally double $MRVL's FY 2025 revenue. If we look at the numbers, it's staggering: Marvell is modeled to take in $10-$12B in revenue (Fubon) in Microsoft Maia ramp in 2027. Alone. To put that in perspective, Marvell's FY 2025 revenue was $5.77B. No matter how I look at it, Marvell looks like a strong Broadcom competitor (after Celestial acqusition) and a terrific long. But... The Information / Benchmark released news that: - Microsoft is negotiating with Broadcom to displace Marvell in future generations. - Amazon Trainium 3 and 4 designs went to Taiwanese competitor Alchip Technologies. causing a massive selloff. After looking into it more, how is this even possible? "Marvell Technology declines after Benchmark Equity Research downgrades the stock, citing a loss of Amazon’s AI chip business." "Benchmark believes Marvell lost Amazon’s Trainium 3 and 4 designs to Alchip, despite Marvell’s recent forecast of strong data-center growth." - Trainium 3 production are locked to Marvell's design. You can't "swap" to Alchip without designing an entirely new chip from scratch. If you want to share scope, sure but the wording is absolutely atrocious. $MRVL already secured Trainium 3 orders from $AMZN throughout 2026, why in the world would Amazon cancel them all and then go with a new vendor? There can be different scope, additional design services, etc. but the framing from all the news sounds like $MRVL is being displaced. Benchmark itself later walked it back materially after causing damage (via more "industry discussions") and said Marvell did not lose Trainium 3/4 outright, but that Amazon added Alchip for additional design support while Marvell retains a position. But the damage has already been done to stock price. As for The Information report on $MSFT Maia loss to Broadcom this is sensational journalism. JP Morgan later literally came out and basically said it's all false, "there has been no ASIC program share loss". On the Microsoft deal, markets completely missed the nuance between "immediate displacement" and "future generation" discussions. The physics of this make an immediate swap impossible from Maia to Broadcom without incredible multiple-year delays. To replace Marvell with Broadcom for the Maia 300 now would require scrapping the entire 2nm mask set, redesigning the I/O ring, and restarting a 3-year validation cycle. Marvell’s specific SerDes Is is also embedded in the chip’s I/O ring (retimers and switches comms). If Microsoft targets a 2027 ramp for Maia 300, the chip must be in validation phase by mid-to-late 2026. This implies the physical design is already frozen or nearing freeze today. A new ASIC design with $AVGO today (Dec 2025) would face this minimum timeline: - Architecture & RTL (12 months): Logic definition and IP selection. - Physical Design to tape ot (9-12 months) - Mask prod & Fab (3-5 months): TSMC 2nm cycle. - Post-Silicon validation (6-9 months) Total Time: ~30–38 months. Even if this is sped up by many months, these "Discussions with $AVGO " not be ready for volume deployment until 2028 at the very earliest. To hit the 2027 ramp timelines, Microsoft must ship the Marvell design. Then there's Microsoft’s rack infrastructure uses Marvell’s "Alaska" retimers and DSPs in the cables. Microsoft would have to: - Re-qualify the signal integrity of every cable and backplane in the rack. - Rip out Marvell retimers from the motherboard and replace them with Broadcom versions As well as change their custom liquid-cooling rack "sidekick", which was specified toward specific distribution of the $MRVL Maia chip’s floorplan. A $AVGO chip would have a completely different physical floorplan and thermal density and they'd need to regineer the rack. So market are completely trolling if when they price in immediate displacement risk and trade on TMZ-style discussion gossip rumors that Microsoft wanted to explore options. $MRVL CEO even went out and said they have "purchase orders in hand" for the entirety of the next year's forecast. 2026 forecasts are probably light because $AMZN and $MSFT both faced delays on their ramp. But if you look at longer term Q4 2026 into 2027, we just from their orders and Maia 300 alone: even just estimating $7.72 EPS from ~$20.0B revenue, 30x P/E (they've traded 35-40 before), would be $231.60/share from $85 if you want to wait 2 years. So TLDR: The whole “replacement” theory is extremely completely misinterpreted and is just noise. Normally, I wouldn't write a whole post on it, but seeing The Information and other news pop up again, and again, spreading misinformation through sensational wording is just crazy. There's a physical impossibility of a mid-cycle swap and market misunderstanding of displacement. So the recent sell-off looks like a solid buying opportunity. As for future discussions, of course any company would want vendor diversification. $MRVL can go do the same with $META MTAI future generations too if Meta wants to multi-source away from Broadcom. If you model just by single project Maia 300 alone and ignore media noise, $MRVL looks incredibly promising.
-
认为英伟达因高盈利进入证明阶段
@BitcoinAIGuy 鉴于 $NVDA 的盈利能力，我认为我们正处于“证明它”的阶段，哈哈。
英文原文
@BitcoinAIGuy I think we’re a little prove it stage with $NVDA given how profitable they are lol
-
论证微软向Iren预付款项的商业逻辑。
@Lazarus_Capital 我召唤 @Agrippa_Inv 作为我的第一只宝可梦，来论证 $MSFT 为 $IREN 超大规模云服务商交易进行预付款的好处。
英文原文
@Lazarus_Capital I summon @Agrippa_Inv as my first Pokémon to argue benefits of $MSFT prepayment for $IREN hyperscaler deal.
-
澄清推文讨论的是IREN微软预付款计算。
@Lazarus_Capital 当然，这条推文具体是在讨论 $IREN 微软预付款的计算。
英文原文
@Lazarus_Capital Sure, so this one was specifically talking about $IREN Microsoft prepayment calculations.
-
因中国出口限制风险，买入AXTI远期看涨期权比持股更安全。
是的，这正是我买入 $AXTI 最远到期日看涨期权(Call Options)的巨大原因。如果存在2027年的行权价，我也会选择。如果中国限制出口，股价可能会被“核爆”式暴跌，此时持有股票可能损失约100万美元，而持有期权仅损失约10万美元。因此，这是极少数情况下看涨期权实际上比股票更安全的时候。
英文原文
Yeah that's a huge reason why I did $AXTI calls furthest date I could go out. I would have done 2027 strikes if they existed. If China restricts exports, it might get nuked, and you would lose like $1M in shares vs $100k in calls. So this is one of the very, very rare times Calls are actually safer then shares.
-
IQEPF虽核心但陷债务陷阱，资产出售时股东权益受损。
难道 $IQEPF 也在出售其公司以及台湾子公司吗？ 它处于一个尴尬的境地：对西方市场至关重要，但某种程度上陷入了债务陷阱。在资产出售中，汇丰银行(HSBC)和贷款方会优先获得偿付，而股东则会被牺牲。 就像这是一家极其稳健的公司，但你投资的对象却像是一家面临破产重组的精神航空(Spirit Airlines)类型的公司。
英文原文
Isn't $IQEPF selling their company as well as Taiwan subsidaries? It's in a weird spot where it's pretty critical to Western markets but there's kinda a debt trap. In an asset sale, HSBC and lenders get paid first and shareholders get shafted. Like it's an extremely solid company but the you're investing in some Spirit Airlines bankruptcy type company.
-
建议关注或尝试交易 $TBBK。
@TylerDurden 试试 $TBBK 吗？
英文原文
@TylerDurden Trya Banks?
-
澄清AXTI风险，探讨InP衬底瓶颈下的估值逻辑。
所以我并不主张买入 $AXTI，因为中国供应链存在重大风险，我只是在澄清一些误解。 如果一只股票“已经涨过了”，重要的是要区分它是否比以往任何时候都便宜，即使在历史高点(ATH)时也是如此。 看看 $MU 的前瞻市盈率(Forward P/E)，我上次查的时候即使在历史高点也小于10。 如果磷化铟(InP)衬底价格因为极端瓶颈而像内存一样翻三倍甚至十倍（鉴于它以前被当作廉价的电信商品定价），它是已经涨过了，还是看起来比以往任何时候都便宜？ 如果一家市值7亿美元的小公司现在成了AI增长周期的单点故障(Single Point of Failure)，因为我们已经触及了InP衬底的天花板，你会如何给它估值？500%的涨幅足够吗？
英文原文
So I’m not advocating for buying $AXTI since there’s significant risks with China supply chains but just clearing up misconceptions. If a stock “already-ran” it’s important to differentiate if the stock is cheaper than ever, even at ATHs. Just look at $MU forward p/e which is <10 last I checked even at ATHs. If inp substrate prices triple or even 10X (given it was priced like a cheap commodity for telecom) like memory because of an extreme bottleneck, did it already run or does it look cheaper than ever? If a $700m tiny company is now the single point of failure for the growth-cycle of AI we’ve hit the ceiling of InP substrates, how would you value it? Is a 500% increase enough?
-
质疑IQE估值，认为其属困境而非便宜，愿重新评估
@NowAwake_ @VulcanMK5 我上次看的时候，IQE 似乎正在进行资产出售？看起来更像是一家陷入困境的公司，而不是“便宜”。 如果情况有变请告诉我，我会重新审视。
英文原文
@NowAwake_7 @VulcanMK5 Last time I took a look, IQE was going through asset sales? Seemed like a distressed company rather than “cheap”. You can tell me if the story is different and I’ll take another look
-
因AI供应链瓶颈逻辑，做多光子股并提示AXTI地缘风险。
披露：我在光子学领域持有 $LITE、$AAOI 和 $AXTI 的头寸。 从我个人的思考过程来看，如果我观察到整个 AI 行业以及像 $LITE 和中际旭创(Innolight)这样的玩家，都被一家市值不足 6 亿美元（甚至低于一家尚未盈利的 LLM 初创公司）的企业所瓶颈制约，我会做多。 下行风险包括：如果中国希望限制 $AXTI 在美国市场，可能会改变其所有权稀释结构。
英文原文
Disclosure: I have positions in $LITE, $AAOI, and $AXTI in the photonics sector. From my own personal thought process, if I see the entire AI industry and players like $LITE and Innolight bottlenecked by some $600m company worth less than a new pre-revenue LLM startup, I’m long. There is downside risk including changing ownership dilution structures if China wants to restrict $AXTI in US markets.
-
看好LITE在ASIC及OCS中的核心地位，以及AAOI在Maia放量中的潜在机会。
@jvthed210725 $LITE 是个人最看好的标的，因为它在每一款 ASIC 中都是核心组件，且在类似 TPU 的光共封装(OCS)架构中，其物料清单(BOM)占比会显著提升。 $AAOI 与 Maia 和 Trainium 的放量呈杠杆关系。根据瑞银(UBS)及其他研究机构的报告，Maia 在 2026-2027 年有望实现百万级以上的出货量爬坡，因此这是一个被低估的潜在机会。
英文原文
@jvthed210725 $LITE is personally my favorite since it’s central to every ASIC + gets much higher BOM share with OCS architectures like TPU. $AAOI is levered Maia and Trainium ramp and we should get Maia 1M+ ramp 2026-2027 as per UBS note + other research firms, so it’s hidden opportunity
-
澄清MRVL与LITE为两只不同股票，将另写MRVL论点。
@bricked_trades $MRVL 是一个很好的做多标的，我会单独写一篇关于他们在 Maia 300 中角色的投资论点(Thesis)。 但这篇论点其实是关于 $LITE 的……所以是两只不同的股票。
英文原文
@bricked_trades $MRVL is a great long, I'll write a separate thesis on their role in maia 300. This thesis was about $LITE though... So two different stocks.
-
某仙股若执行成功，当前定价或低估其在AI集群中的价值。
嗯，不完全是。他们有过执行失败的历史，$AMZN 多年前就将其作为客户剔除。但如果考虑到他们现在手握铜缆到光子学（CPO）的“金票”，以及内存等供应常量，且能在此次成功执行，其当前的仙股定价可能并未计入他们在每个 Trainium 和 Maia 集群中的物料清单（BOM）份额，以及可选的美国磷化铟（InP）晶圆厂。
英文原文
Well not exactly, they have a history of execution failures with $AMZN dropping them as a client many years ago. But if they can execute this time around given their new golden ticket of copper -> photonics and supply constants like memory, their current penny stock pricing probably doesn’t price in their BOM share of every Trainium and Maia cluster they’re a part of + InP fab optionally in US.
-
$AAOI 作为 Trainium/Maia 杠杆标的，随微软 ASIC 量产预期重估。
呃，$AAOI 本质上只是 Trainium 和 Maia 的杠杆多头头寸，因为公开信息显示他们不像 $LITE 那样参与 TPU/Blackwell 业务。 Maia 已推迟至 2026 年，量产爬坡可能要到年底并延续至 2027 年。 特别是鉴于传闻 $AAOI 正与微软共同开发新架构，他们很可能会同步扩大规模。 总之，当这两款芯片启动时，我们可能会看到更多相关标的被重新评级。鉴于每家超大规模云服务商都想要自己的 ASIC（专用集成电路），这似乎是大势所趋。但关键在于 $AAOI 方面的量产爬坡与执行能力。
英文原文
Uhh $AAOI is just a levered long in Trainium and Maia, since they’re not involved in TPU/Blackwell unlike $LITE from public info. Maia was pushed backed to 2026, with ramp up in likely EOY into 2027. Esp with rumors $AAOI co-developing by new architecture with Microsoft, they’ll likely scale up together. Anyway whenever those two kick into gear, we’ll likely see more things be re-rated. It seems more of an inevitability given every hyperscaler wants their on asic. But main thing is production ramp/execution on the $AAOI side.
-
回应批评，重申NBIS长期价值及短期错判逻辑。
批评很中肯。我最初在 $NBIS 86美元时发布观点，即便在125美元我仍会买入。 我未曾预期一家远期年度经常性收入(ARR)达70-90亿美元、同比增长超700%且拥有4家各自增长超100%的子公司企业，市值会暴跌40%至190亿美元。 存在1-3个月的短期定价错误（尤其是受宏观压力、套息交易平仓引发的高贝塔资产抛售及临时信贷收紧影响），但我预计所有高确信度标的在6-12个月内将跑赢大盘。
英文原文
Very fair criticism. I first posted about $NBIS at $86 and would still be buying at $125. I did not expect a $7-9B forward ARR company growing 700%+ Y/Y, with 4 different subsidiary companies each growing 100%+ Y/Y to drop 40% to a $19B MC. There's things that are mispriced 1-3 month short term (especially with larger macro overhang with high-beta asset selloffs from carry trade unwind and temporary credit tightening) but I expect all my high conviction picks to outperform 6-12 months out.
-
$AAOI 是 $MSFT/$AMZN ASIC 扩产的高弹性隐藏宝石，具极高不对称收益。
在 $LITE、$COHR、中际旭创等公司中，我发现 $LITE 的不对称性（asymmetrical setup）最强，因为它是 Ironwood（含共封装光学(CPO)）、Blackwell 及其他超大规模客户专用集成电路(ASIC) 的核心。然而，$AAOI 是极致的超额收益（alpha）。尤其是作为 $AMZN Trainium 和 $MSFT Maia ASIC 扩产的对冲工具，以及美国少数磷化铟(InP) 晶圆厂之一。尽管拥有 $40 亿 $AMZN 认股权证和采购协议，12 月的订单似乎确实是基于 Maia 的集群。有传闻称 $AAOI 似乎正在为 $MSFT 的硅片开发新的光互连架构。尽管超大规模客户 ASIC 炒作和光子学反弹，$AAOI 股价毫无动静，因此在 $20 亿市值下绝对是隐藏宝石。但它高度依赖于 $MSFT 和 $AMZN（鉴于其云部门，它们很可能成功并大规模部署自有 ASIC）。
英文原文
Among $LITE, $COHR, Innolight, etc. I've found $LITE to be the most asymmetrical setup since it's central to ironwood (with OCS), blackwell, and other hyperscaler ASICs. However, $AAOI is extreme alpha. Especially as a levered play on $AMZN Trainium, $MSFT Maia ASICs scale up + one of the only InP fab in US. Although they have $4B $AMZN warrants + purchase agreements, it does seem likely that Dec's orders were Maia-based clusters. There's rumors that $AAOI seems to be developing a new optical interconnect architecture specifically for $MSFT's silicon too. $AAOI hasn't moved at all, despite hyperscaler ASIC hype + photonic rallies, so it's a hidden gem for sure at $2B MC. However, it is highly levered to $MSFT and $AMZN (which will likely succeed + deploy their own ASICs at scale given their own cloud departments)
-
JPM无需收购交易所，合作即可获利。
可能不会，没有理由这样做。鉴于$JPM提供的巨大做市商流动性，他们几乎肯定能享受0%的交易手续费。他们没有理由去收购一家同质化的交易所，从而承担额外的监管麻烦和资产负债表波动。通过与Coinbase合作获得免费执行，他们就能获得交易所的同等优势，然后通过增加一个垂直业务来赚取客户价差利润。
英文原文
Probably not, no reason to. $JPM almost certainty has access to 0% fees for trading given their immense maker liquidity they provide. There's no reason for them to acquire a commoditized exchange, then take on additional regulatory headache and balance sheet volatility. They get the same benefits of an exchange by partnering with Coinbase for free execution, then just profit off customer spread with an additional vertical.
-
分析LITE、COHR光子学双寡头地位及AAOI对微软亚马逊ASIC的杠杆看涨逻辑。
我手里只有 $LITE 和 $AAOI 哈哈。但我对 $LITE 和 $COHR 的看法是，它们在光子学(Photonics)领域基本处于双寡头地位。其中 $LITE 额外受益于 Google TPU 的增长杠杆。然后是 $AAOI，它具有极高的超额收益(Alpha)潜力，且是对 $MSFT 和 $AMZN 超大规模云计算厂商专用集成电路(ASIC)的杠杆式看涨押注。目前尚未有行动，也没有大幅上涨（尽管作为光子学玩家，今年反而下跌了12%），因为市场尚未对 Amazon Trainium 或 Microsoft Maia 产生炒作热度。
英文原文
I only have $LITE and $AAOI lol. But how I'd frame it is $LITE and $COHR are basically duopolies in photonics. With $LITE extra leveraged on Google TPU growth. Then there's $AAOI, extreme alpha + leverage call on $MSFT + $AMZN hyperscaler ASICs. No action yet, no major runup (even down 12% this year despite being a photonics player) because there's no investor hype about Amazon Trainium or Microsoft Maia (yet).
-
LITE因OCS垄断地位及供应瓶颈，受益于AI巨头需求。
$LITE 正处于“刚好合适”的黄金时刻，因为所有的 $GOOGL TPU、超大规模云厂商的 ASIC 以及 $NVDA Blackwell 架构都依赖它。目前他们因需求旺盛而面临供应瓶颈（类似于美光 Micron）。据我了解，光子学领域目前呈双寡头垄断格局（$COHR 和 Lite）。然而，在光电路交换(OCS)领域——这是 Google TPU 架构为实现更高性能所采用的技术——凭借专利组合和技术优势，$LITE 处于垄断地位。如果 $MSFT 和 $AMZN 最终也采用 OCS，那对 LITE 来说将是绝佳时刻。超大规模云厂商在磷化铟(InP)晶圆厂方面的垂直整合在短期内难度极大。未来几年内不太可能有人能取代他们（也许 2027-2028 年 $AVGO 的竞争会加剧），但就目前而言，他们只是受限于供应能力。
英文原文
$LITE is in the goldilocks moment where every single GOOGL TPU + hyperscaler ASIC + NVDA blackwell depends on it. Right now they're supply constrained from demand (similar to Micron). From what I understand, photonics right now is a duopoly ( $COHR and Lite). However, for OCS, which is what Google TPU architecture uses for better performance, $LITE is a monopoly from their patent portfolio + tech. If $MSFT, $AMZN end up adopting OCS, it's just a holy moment for LITE. Hyperscaler vertical intergration of InP fabs is just way too difficult near term. Nobody will likely replace them next few years at least (maybe 2027-2028 increasing competion from $AVGO) but as of now, they're just supply constrained right now.
-
推荐光子材料供应商AXTI，市值5亿美元。
@manukafirth 另一个不错的标的是 $AXTI，它是 $LITE / 光子学(Photonics) 的材料供应商。市值5亿美元。
英文原文
@manukafirth Another good one is $AXTI which is the materials supplier for $LITE / photonics. $500m mc
-
对比LITE与AAOI，看好AAOI在MSFT和AMZN ASIC中的重估潜力。
“太晚了”。 $LITE 是一家市值 260 亿美元的公司在未来 $GOOGL TPU 部署以及 $NVDA / $AMZN 等其他 GPU/ASIC 的核心位置。 它占据 Google TPU v7 8-12% 的物料清单(BOM)份额。这就像说当 $NVDA 市值达到 1 万亿美元时它“太晚了”一样。 但如果你想要另一个参与者，那就是 $AAOI。 $MSFT 正在 Maia ASIC 中使用它们，并且很可能正在与 $AAOI 创建一种新架构。 它们还与 $AMZN 拥有 40 亿美元的认股权证+购买协议。 所以这更像是一个做多 $MSFT Maia + $AMZN Trainium ASIC 的玩法，但市场尚未真正关注这两家超大规模云服务商的 ASIC。 如果有任何关于这两者的利好消息，$AAOI 将极快地重估。 尽管据报道 $META 将购买大量产品，但每个人都跳上了 $GOOGL TPU 的顺风车。
英文原文
"Late". $LITE is a $26B MC in the center of every future $GOOGL TPU deployment, $NVDA/ $AMZN / other GPU/ASICs. 8-12% BOM share of every Google TPU v7. This is like saying $NVDA was "late" when price it was a $1T MC. But if you want want another player, $AAOI. $MSFT is using them for Maia ASICs and are likely creating a new architecture with $AAOI. They also have a $4B warrant + purchase agreement with $AMZN. So it's more of a long $MSFT Maia + $AMZN Trainium ASIC play but markets haven't really cared about those two hyperscaler ASICs yet. If there's any positive news about those two $AAOI will revalue extremely fast. Everyone jumped on the $GOOGL TPU bandwagon after $META reportedly was going to buy a bunch though.
-
Lumentum凭借AI光网络垄断地位，有望随巨头资本支出增长实现市值翻倍。
$LITE 的投资逻辑：AI 领域的隐形垄断。 Lumentum 今年迄今上涨 316%，但到 2027 年可能上涨 1000%+。 美光 ($300B) 或台积电 ($1.5T) 位于每个部署的 TPU/GPU 中心。 但 $LITE 也是如此，而其市值仅为 $26B。 在谷歌的每一个 TPU 中，$LITE 为其市值创造了惊人的利润。 这是因为它是光路交换 (OCS) + 光网络的标准。 它也存在于： - $NVDA Blackwell - $AMZN Trainium - 以及其他超大规模云服务商的 ASIC 中。 Lumentum 位于光子学芯片部署的“神圣三位一体”中。 对于每笔 TPU 资本支出，$LITE 占据 8-12%。对于每个 Nvidia GPU，$LITE 占据 ~2-3%（在 InnoLight 和其他公司之间分配，所以计算有点复杂）。 仅针对 NVDA GPU 部署的 BOM 粗略估算： NVIDIA Blackwell (GB200)： HBM 内存：~50–55%（SK 海力士主导，美光，三星） 逻辑（GPU 裸片）：~25-30%（$TSM 4NP） CoWoS 封装：~13-18% $TSM 光网络：~3–5%（InnoLight, Lumentum, Coherent） PCB/电源：5% 谷歌 TPU “Ironwood” TPU v7： HBM 内存：38-42% 三星 / SK 海力士 逻辑裸片：TSM ~28-33% 设计/I.O：8-10% 联发科 光网络：10-14%（$LITE 主要，$COHR 次要） 光交换：2-4% $LITE $LITE 估计的总集群份额：~8–12% 仅供参考，谷歌的“光”BOM 份额（8–12%）是一个异常值，源于其独特的光路交换 (OCS) 垄断。 粗略估算： 到 2027 年谷歌 TPU 支出 $40B。$LITE 占据 10%（30-40% 利润率），仅来自谷歌的自由现金流超过 $1.5B，仅其主要客户就带来 17 倍市盈率。（分析师在预测 TPU 支出增长时可能严重偏离。） 这甚至不包括其在 $AMZN Trainium、$NVDA Blackwell、$MSFT Maia 和其他芯片部署中的份额。 $LITE 目前位于每个 TPU/GPU 未来芯片部署的中心，并从中分得一杯羹。 唯一的 downside 是，虽然他们现在是明确的市场领导者，但 $AVGO 和 $COHR 很可能在 2027-2028 年准备好竞争。然而…… 人们说“$26B，历史新高，为什么现在买”。 这就是原因。他们参与了未来每一个部署的 TPU/GPU/ASIC。 如果谷歌 TPU 和其他芯片支出增加，且 LITE 从每个美元支出中占据 2-3%（来自 $NVDA, $AMZN, $MSFT）或 8-12%（来自 $GOOGL），$LITE 最终很容易超过 $60B+。
英文原文
The $LITE thesis: The hidden monopoly in the AI. Lumentum is up 316% YTD, but might be 1000%+ by 2027. Micron ($300B) or TSM ($1.5T) sit in the center of every TPU/GPU deployed. But same with $LITE, but it's a $26B MC. In Every, Single, TPU from Google, $LITE makes unbelievable amounts of profit for their marketcap. That's because it's the standard for Optical Circuit Switching (OCS) + optical networking. It's also in - $NVDA Blackwell -$AMZN Trainium - and other hyperscaler ASICs. Lumentum sits in the holy trinity of every single chip deployment for photonics. And for every TPU capex spent, $LITE takes 8-12%. For every Nvidia GPU, $LITE takes ~2-3% (split between Innolight and some others, so the math gets a bit complex). But some napkin math on NVDA GPU deployments alone for BOM: NVIDIA Blackwell (GB200): HBM memory: ~50–55% (SK Hynix (Lead), Micron, Samsung) Logic (GPU Die): ~25-30% ( $TSM 4NP) CoWoS Packaging: ~13-18% $TSM Optics/Network: ~3–5% (Innolight, Lumentum, Coherent) PCB/Power: 5% For Google TPIU "Ironwood" TPU v7: HBM Memory: 38-42% Samsung / SK Hynix Logic Die: TSM ~28-33% Design/I.O: 8-10% MediaTek Optical Network: 10-14% ( $LITE (primary), $COHR secondary) Optical Switch: 2-4% $LITE $LITE est. total cluster share: ~8–12% Just an FYI, Google's "Optical" BOM share (8–12%) is an anomaly due to their unique Optical Circuit Switch (OCS) monopoly. Just for some napkin math: $40B Google TPU spend by 2027. $LITE captures 10% (30-40% margins), $1.5B+ FCF from Google alone, 17x earnings from just their primary customer. (analysts are probably extremely off with projecting TPU spend scaling). Not even including their split from $AMZN Trainium, $NVDA Blackwell, $MSFT Maia, and other chip deployments. $LITE is in the center of every single TPU/GPU future chip deployment for now and takes a cut. The only downside is they're the clear market leader now, but $AVGO and $COHR are likely set up to compete by 2027-2028. However... People say "$26B, ATH, why are you buying now". This is the reason. They're involved in every future single TPU/GPU/ASIC deployed. $LITE could end up easily over $60B+ if Google TPUs, and other chip spend ramps up and LITE takes a 2-3% (from $NVDA, $AMZN, $MSFT) or 8-12% cut (from $GOOGL) for every single dollar spent.