全部推文
共 6424 条 · 第 147 / 215 页 · 滚到底自动加载更多
-
指出盈透证券提供法兰克福上市的SK海力士交易代码。
@Ren_aramb $IBKR 提供 $HY9H，即法兰克福证券交易所上市的 SK 海力士(SK Hynix)。
英文原文
@Ren_aramb $IBKR has $HY9H which is SK Hynix on Frankfurt.
-
推荐AI供应链关键外国垄断股及美股替代投资渠道。
我首选的外国股票摘要： 1. 日本帝人博西(Nitto Boseki) - T-玻璃(T-Glass) 2. 台湾欣兴电子(Unimicron) - 基板/玻璃核心(Substrates/Glass Core) 3. 三星/SK海力士 - 高带宽内存(HBM) 4. 日本铠侠(Kioxia) - 与 $SNDK 合作的 NAND 闪存 5. 台湾南亚科技(Nanya Technologies) - 内存 我通常更偏好“美国制造”瓶颈及供应链相关的美国标的，如 $INTC (晶圆代工)、$MU (内存) 或 $LPTH (锗)。但这些外国公司要么对整个人工智能(AI)基础设施建设至关重要，要么能捕获大量溢出需求。 如果你想知道如何进入外国市场？ - $IBKR 像帝人博西这样的公司有美国场外交易(OTC)等价物，如 $NBCLF。如果你使用 Robinhood，$EWY / $FLKR 可以提供对韩国大型内存公司的敞口。 像帝人博西这样的公司接近垄断，因此这是我特意去投资的极少数外国公司。
英文原文
TLDR of my top foreign stocks: 1. Nitto Boseki - T-Glass (Japan) 2. Unimicron - Substrates/Glass Core (Taiwan) 3. Samsung/SK Hynix - HBM (Korea) 4. Kioxia - NAND w/ $SNDK (Japan) 5. Nanya Technologies - Memory (Taiwan) US supply chains like $INTC (foundry), $MU (memory), or $LPTH (germanium) are typically my longs of preference "Made in America" bottleneck + supply chains. But some of these foreign companies are either critical to the entire AI buildout or capture a lot of the overflow. If you're wondering how you get access to foreign markets? - $IBKR There are US OTC equivalents for companies like Nitto Boseki like $NBCLF. If you're on Robinhood, $EWY / $FLKR provides access to large KR memory exposure. Some like Nitto Boseki are near monopolies so these are the very rare foreign companies I go out of my way to invest in.
-
澄清关于BBW部件用于国防建设的虚假报道。
@ryn_xue 仅作免责声明：关于 $BBW 的泰迪熊部件被用于国防部 HIMAR 建设的报道并不属实（乍一看我以为是真的，结果是个笑话）。
英文原文
@ryn_xue Just a disclaimer: the report about $BBW's Teddy Bear parts being used for DoD HIMAR buildout is not true (I thought it was at first glance, but turned out to be funny).
-
调侃AI受益者可能出人意料，如BBW语音技术传闻。
@ryn_xue 去年有未经证实的报道指出，$BBW 旗下泰迪熊公司的高级语音盒技术被用于乌克兰的海马斯(HIMARS)系统。就像日本 TOTO 马桶因高带宽内存(HBM)而受益一样，有时最大的 AI 受益者往往是你意想不到的。 https://t.co/iZT1gNl7lu
英文原文
@ryn_xue There were unverified reports in last year that $BBW's Teddy Bear's advanced voice box technology was used for Ukraine HIMAR systems. Like the Japanese Toto toilets benefiting from HBM, sometimes the largest AI beneficiaries are the ones you never expect. https://t.co/iZT1gNl7lu
-
博主幽默回应，认为泰迪熊胜过HBM3e。
@CrypticAether 确实，泰迪熊 > 高带宽内存(HBM3e)。
英文原文
@CrypticAether True, Teddy Bears > HBM3e.
-
调侃AI算力焦虑实为童年怀旧，非技术瓶颈。
@CrypticAether “瓶颈”并非高带宽内存(HBM3)或关键材料，而是童年怀旧。我们曾以为需要10万个$NVDA H200集群来寻找生命意义，结果发现我们只需要泰迪熊。
英文原文
@CrypticAether The "bottleneck" wasn't HBM3 memory or critical materials. It was childhood nostalgia. We thought we needed 100k clusters of $NVDA H200s to find meaning in life. Turns out, we just needed teddy bears.
-
作者认为$BBW五年回报超英伟达，讽刺热门赛道拥挤。
人们做多 $MU / $SNDK 押注“存储超级周期(Memory Supercycle)”。 或者因为需求做多“铜(Copper)”。 但显而易见的交易就在你面前吗？ $BBW - Build a Bear（泰迪熊工厂）。 5年回报率： 泰迪熊：1301.55% 英伟达(Nvidia)：+1,258.68% 当你在 GPU/存储/铜供应链中为残羹冷炙争斗时，你本可以像我一样买一只泰迪熊。
英文原文
People are long $MU / $SNDK for "The Memory Supercycle". Or "Copper" because of demand. But is obvious trade in front of you? $BBW - Build a Bear. 5y Return: Teddy Bear: 1301.55% Nvidia: +1,258.68% While you're fighting for crumbs in the GPU/Memory/Copper supply chains, you could have just bought a Teddy Bear like I did.
-
博主回应感谢，表示分享观点能带来成就感。
@Ren_aramb 谢谢！我做这些是为了获得成就感，所以在发布我对某些话题的看法时，我确实能从中得到一些回报。
英文原文
@Ren_aramb Thank you! I do this for fulfillment, so I do get something out of it when I post my take on topics.
-
感谢互动，并指出新云投资逻辑获市场认可。
@EndicottInvests 谢谢！我记得经常看到你的名字出现在 $NBIS 相关的讨论中。很高兴很多人觉得关于新云(neoclouds)的投资逻辑很有说服力。
英文原文
@EndicottInvests Appreciate it! I remember seeing your name pop up a lot with $NBIS. I'm glad a lot of people found the thesis compelling with neoclouds.
-
感谢早期粉丝的长期关注与支持。
@adenois 哇，非常感谢你成为最早阅读我帖子并一直关注的人之一！我在心里为所有我的 $RDDT 朋友们保留了一个特别的位置。
英文原文
@adenois Wow thanks for being one of the first people to read my posts and still following along! I keep a special place in my heart for all my $RDDT friends.
-
引用三星NAND涨价100%报道，印证供应短缺。
@srikyada 帖子中已经明确引用了。但这是三星(Samsung)的版本：“三星电子，第一季度NAND价格上调100%……‘供应跟不上需求’”——电子新闻(ET News)
英文原文
@srikyada It’s literally cited in the post. But this is the SK version: “삼성전자, 1분기 낸드 가격 100% 인상…“공급이 수요 못 따라간다” - 전자신문 (ET News)
-
博主感谢粉丝评论，表示已阅读但无法一一回复。
@damnang2 非常感谢！ 借此机会，我很想回复这里每一条友好的评论，但实在太多了。 不过我确实已经阅读了大家到目前为止的所有留言。
英文原文
@damnang2 Thank you so much! Just piggybacking off this, I’d love to every other nice comment here but there’s just way too many. But I did read everyone’s message so far though
-
美光涨价打破传统模型，估值难追股价。
@sarfatti_IR 是的，这对 $MU 来说太惊人了，而且所有这些涨价都打破了传统建模。我又做了一些计算，前瞻市盈率(Forward P/E)可能会压缩至约 8.2-8.5 倍，对应 77.2% 的毛利率。估值无法跟上股价的步伐。
英文原文
@sarfatti_IR Yeah this is incredible for $MU and all these price hikes break traditional modeling. I did some calculations again and forward p/e might compress to ~8.2-8.5 of 77.2% gross margins The valuation can’t keep up with the stock price
-
美光涨价利好，前瞻PE降至8.9倍，股价滞后于基本面改善。
@TechEquityEng 随着这一疯狂的涨价，一些粗略估算(quick estimates)可能将 $MU 的前瞻市盈率(forward p/e)从之前的预期拉回到约 8.9 倍。 美光(Micron)的基本面每天都在变得更便宜，而股价却无法跟上。
英文原文
@TechEquityEng With this insane price hike, some napkin math calculations might bring $MU back to forward p/e ~8.9 from previous expectations. Micron fundamentally keeps getting cheaper every day and stock price can’t keep up.
-
对比SICC、COHR与WOLF在300mm晶圆技术路线上的差异。
@Zlack00 SICC 在 2024 年进行了展示，$COHR 在 2025 年 12 月拥有其 300mm 里程碑。 $WOLF 声称是首个 300mm “单晶”晶圆。其他则被声称是多晶或低质量衬底。因此措辞很微妙。https://t.co/fH2GCc2G5c
英文原文
@Zlack00 SICC had a showcase in 2024, $COHR has its own 300mm milestone in December 2025. $WOLF claimed to be first 300mm "single crystal" wafer. Others were claimed to be poly-crystal or low-quality substrates. So wording is nuanced. https://t.co/fH2GCc2G5c
-
将COHR比作西方欣兴，强调其在多领域供应链瓶颈中的关键地位。
@platochi $COHR 就像西方的欣兴电子（Unimicron）（指它们在未来瓶颈环节如碳化硅(SiC)到磷化铟(InP)中无处不在，并非比较上下游关系）。当我在太空、机器人、人工智能领域进行供应链映射时，我喜欢投资那些在多个地方频繁出现的公司。
英文原文
@platochi $COHR is like the Western Unimicron (in the sense they show up everywhere on future bottlenecks like SiC to InP, not comparing downstream/upstream). When I do these supply chain mapping in space, robotics, AI, I like to invest in ones that just pop up multiple places.
-
澄清ACLS用于SiC功率器件而非中介层，或涉及电源链。
@FrontRangeHQ 我相当确定 $ACLS 用于碳化硅(SiC)功率器件，而不是碳化硅(SiC)中介层？ 不过它们可能会出现在电源输送链中。
英文原文
@FrontRangeHQ Pretty sure $ACLS is used in SiC power devices, not SiC interposers? They might show up in the power delivery chain though.
-
梳理SiC中介层供应链瓶颈及2027年Rubin架构落地时间线。
如果你不喜欢 $WOLF，以下是 $NVDA Rubin 和超大规模客户 ASIC 潜在的碳化硅(SiC)中介层瓶颈股组合： 下游 SiC： $TSM - 用于 CoWoS 中介层的 SiC $AMKR - 用于 SiC 集成的 S-SWIFT 封装 $ASX - VIPack 平台及光学集成 中游 SiC： $KLAC - SiC 缺陷检测 $ONTO - 面板级封装+检测 $AMAT - SiC 沉积 $AEHR - SiC 检测 $ENTG - 化学机械抛光(CMP) $BESI (AS) - 混合键合设备 Disco - 70-80% 市场份额的精密切割 SiC 晶圆： $WOLF - 300mm SiC 晶圆 $COHR - 150mm 和 200mm SiC，仍在开发中 SK Siltron CSS (KRX: 034730) - 建设至 200mm Resonac (4004 TYSE) - SmartSiC 键合 NGK Insulators - 晶体生长 Kyocera - SiC 及陶瓷封装 这些都是早期供应链+活跃研究标的。 那么时间线何时到来？ 2025-2026 (Blackwell Ultra)：继续依赖硅中介层 2027 (Rubin / Rubin Ultra)：在最高端 AI SKU 中引入 SiC 中介层。 2028+：玻璃基板/SiC/金刚石中介层？ 披露：我持有上述 $AMKR, $AEHR, $ONTO, $COHR 和 $TSM。
英文原文
The potential SiC (Silicon Carbide) Interposer bottleneck stack for $NVDA Rubin and Hyperscaler ASICs (if you didn't like $WOLF): Downstream SiC: $TSM - SiC for CoWoS interposers $AMKR - S-SWIFT packaging for SiC integration $ASX- VIPack platform and optical integration. SiC Midstream: $KLAC - SiC defect inspection $ONTO - Panel-level packaging + inspection $AMAT - SiC deposition $AEHR - SiC inspection $ENTG - CMP $BESI (AS) - Hybrid Bonding equipment Disco - 70-80% marketshare precision cutting SiC Wafers: $WOLF - 300mm SiC wafers. $COHR - 150mm and 200mm SiC, still developing SK Siltron CSS (KRX: 034730) - Building to 200mm Resonac (4004 TYSE) - SmartSiC bonding NGK Insulators - Crystal growth Kyocera - SiC and ceramic packaging These are all early supply chain + active research. So when does this timeline hit? 2025-2026 (Blackwell Ultra): Continued reliance on Silicon Interposers 2027 (Rubin / Rubin Ultra): Introduction of SiC interposers in the highest-end AI SKUs. 2028+: Glass Substrates/SiC/Diamond Interposers? Disclosure: I own $AMKR, $AEHR, $ONTO, $COHR and $TSM above.
-
指出针对WOLF的可转债套利策略。
@foogleman1234 那可能是针对 $WOLF 的可转债套利(convertible note arbitrage)。
英文原文
@foogleman1234 That's probably convertible note arbitrage with $WOLF
-
发布西方供应链瓶颈小盘股清单，警示瓶颈不等于好投资。
@__EthanHunt__ 我在潜在瓶颈列表中列出了 $WOLF，这家公司确实很有趣。但“瓶颈”和“好投资”之间是有区别的。我觉得我做的尽职调查(DD)不能白费，所以干脆把它发出来了。 https://t.co/XCICxxODXQ
英文原文
@__EthanHunt__ I listed $WOLF in my list of potential bottlenecks, and the company genuinely is interesting. But there's a difference between "bottleneck" and good investment. Felt my the DD I did would go to waste, so ended up just posting it. https://t.co/XCICxxODXQ
-
WOLF虽具300mm SiC垄断优势，但高债与稀释使其成财务陷阱。
$WOLF（市值4.64亿美元）是“看空-看多”悖论的典型代表。 Wolfspeed凭借碳化硅(SiC)晶圆被视为AI领域的圣杯。 这家晶圆代工厂上周刚刚制造出全球首片300mm单晶SiC晶圆。 $NVDA的Rubin架构及后续架构很可能需要300mm SiC中介层(interposers)，而Wolfspeed是目前全球唯一能实现这一点的公司。 他们获得了7.5亿美元以上的《芯片法案》拨款，对美国国家安全至关重要。 我将此列为一个瓶颈环节，但关键在于： 这是一个负企业价值(EV)陷阱： - 9.26亿美元现金，21亿美元债务，4.65亿美元市值 - EV约为16亿美元。 此外还有： -> 可转换票据#1 约3.314亿美元。基于10亿美元隐含股权价值转换为新股。 -> 瑞萨(Renesas)可转换票据#2 约2.04亿美元（占总股份的13.6%） 2025年12月15日，Wolfspeed提交招股书，注册1130万股供“出售股东”使用。 这些是破产期间以股权代替现金的前债权人。 所以... Wolfspeed是2027年SiC中介层论点的核心，因为它们在美国制造且拥有巨大护城河。 但在他们能实现大规模量产并掩盖可转换债券、债务和稀释影响之前，这看起来像个陷阱。 从纸面上看，由于拥有全球首片300mm SiC晶圆且作为美国公司具有垄断地位，这是AI愿景股中的圣杯。 但财务上这是一颗地雷。 披露：我未持有任何头寸，只是想分享我的研究结果。
英文原文
$WOLF ($464M) is the paradigm of the "Bear-Bull" Paradox. Wolfspeed is the holy grail of AI with SiC wafers. And this foundry just made the world’s first 300mm single-crystal SiC wafer last week. $NVDA Rubin + architectures likely need 300mm SiC interposers and Wolfspeed is the only company in the world currently that can do it. They're critical to US national security with $750M+ in chips act grants. I listed this as a bottleneck but here's the catch: This is a negative EV trap: - $926M cash, $2.1B debt, $465m MC Ev would be roughly ~$1.6B. Then you have -> Convertible Notes #1 ~$331.4 Million. Convertible into New Common Stock based on an implied equity value of $1.0 Billion. -> Renesas Convertible Notes #2 ~ $204 Million. (13.6% of the total shares) December 15, 2025, Wolfspeed filed a prospectus to register 11.3 million shares for "selling stockholders" These are the former creditors who took equity instead of cash during the bankruptcy. So... Wolfspeed is central to the SiC Interposer thesis for 2027 as they're made in America and are have a massive moat. But this just looks like a trap right unless they can get this to massive scale and overshadow any convertibles, debt, and dilution. This is on paper the holy grail of AI moonshots since they have the world's first 300mm SiC wafer and are a monopoly in as a US company. But financially this is a landmine. Disclosure: I own no positions, just wanted to post my findings.
-
OPTT虽受Andruil概念炒作波动大，但高负债下稀释风险极高。
关于 $OPTT，我曾向他人评论称，由于 Andruil 相关的合同与合作伙伴关系炒作，该股极易出现大幅波动（因此有利可图）。但他们仅有 1100 万美元现金，却背负 1000 万美元债务和 1700 万美元负债，因此一旦有散户兴趣，稀释股本的可能性极高。除非我遗漏了某些超额收益(alpha)信息，否则资本稀释似乎迫在眉睫。
英文原文
For $OPTT, I made this comment to someone else where it's really susceptible to large movements (so profitable), bc there's hype on contracts like partnerships feeding into Andruil. But they have $11m cash and $10m debt with $17m liabilities so it seems incredibly likely they’ll dilute if there's any retail interest. Feels like capital dilution is imminent unless there's any alpha i missed
-
看好AEHR作为HBM4等关键技术的收费亭角色。
@PhotonCap 同意！我特别喜欢 $AEHR，尤其是因为它是许多领域（如 HBM4、碳化硅(SiC)晶圆、光子学等）的“收费亭”。
英文原文
@PhotonCap Agreed! I really like $AEHR especially, since they're a toll booth for a lot of things like hbm4, SiC wafers, photonics, etc
-
WLAC IPO延期至2026年中，但获1.27亿合同显利好。
@Sumojoe118 我看好 $WLAC，但他们从原定的2025年第四季度IPO严重延期，推迟到了2026年6月30日。因此，由于目前处于沉寂期，我此前并未过多评论，但最近获得的1.27亿美元FluidStack合同确实非常看涨。
英文原文
@Sumojoe118 I liked $WLAC, but they got heavily delayed from their Q4 2025 IPO, and extended it to June 30, 2026. So haven't really commented since it's in a lull state, but the $127m fluidstack contract recently is really bullish.
-
长期看好RKLB，但因短期高估暂不纳入短中期名单。
@AstronautCoin 长期来看，$RKLB 可能是我最看多的股票。不过短期来看它被严重高估。所以我暂时将其排除在外。这更像是一份我的短中期观察名单。
英文原文
@AstronautCoin Long term, $RKLB is probably my most bullish stock. Short term though it's pretty overvalued. So I just leave them out. This is more like a short-med term list for me.
-
提及无人机及战争板块个股机会，强调研究充分后再评论。
没问题！像 $RCAT、$PDYN 这样的无人机板块股票也有很多，从主题上看应该也是看涨的。只是需要关注的股票太多了，如果我对某只股票的研究还不够深入，我尽量不去评论。$OSS 未来应该会很有趣，因为有战争板块的顺风因素。
英文原文
No problem! There's a lot of drone sector stocks like $RCAT, $PDYN too that should be thematically bullish. There's just too many stocks to look at and I try not to comment if I didn't do enough research into them yet. $OSS should be interesting going forward with War sector tailwinds
-
看好MRAM高上行空间，修正VELO资产负债表数据但仍认为其一般。
$MRAM 很有意思。资产负债表良好，且拥有极高的上行垂直空间。 另外需要纠正一下上面关于 $VELO 的说法，他们手头约有 4990 万美元现金 - 1100 万美元（第四季度支出），所以现在可能持有 3800 万美元，负债 2300 万美元 | 总负债 6400 万美元，看起来稍好一些。但资产负债表仍然不是最好的。
英文原文
$MRAM is interesting. Good balance sheet, and very high upside vertical. Also need to make a correction to $VELO above, they have around $49.9M in cash -$11m Q4, so they're probably sitting on $38m now with $23m debt | $64m in liabilities, which looks slightly better. But still not the best balance sheet.
-
分析SHMD、COPX及UAMY的投资逻辑与潜力
我之前一直在研究 $SHMD，因为其客户据报包括 $AVGO。但我至今仍未理清做多该股的逻辑。 我将 $COPX 归类为铜板块。 $UAMY 是做多锑的绝佳标的，它是美国少数玩家之一。其 2.45 亿美元的独家供应合同对于战略储备而言规模巨大。
英文原文
I've been looking into $SHMD awhile back since one of their customers was reportedly $AVGO. But still can't wrap my head around taking a long position yet. I put $COPX up there as Copper. $UAMY is an amazing long for antimony one of the only US players. $245m sole source contract is huge too for stockpiles.
-
博主计划研究$EOPSF，但坦言对定向能武器领域不熟悉。
@ShortSeller 我还没研究过，但鉴于 $EOPSF 看起来很有趣，我很快就会去研究。不幸的是，我对定向能武器（energy weapons）领域不太熟悉。https://t.co/6XddprIHda
英文原文
@ShortSeller Haven't research it yet but I will soon since $EOPSF seems interesting. I'm not well versed in energy weapons unfortunately. https://t.co/6XddprIHda
-
Oracle因过度依赖OpenAI面临巨大交易对手风险，长期前景堪忧。
尽管$ORCL股价在300多美元时，我也曾表达过类似观点，当时他们预测资本支出(Capex)将在一年内从250亿美元激增至500亿美元以上。这是为了履行5000多亿美元的积压订单。但当客户是OpenAI（例如3000亿美元的Stargate项目）时，这些积压订单的意义就变了。他们借入巨额资金只为单一客户建设，而鉴于OpenAI的其他合同义务，这构成了最明显且极端的交易对手风险。加上在Gemini面前失去收入份额，他们最终违约的可能性越来越大。我高度怀疑Sam Altman能否维持1万亿美元以上的合同义务，即使进行IPO（他们前几天刚与Cerebras签署了另一项100多亿美元的合同）+ 他们只是在传播风险传染，试图在资金耗尽时获得美国政府的兜底。这种对可疑交易对手的债务扩张掩盖了其其他业务板块，我认为市场上有很多更好的做多标的，而不是去赌OpenAI。当然这只是我的观点，我们可能会看到因Softbank等近期对OpenAI的注资而带来的Oracle短期反弹，但我不认为这种势头能持续多年。
英文原文
I've been saying this about $ORCL in the 300's though too when they projected $25B capex to $50B+ in just 1 year time. This is to fulfill $500B+ in backlog. But that backlog doesn't matter when it's OpenAI (eg. $300B Stargate). They're borrowing so much to build out on a sole customer, which happens to be the most obvious + extreme counterparty risk given OpenAI's other contractual obligations. When you compound that with losing revenue share to Gemini, it looks increasingly likely they'll default at some point. I have high doubts Sam will find a way to sustain $1T+ in contractual obligations, even with an IPO (they just signed another $10B+ one with Cerebras the other day) + are just spreading contagion so they try and get backstopped by US government if they run out of funds. This type of debt buildout on a questionable counterparty overshadows their other business verticals, and my opnion is that there's a lot better longs out there than taking a risk on OpenAI. Just my opnion though, we might see a short term oracle recovery bc of openai's recent funding from softbank and others but I don't think it can be sustained over the years.