个股论点
围绕单一股票的投资逻辑与仓位表态 · 共 2373 条 · 滚到底自动加载更多
-
只有一个股票代码。
@mi20483980476 $AAOI
英文原文
@mi20483980476 $AAOI
-
虽然看多存储，但宏观和战争会盖过个股表现。
没错，我总体上还是看多存储，因为我觉得需求是结构性的。价格上涨不是结构性的，但未来 $MU 到 SK 海力士的毛利率会一直很高。 不过宏观 / 战争会压过 $EWY 和其他名字。 所以如果有全球能源危机，个股表现也就没那么重要了。
英文原文
No, I'm largely bullish on memory since I think demand is structural. Price hikes are not, but gross margins will be high for $MU to SK Hynix moving forward. However, Macro/War overshadows $EWY and other names. So individual performance won't matter as much if there's a global energy criss.
-
警告$IREN的60亿美元ATM大规模稀释将清算散户股权，历史先例已证明结局惨烈
$IREN 提交了文件，计划以 117 亿美元市值稀释 60 亿美元。 这不是噪音。 这是 Iren 通过向公开市场发售所有新股来变现 4.5GW 容量的方式。 如果你想了解这类操作的历史结果： 看看 $BKKT 在 Mike 手下暴跌 99% 的先例，以及 $IREN 董事会过去大规模 ATM（按市价发行）的历史。或者他最近的公司 $ASST。 这对公司和高管是增值的：因为这清算了所有散户股东，而他们总可以发行 SBC（股权激励）。 所以他们实际上并不在乎股价需要达到什么水平才能发售。 完成之后，他们将拥有 60 亿美元新现金用于扩张，而无需支付利息。 但可转换票据附带利息、以及 $NVDA 资助资产负债表的模式对散户资金好得多的原因是： 因为它不需要清算散户股权就能实现这一点。因为目前看来 $IREN 像是数据中心版的 $AMC，拥有不断缩小的护城河，以及即将有 60 亿美元新股涌入公开市场。 我发帖讨论 $IREN 的原因是： - 人们将 60 亿美元 ATM 视为"噪音"而不予理会 - 它是最受散户欢迎的"抄底"标的之一，而这些散户正在买入一个 60 亿美元稀释机器 - 人们仍然完全不了解其中的风险 - 他们目前的资金量不足以支撑 GPU/GW 容量的货币化 - 他们很可能必须使用 ATM，这不是"选择性" 再次声明：我在这家公司零仓位。 我只是警告散户投资者，这个 ATM 的结构性机制如何通过 60 亿美元以上的 ATM 运作来从结构上抹杀你的股权增值。 因为 $IREN 可能需要在任何价格出售新股来变现他们的 GW，否则就没有必要提交这个文件。 高管们实际上不需要担心，因为他们可以通过发行 SBC 来弥补股价下跌，就像 $SNAP 一样。 如果你还在疑惑你的股权是否会被大规模 ATM 清算： 还有更好的做多选择，比 $IREN 更值得。
英文原文
$IREN filed to dilute $6,000,000,000 at a $11.7B MC. That is not noise. This is Iren's way to monetize their 4.5GW capacity by selling all those new shares onto the open market. If you want some history on how this turns out: Look at $BKKT that crashed 99% with Mike and $IREN board of directors history with excessive ATMs. Or his recent company $ASST. It’s accretive to the company and executives: Because it wipes out all retail shareholders and they can always issue SBC. So they don’t actually care what stock price it needs to be at to sell. After they’re finished, they have $6B in new cash to use for scaling without paying interest. But the reason why convertible notes with interest, and $NVDA funding balance sheets is much better for retail capital: Is because it doesn’t wipe out retail equity to achieve this. Because at this point $IREN looks like the $AMC of datacenters with a dwindling moat, and looming $6B in shares sold into the open market. Reason I post about $IREN is because - people dismiss a $6B ATM as “Noise” - it’s one of the most popular retail “buy the dip” companies that they’re buying into a $6B dilution machine - people still don’t understand the risk at all. - the amount they have now is not enough to finance GPUs/GW capacity monetization. - they likely will have to use the ATM, it’s not “optionality” Again: I have zero positions in the company. I’m just warning retail investors that this ATM structurally wipes out your equity appreciation by how structural mechanics of $6B+ ATMs work. Because $IREN likely needs to sell new shares at any price to monetize their GW, otherwise there would be zero need to file it. Executives actually don't need to care because they can make up for stock price dropping by issuing SBC like $SNAP. If you have to wonder if your equity gets wiped out from an excessive ATM: There are better longs out there than $IREN.
-
补充说 GLNG 也是不错的选择。
@NishantChandra2 $GLNG 也是个不错的选择。
英文原文
@NishantChandra2 $GLNG is a good pick as well
-
认为 RDDT 掌握用户信息后，定向广告收入会像 META 一样提升。
@SidG1293 如果他们 $RDDT 掌握了用户信息来做账号验证，那定向广告收入上升的现实就会像 $META 一样。
英文原文
@SidG1293 If they $RDDT knows people’s info for account verification. The sad reality is that their targeted ad revenue goes up like $META.
-
认为战时最直接的做法可能是做多 LNG 和石油出口，而不是逆势抄底。
老实说……最明显的想法，比如做多 $LNG，或者做多 $CVX，可能就是最好的，而不是在战时去逆势做多别的东西？ 尤其是在 18-30 的低 IV 下，这些到现在本来都能轻松涨几百个百分点了。 如果局势继续升级，石油和 LNG 出口商大概率还有更多上涨空间。
英文原文
Honestly... the most obvious ideas like long $LNG. Or going long on $CVX. Are probably the best ones instead of contrarian longs in Wartime? Especially with low 18-30 IV, this would have easily been a few hundred percent gain by now. There’s probably a lot more for oil and LNG exporters to run if tensions escalate.
-
看好 NEXT 的三到五年长期逻辑，LNG 出口商会受益于伊朗局势。
对 $NEXT 来说，这是一个极其扎实的 3 到 5 年长期多头，尤其是当产能已经卖光了的时候。 不过 LNG 的产量要到 2027 年上半年？ 但我确实认为整个板块都会受到关注。 伊朗事件的主要受益者就是像 $LNG 这样的 LNG 出口商，现在它们在为全球其他地区托底。
英文原文
Extremely solid multi-year long (eg. 3-5 years) for $NEXT, especially when capacity is sold out already. LNG production is first half of 2027 though? Though I do think the sector as a whole will get attention. The main beneficiaries of Iran are the LNG exporters like $LNG now backstopping the rest of the world though.
-
认为 NOC、LMT 已提前计入冲突，油价也早已先跑，后续仍可能继续上涨。
$NOC、$LMT 在冲突发生前就已经先跑了，因为市场大体上早就预期到伊朗这件事。Lockheed 年初至今涨了 23%，即使回落后也已经非常夸张。 油价也提前交易过了，但伊朗把一切都炸开的二阶、三阶影响远超预期。 所以这些标的还在继续走强，因为如果局势进一步升级，前面可能会有一场巨大的危机。
英文原文
$NOC, $LMT already frontran ahead of conflict since Iran was largely expected. Lockheed up 23% YTD is enormous already, even after the drop. Oil was frontran as well, but the third order effects of Iran blowing everything up was way beyond expectations. So those kept going up, since there's a possibility of a massive crisis ahead if things escalate even more.
-
说可以把 MU/SNDK 换成海力士，把 COHR 换成 LITE，强调自己偏好的标的。
@ajouannic 你可以把 $MU / $SNDK 换成海力士。或者把 $COHR 换成 $LITE。 我只是说说我个人更看好的替代选择。
英文原文
@ajouannic You can replace $MU / $SNDK with Sk Hynix. Or replace $COHR with $LITE. I'm just saying my personal preferences in what I expect to outperform.
-
分享小账户分散化投资组合，涵盖AI半导体、光子学、机器人、加密等领域的高增长和杠铃策略标的
快速增长赛道： $AAOI - 光收发器2027年下半年收入增长10倍 $NBIS - 2026年第四季度收入增长10倍 $ARM - 全新AI CPU带来5倍收入增长 $MRVL - 来自$MSFT Maia芯片项目2-3倍收入增长 $AVGO - 长期持有超大规模云厂商ASIC $LITE - 长期持有OCS/谷歌TPU Win Semi - 对前沿行业的代工敞口 $TSEM - 长期持有光子业务，订单积压 SK Hynix - 存储业务敞口，极端营业利润增长 同时配置一些杠铃策略，远离超大规模云厂商资本支出（除亚马逊外）： $VNP - 长期持有西部供应链的稀土 $NEO (TCX) - 机器人供应链 $AMZN - 机器人/AI降低运营成本 $CRCL - 稳定币多头 $RDDT - 高得离谱的利润 $GLD - 安全对冲 $IBIT - 2028年减半 $CVX 看涨期权 - 石油对冲 也许还有长期配置（你知道迟早会来）： $INTC/$AMKR - 美国制造供应链 $SOI - 硅光子学/共封装光学(CPO)基板 $RKLB - 对太空行业的长期看涨期权 然后挑选一两支小盘股博暴涨： $SIVE - CW激光关键点或$I QE - Landmark重组重估，这是我最喜欢的两个 还有其他我提过的，比如$AEHR用于测试或$VPG用于Optimus 我自己在$AXTI等股票上的主动管理模式与其他人应该做的完全不同风险敞口配置。 在这种宏观环境下全仓投入高贝塔标的不太是最佳选择。
英文原文
Faster compounds: $AAOI - 10x revenue ramp from optical transcivers h2 2027 $NBIS - 10x revenue ramp Q4 2026 $ARM - 5x revenue growth from their new AI CPU $MRVL - 2-3x revenue growth from $MSFT Maia Ramp. $AVGO - Long hyperscaler ASIC $LITE - Long OCS / Google TPU Win Semi - Foundry exposure to frontier industries $TSEM - Long photonics, backlogged SK Hynix - Memory exposure, extreme operating income ramp With some barbell exposure away from Hyperscaler capex aside from Amazon: $VNP - Long term rare earths for Western Supply chains $NEO (TCX) - Robotics Supply chains $AMZN - Robotics/AI cutting opex $CRCL - Stablecoin long $RDDT - Ridiculously high profit $GLD - Safe Hedge $IBIT - Halving 2028 $CVX Calls - Oil Hedge And maybe long term (you know it's coming): $INTC / $AMKR- Made in America supply chains $SOI - Silicon Photonics / CPO substrates. $RKLB - Long term call on Space industry Then pick one or two small cap moonshots: $SIVE - CW Laser Chokepoints or $IQE for Landmark rerating on restructuring were my two favorites. There's others I've mentioned like $AEHR for testing or $VPG for Optimus. How I actively manage my own stuff from $AXTI and others is a lot different risk profile than what others should do. Going full port into high-beta in this macro environment is not the best idea.
-
说如果连 CRWV 都快撑不住了，市场上大概还有更好的多头。
@BelfortsTaxGuy 如果你已经开始想，$CRWV 几乎都要撑不住了……那市场上大概还有更好的多头，我只是说一下。
英文原文
@BelfortsTaxGuy If you have to think about almost not sinking for $CRWV... there's probably better longs out there in the market, just saying?
-
对比IREN与SIVE稀释规模，认为SIVE风险回报更优
@Jornka329996 @gude57856 $IREN正在稀释超过60亿美元的股份，却无人关注。$SIVE可能需要稀释1500-2000万美元来扩大连续波(CW)激光业务板块，开始成长为下一个$LITE？风险回报特征对我而言更青睐像$SIVE这样的名字。
英文原文
@Jornka329996 @gude57856 $IREN is out there diluting $6B+ and nobody bats an eye. $SIVE could potentially dilute $15M-20M to scale up CW laser segments and start growing into the next $LITE? Risk reward profiles favor names like $SIVE for me.
-
对比三只Neocloud股票，认为$NBIS是最佳选择，$CRWV融资结构致命，$IREN长期增值但短期散户不利
我对$NBIS、$IREN、$CRWV以及当前Neocloud市场的看法。 其中一家将在5年内成为下一个AWS： 我的猜测是Nebius。 这不是赢家通吃（DigitalOcean与亚马逊并存），但显然有更优越的结构和可能的赢家。 不利因素： ->伊朗冲突导致降息的可能性很低。 ->更广泛的市场似乎不愿意资助CapEx周期。但又想从中获益。 关于$IREN： 我们理解，4.5GW = X收入。但谁来资助GPU？ 无论谁现在买入$6,000,000,000的ATM。 赢家将是稀释完成后进入的任何人。 现实是，如果没有colo模式，他们没有足够的资金来通过GPU货币化其容量。 而且他们没有找到其他融资方式，因此由于一个会买入他们出售的任何东西的邪教社区，他们选择了ATM。 然而，我同意长期来看这是增值的。只是对现在买入的散户没那么有利。 关于$CRWV： 他们做对了一切... $NVDA支持。超大规模客户... 但他们融资完全错了。现在，超过$1.5B+的年度债务利息正在吞噬Coreweave的利润并侵蚀FCF。 就像信用卡债务一样，Coreweave找到工作来偿还债务，但最终债务利息太高，工作收入不足以覆盖债务和扩张。 如果有任何公司倒闭，$CRWV是第一个因巨额债务和利息而倒下的。 关于$NBIS： 他们正在尽可能做好... $NVDA提供$2B资助capex。 可转债发行（可转债做空对冲对短期价格上涨不利）。 但这是比Coreweave更好的融资结构，利息低得多。 他们现在有来自$META和$MSFT的约$460亿+订单积压，这两家是最盈利的超大规模云服务商之一，且没有像Coreweave那样直接关联OpenAI的风险传染。 与其他公司不同的是；他们的其他公司也有增值（Clickhouse股权增值：avride机器人出租车规模化；toloka三位数增长）。 从我来看：Nebius是明显的赢家。 然而，当前宏观环境不利于整个板块的短期持有者，指数下跌7%。 尤其是如果他们在ATM活跃时买入的话。 长期来看，当他们规模化后获益时（例如$NBIS 2026年Q4，是的，甚至$IREN），收益将是巨大的。
英文原文
My thoughts on $NBIS, $IREN, $CRWV and the current Neocloud market. One of them ends up as the next AWS in 5 years: My guess it’s Nebius. It's not winner takes all (DigitalOcean is there with Amazon), but there's clearly superior structures and likely winners. The downside: -> Low chance of rate cuts from Iran conflict. ->Broader market doesn't appear to want to fund the CapEx cycle. But want to reap the benefits With $IREN: We get it, 4.5GW = X revenue. But who is funding the GPUs? Whoever is buying into the $6,000,000,000 ATM right now. The winners will be whoever enters after holders get fully diluted. The reality is, they don't have enough funding to monetize their capacity through GPUs without colo models. And they didn't find other financing methods, so they went through ATMs because of a cult community that will buy into anything they sell. However, I agree it will be accretive long term. Just not as much for the retail buying in now. With $CRWV: They did everything right... $NVDA backing. Hyperscaler clients... But they financed completely wrong. Now, $1.5B+ yearly debt interest is eating Coreweave alive and cuts into FCF. Almost like credit card debt, Coreweave gets a job to pay off that debt, but eventually, the debt interest is too high that working doesn't really cover that and expansion too. If any company goes down, $CRWV is the first to go the massive debt load and interest. With $NBIS: They're doing as much as they can right... $NVDA funding $2B to fund capex. Convertible note offerings (convertible note short hedging is annoying for short term price appreciation). But this is the best way to do financing structures with much lower interest than Coreweave. They now have ~$46B+ in backlog from $META and $MSFT, two of the most profitable hyperscalers out there, without direct OpenAI linked contagion like Coreweave. And unlike others; there’s appreciation from their other companies (Clickhouse equity appreciation: avride robotaxi scale up; toloka triple digit growth) From my take: Nebius is the clear winner. However, current macro environments does not favor short term holders across the board with indexes dropping 7%. Especially so if they're buying into active ATMs. Long term, the benefits when they scale up eg. $NBIS Q4 2026 (yes, even $IREN), will be immense.
-
SIVE作为CPO光源瓶颈被低估，2500万市值太低，坚信将成下一个AXTI
这就是为什么在进入交易前需要有信念。如果你了解$SIVE在CW激光领域为Jabil、$MRVL Celestial等共封装光学(CPO)供应，就会明白2.5亿美元市值作为光源瓶颈来说太低了。我们高度相信，一年后这会像$AXTI一样，因为这将是CPO规模化扩展的架构范式。在上涨过程中我不关心波动，因为我对光子学的发展如何演绎有坚定的信念。
英文原文
This is why you need to have conviction before entering a trade. If you knew $SIVE positioning in the CW laser space to Jabil, $MRVL Celestial, and others for CPO. $250M MC as the light source chokepoint would be a joke. High confidence we’ll see this end up like $AXTI in a years time since it this will be the architectural paradigm for cpo scale up. Don’t care about volatility in the way up because I have conviction in how this plays out with photonics.
-
看多稳懋3105：位于LEO卫星/光学/人形机器人三大高增长领域垄断瓶颈点，战略价值被市场低估。
我在稳懋（Win Semi）(3105.TWO) 4.1B市值时做多。我认为市场忽略了这家全球最重要的晶圆厂之一（在台积电$TSM之外）。我认为他们的战略定位远超40亿美元市值。他们几乎位于所有主要瓶颈点：-> SpaceX Starlink 低轨卫星供应链。-> 作为$AVGO、$LITE、$MTSI、$SIVE的磷化铟(InP)光学收发器晶圆厂。-> 作为人形机器人的身体/眼睛，是用于飞行时间(TOF)激光器的砷化镓(GaAs)晶圆厂，可能用于Boston Dynamic Atlas。-> 来自之前与联发科(MediaTek)/高通(Qualcomm)/$AAPL的业务遗留。但稳懋似乎受到legacy业务的拖累（如$SOI），光学是其最大增长向量之一。然后...稳懋拥有几乎任何晶圆厂中最大的TAM扩张/收入加速：随着：LEO、人形机器人/连续波(CW)激光、800g、1.6t、3.2t光学收发器在未来几年大幅增长。特别是博通(Broadcom)作为其锚定客户（$AVGO持有约5%的稳懋股份）。$NVDA不在乎谁制造激光器，无论是$LITE还是$COHR。他们只在乎是否有足够的供应。供应不够。-> CW激光需求可能呈抛物线增长。（他们制造像$SIVE设计的激光器）-> LEO卫星需求（SpaceX Starlink）可能呈抛物线增长。-> 人形机器人需求可能呈抛物线增长。因为稳懋位于三个最前沿和增长最快行业的半垄断瓶颈点：光子学/AI、机器人/人形机器人和太空。特别是光学TAM爆发：稳懋2027年远期收益大约在35倍范围，我认为这是保守估计，远期估值最终会非常便宜。稳懋将大大受益于TAM扩张和加速的收入增长。当然：稳懋会赢。所以我做多稳懋。
英文原文
I am long Win Semi (3105.TWO) at $4.1B MC. I believe markets are sleeping on of the most important foundries in the world (aside from $TSM). IMO their strategic positioning exceeds far beyond $4B MC. They sit in almost every major chokepoints: -> In the SpaceX Starlink LEO supply chain. -> As $AVGO, $LITE, $MTSI, $SIVE InP foundries for optical transceivers -> then as the body/eyes of humanoids as the GaAs foundry for TOF lasers possibly mapping to Boston Dynamic Atlas -> With legacy from MediaTek / Qualcomm / $AAPL from their previous business. But Win appears to be bottom of the legacy drag (like with $SOI), with optical as one of their largest growth vectors. Then... Win has the largest TAM expansion/revenue acceleration out of almost any foundry: With: LEO, humanoids / CW laser, 800g, 1.6t, 3.2t optical transceiver massive ramp up over the next few years. Especially with Broadcom as their anchor client ( $AVGO owns ~5% of Win). $NVDA doesn't care who makes the lasers, whether it's $LITE or $COHR. They just care if there's enough. There's not enough. -> Demand for CW lasers will likely go parabolic. (they make the lasers that companies like $SIVE designs) -> Demand for LEO satellites (SpaceX Starlink) will likely go parabolic. -> Demand for humanoids will likely go parabolic. As, Win Semi sits as a semi-monopoly chokepoint in the three most frontier and fastest growing industries for photonics/AI, robotics/humanoids, and space. Especially with Optical TAM explosion: Win fwd earnings for 2027 roughly in ~35x range, I do think this is sandbagging it and forward multiples will end up dirt cheap. Win will largely benefit from TAM expansion and accelerated revenue growth. Of course: Win will win. So I am long Win.
-
警示VCX ETF存在严重溢价泡沫，建议回避
关于$VCX的警告。 这是一只权重偏向Anthropic、Databricks、OpenAI、Anduril和SpaceX的热门ETF。 问问你自己：SpaceX的估值是$META的23倍，合理吗？ 该股本月上涨1288%：VCX目前交易价格约为其NAV的20倍。也就是说存在1500-2000%的溢价？ 这就相当于以34万亿美元的价格买入SpaceX。或以16万亿美元买入OpenAI。 你会做空它吗？ 不能：它的流通股极少，且缺乏对冲工具。 但你应该买入吗？ $VCX现在看起来就是"博傻理论"的典型案例，你只是在玩接鼓传花的游戏。
英文原文
Warning about $VCX. A very popular ETF with weighting toward (Anthropic, Databricks, OpenAI, Anduril, and SpaceX). Ask yourself this: Is SpaceX is worth 23 times $META valuation? With the stock going up 1288% this month: VCX is trading at roughly ~20x its NAV. So 1500-2000% premium? This is like buying SpaceX at 34 Trillion USD. Or OpenAI at 16 Trillion USD. Would you short it? No: It's extremely low float and lack of options for hedging. But should you buy into it? $VCX looks like greater fools theory right now and you’re playing hot potato.
-
角川书店(Kadokawa)年内涨幅15.93%，用户戏称靠动漫品味跑赢科技大盘。
很多人问我最喜欢的三部动漫： 1）《路人女主的养成方法》（《路人女主的养成方法》）- 角川书店（TYO: 9468） 2）《欢迎来到实力至上主义教室》- 角川书店（TYO: 9468） 3）《Re:从零开始的异世界生活》- 角川书店（TYO: 9468） 角川书店年内至今涨幅：15.93% 经验教训： 如果你对动漫有好的品味，就能跑赢$GOOGL和$META以及大盘指数吗？
英文原文
Lot of people were asking about my top 3 anime: 1) Eminence in Shadow - Kadokawa (TYO: 9468) 2. Classroom of the Elite - Kadokawa (TYO: 9468) 3. Re:Zero - Kadokawa (TYO: 9468) Kadokawa YTD: 15.93% Lesson learned: You can outperform $GOOGL and $META + indexes if have good taste in anime?
-
认为 VLN 低于账面价值，且市场对以色列相关股票偏冷，但估值仍显著低估。
关于 $VLN，我不觉得自己错了。但市场显然不同意我的看法。 我学到了两件事： - 一些股票，像日本股，确实会交易到远低于账面价值 - 市场不太喜欢以色列股票，比如 $ETOR，我在那上面也亏了不少。 我还是觉得它被低估了。净资产（2.926 亿美元纯现金）对比 1.32 亿美元市值。 净资产是 1.05 亿美元，而且零负债。 那企业价值不就大概 2700 万美元？对于一家 2025 财年营收 7060 万美元、GAAP 毛利率 60% 的成长公司来说。 所以估值看起来很离谱。但股票市场偏偏就是这么运作的。
英文原文
As for $VLN, I don't think I'm wrong. But markets clearly disagree with my opnion. I've learned two things: - Some stocks like Japanese ones can trade way under book value - Markets don't like stocks from Israel like $ETOR for some reason (lost a lot on that too). I still think it's undervalued. Net assets ($92.6M pure cash) vs. $132M MC. Net assets is $105M with zero debt. So EV is ~$27M? For a growing company with 60% GAAP gross margins on $70.6M FY 2025 revenue. So valuation looks stupid. But somehow stocks operate that way.
-
认为 LASR 与无人机股不同，因定向能武器和金穹扩展带来新催化。
@frsinvesting $LASR 和无人机板块的股票不一样，因为定向能武器现在因为 Iron Beam 的实战使用，拥有巨大的应用场景。 关键问题是它们是不是金穹扩展项目的承包商。我现在就在等这方面的消息。
英文原文
@frsinvesting $LASR is different than drone sector stocks, since energy directed weapons have a massive use case now from Iron Beam live usage. The main question is if they're the contractor for golden dome expansion or not. I'm just waiting for news surrounding that.
-
认为 SIVE 和 IQE 可能都能成为 10 倍股。
@_Livesofourdays 是啊，$SIVE 和 $IQE 对我个人来说都有可能成为 10 倍股。 很期待看看它们最后会走到哪里。
英文原文
@_Livesofourdays Yeah $SIVE and $IQE are both possible 10x’s for me personally. Excited to see where they head
-
说 AAOI 自己已经涨了 3 倍以上，对其仍然高信念。
@usbastats $AAOI 自从我做多之后已经涨了 3 倍多……我只是因为名字太多，一时忘了它。 但对 $AAOI 依然是高信念。
英文原文
@usbastats $AAOI is up over 3x since I went long… I just forgot about it since there were too many names But yes high conviction with $AAOI
-
讨论 Google TurboQuant 对 DRAM/NAND 的影响，认为更像效率提升而非需求坍塌。
谷歌的 TurboQuant…… 以及它对 $SNDK、$MU、海力士和其他公司的影响： 它做的事情是： -> KV cache 内存占用减少 6 倍 -> H100 GPU 速度提升 8 倍 它本质上是一个压缩算法。 那现在问题来了……它会把存储压下去吗？ -> 大概率不会。 不过这也许对 $ARM 和其他公司是利好，因为你可以本地跑 AI，而不是依赖 DRAM-heavy 的数据中心。 但话说回来： -> 这基本上就是 DeepSeek round 3。你能让算法更高效，但这并不会替代存储或者 GPU。 -> 它可能会在结构上略微降低 DRAM 需求。 -> 而且到目前为止，好像也只在 Gemma、Mistral 和 Llama-3.1 这些小模型上测试过（而且那篇论文已经发了一年了） 另外，市场还把 DRAM 和 NAND 混在一起看……这个算法压的是 KV cache（DRAM），并不会对 NAND 存储有什么作用？ 不管怎样： 算法总会变得更高效。大家老说杰文斯悖论，这没错，因为这只是把用途规模继续放大。 真正该看的还是超大规模云厂商的 CapEx 预期，而不是让事情更高效的 Google 算法。 我更觉得这是个叙事层面的逆风，而不是对盈利有实质影响。
英文原文
Google's TurboQuant... And it's effect on $SNDK, $MU, SK Hynix, and others: What it does: -> 6x reduction in KV cache memory footprint -> 8x Speedup on H100 GPUs It's a compression algorithm. Now... Will it beat down memory? -> Prob not. Implications might be bullish for $ARM and others though where you can run AI locally, rather than DRAM heavy DCs. However: ->This is basically DeepSeek round 3. You can make algorithms more efficient. But that doesn't replace either memory or GPUs. -> It could structurally (and slightly) reduce DRAM demand. -> think it's only been tested on small models so far like Gemma, Mistral, and Llama-3.1 (and paper's been out for a year) Also, markets conflated DRAM with NAND... this algo compresses the KV cache (DRAM). Doesn't do anything to NAND storage? Regardless: Algorithms will always get more efficient. People keep saying Jevons Paradox, which is true since this just scales use cases. Main thing to look out for is hyperscaper capex projections, not Google Algorithms that made things more efficient. Feels more like a narrative headwind than anything material to earnings.
-
说自己对 SIVE 高信念，同时也看好 ARM、AAOI。
@Jornka329996 谢谢！$SIVE 对我个人来说是高信念标的。即便现在市值大概只有 4 亿美元左右，我也是这么看。 $ARM 和 $AAOI 这种价位下，长期看起来也很有潜力！ 这周我还在看另外两家大盘公司。
英文原文
@Jornka329996 Thanks! $SIVE is high conviction for me personally. Even so at these levels around ~$400m MC. $ARM and $AAOI seems promising long term at these levels too! I did have two more large cap companies I'm looking at this week.
-
说超大规模 ASIC 正在吞掉推理市场，如果只剩训练会有麻烦。
@yuntungshieh 超大规模云厂的 ASIC 正在把推理市场吃掉。 如果你被稀释得只剩训练业务，那可能就麻烦了。 另外我还有别的想法，但不想公开发出来。
英文原文
@yuntungshieh Hyperscaler ASICs eating up inference market. If you’re diluted down to just training could be trouble. Also there’s something else but don’t want to post it publicly
-
认为 AAOI 的异动大概与期权流有关。
@O_bhaina $AAOI 的走势大概还是和期权流有关。
英文原文
@O_bhaina Probably option flow related on $AAOI
-
认为它可能五个月后再翻倍。
@vatsalBN 大概五个月后还能再翻倍一次。
英文原文
@vatsalBN Probably see it doubling again in 5 months
-
说 ARM 其实也是一只 1400 亿美元市值的“大盘股”。
@Rationalmind__ 兄弟，$ARM 现在可是 1400 亿美元市值的公司。
英文原文
@Rationalmind__ Bro $ARM is a $140B MC company
-
认为 TTD 和 RDDT 是不错的修复多头，但市场似乎不再在意软件增长。
@HAkbarnataj $TTD 和 $RDDT 都是不错的修复多头。 不过感觉市场现在已经不怎么在意那些年增 20% 到 50% 的软件公司了？
英文原文
@HAkbarnataj $TTD and $RDDT are good recovery longs. However feels like markets just don’t care about software companies growing 20-50% Y/Y anymore?
-
用日文感叹自己很久没对大盘股这么兴奋，并看好 ARM 的 AI CPU 收入扩张。
自从海力士和三星以来，我已经很久没有对这么大盘的股票这么兴奋了！ $ARM 的 AI CPU 收入扩张预期，真的太夸张了。
英文原文
@beauty_oe SKハイニックスやサムスン以来、大型株でこんなにワクワクしたのは久しぶり! $ARM のAI CPUの収益拡大予想、マジでヤバいね。
-
说自己不是因为 RISC-V 抢食而做多 ARM，而是看好其转向 AI CPU。
@piyush1337 我做多 $ARM 不是因为 RISC-V 会蚕食它的商业模式，而是因为它自己转向了 AI CPU…… 这真的是一个极其巨大的游戏规则改变，尤其是如果你看收入预测的话。
英文原文
@piyush1337 I didn't long $ARM bc of RISC-V cannibalizing their model, but pivot to AI CPUs themself... It's an incredibly massive game changer, esp if you look at revenue projections.