全部推文
共 6749 条 · 第 69 / 225 页 · 滚到底自动加载更多
-
瑞典媒体看不懂光子学
我真的笑了。这大概是我目前见过最搞笑的标题了，自从 $RPI 那次之后。 你会看到瑞典那些不懂技术的记者，对半导体 / 光子学一窍不通。 他们在分析 $SIVE，还告诉所有本地人（大多数持股者）赶紧卖掉，因为它只是个“meme”。 他们不懂围绕即将到来的 $JBL 1.6T 或 $MRVL CPO 放量的超大规模云厂商认证周期吗？ 也不懂它刚刚被验证为 $GFS 生态里的激光供应商，而且 $AMD 正在推动新的 CPO 需求吗？ > 一定是 meme。只能按 2025 年 TTM 资产负债表和旧营收预测来估值，对吧？ 至少这种波动有助于在明年 CPO / 1.6T 关键拐点之前，把控制权转给美国投资者 / 机构。
英文原文
I'm genuinely laughing. This is gotta be the funniest headline I've seen to date since $RPI. You have non-technical journalists in Sweden, with no understanding of semiconductors/photonics. Doing an analysis on $SIVE and telling all the locals (who hold majority) to sell, because it's a "meme". Don't understand hyperscaler qualification cycles from upcoming $JBL 1.6T or $MRVL CPO ramp? Or the fact they just got validated as the laser supplier in $GFS ecosystem with $AMD driving new CPO demand? > Must be a meme. It's only possible to value a company based on TTM 2025 balance sheets and old revenue projections right? At least this volatility helps transfer control over to US investors/institutions before CPO/1.6T inflection point next year.
-
SIVE 约 2B 比较合理
@itai_angel 我觉得大约 20 亿美元是合理的。 这只是来自不懂技术的记者推动本地人抛售所带来的短期波动。 这只是股份从散户转移给西方机构 / 投资者的过程。我不会期待它天天都是绿的。
英文原文
@itai_angel Around ~$2B sounds reasonable to me. It's just short term volatility from non-technical journalists scaling all the locals to sell. This is just the process of how shares transfer from retail to Western institutions/investors. Not expecting green every day.
-
本地人抛售对机构有利
他们其实是在帮助美国机构，鼓励本地投资者卖出。 这原本是一只主要由本地散户持有的股票。它最终只会把控制权转给西方机构。 我不太明白“meme stock”怎么会去支撑 $JBL 1.6T、Ayar、$MRVL 以及其他 CPO / 光子项目，哈哈。
英文原文
They're actually doing good work for US institutions encouraging local investors to sell. It was a majority retail-locally owned stock. It will just transfer control over to institutions in the West. Not sure what "meme stock" powers $JBL 1.6T, Ayar, $MRVL, and other CPO/photonics programs lol.
-
AMD/GFS 的 CPO 反而利好 SIVE
$AMD 和 $GFS 的 CPO 公告，对 $SIVE 的影响可能比市场预期更大。 在这个消息下，$SIVE 激光很可能会支持 AMD 的 CPO 项目。 可能有两条路径： 1. Enosemi（AMD 在收购后做的内部 PIC 设计）。Enosemi 的芯片小片由 GF 代工，但对于 ELS，$AMD 可能会从多家供应商采购，而 $SIVE 可能是底层多来源激光源。 2. Ayar（AMD 在 2026 年 3 月投资，Series E）。Ayar 的 SuperNova 光源已经和 $LITE 一起使用了 $SIVE 的 DFB 激光阵列。 由于时间线和上个月那次巨额融资，Ayar 的 SuperNova 很可能是 MI500 在 2027 年的第一代 CPO 路径。那条路径已经把 $SIVE 设计进去了，而且它和 $LITE 都出现在 $GFS 的幻灯片里。 Enosemi 也许会在 2028 年及以后几代里更重要？ 不管怎样，$AMD 通过 Enosemi / Ayar 还是需要激光器来支撑他们 2027 年的 MI500 上市…… 看起来 Sivers 很可能会作为光源，支持 AMD 的 CPO 项目，因为它已经被设计进 Ayar 了。 $AMD / $GFS 这条线对 Sivers 的重要性看起来很大。
英文原文
The $AMD and $GFS CPO announcement is probably bigger than markets expect for $SIVE. With the news, it's likely $SIVE lasers power $AMD's CPO program. Either through two potential paths: 1. Enosemi (AMD's in-house PIC design post-acquisition). Enosemi's chiplets are fabbed at GF but for the ELS, $AMD could source it from multiple players with $SIVE as the underlying multi-source laser source. 2. Ayar ( $AMD invested in March 2026, Series E). Ayar's SuperNova light source already uses $SIVE DFB laser arrays alongside $LITE. Ayar's SuperNova is the most likely first-gen CPO path for MI500 in 2027 given timelines and the enormous fundraise last month. That path already has $SIVE designed in alongside $LITE and they both appear with $GFS's slide. Enosemi becomes more relevant for 2028+ generations? Regardless, $AMD through Enosemi/Ayar needs lasers for their 2027 MI500 rollout... It seems likely Sivers ends up powering $AMD's CPO program as the light source since they're designed into Ayar. The $AMD / $GFS materiality looks large for Sivers.
-
只是表示同意
@itsDanielWu basically
英文原文
@itsDanielWu basically
-
LPK 是玻璃基板卡点
现在很多人不停问我对 $LPKK / $LPK 的看法。 是的，我之前说过，它们就像 LIDE（激光诱导深蚀刻）里玻璃核心基板的卡点。 已知最大的合作伙伴是 $ONTO（LPKF 和 Onto 的计量检测合作，用于玻璃核心的大规模生产）。 至于市场份额："超过 80% 的全球主要客户都选择了 LPKF 设备" 用于流程验证。 所以这大概包括： - Samsung Electronics / Electro-Mechanics - $INTC（收到一家领先芯片制造商的大单……在 2020 年初安装了第一套 LIDE 系统……现在又下单更多 LIDE 系统以开始量产） - SKC（Absolics） - $GLW、AGC、Schott。 - Nippon Electric Glass。 当然，这里说的是评估，所以实际放量时那个 80% 可能会更低。 再说一些我自己的前瞻市盈率计算： - 2027：大约 11-12.5 倍，2028 大约 7.8 倍，看起来非常有吸引力。 - 总现金：约 1000 万欧元，债务大约 300 万欧元。负债权益比：约 3.8% 资产负债表非常干净、轻资产，没有像 $SHMD 那样的稀释悬压。 市值约 3.62 亿美元，我的结论是：我认为上行空间很大，迟早会看到机构买入这个名字。 即便 80% 的玩家最终能设计出别的方案，只要有一部分转化，市值也会很有意义。 这可能几个月前还偏早，但玻璃核心路线图现在正像 CPO 一样加速。 披露：我确实有仓位。这只是我的想法。 X 上的人确实做了功课。
英文原文
People nonstop ask me about $LPKK / $LPK for my opinion Yes, I mentioned they're like a chokepoint for glass core substrates for LIDE (laser induced deep etching) way back when. Biggest known partner is $ONTO (LIDE with Onto metrology for glass core mass production). Then as for market share: "more than 80% of customers among major global players have selected LPKF equipment" for process validation. So that probably includes: - Samsung Electronics/Electro-Mechanics - $INTC (Receives a Major Order from a Leading Chip Manufacturer... installed a first LIDE system at the beginning of 2020... now ordered further LIDE systems to start volume production) - SKC (Absolics) - $GLW, AGC, Schott. - Nippon electric glass. Of course this is evaluation, so that 80% could be lower in actual ramp. As for some personal FWD P/E calculations: - 2027: ~11-12.5x and ~7.8x for 2028, which looks very compelling. - Total Cash: ~€10.0M, debt was around ~€3.0M. debt to equity: ~3.8% So very clean-asset light balance sheet, no dilution overhang like $SHMD. ~$362m MC, conclusion: great upside long imo, hard to see institutions not buying this name down the road. Even if the 80% of players managed to design another way, even a fraction would probably be very material to the MC. It was probably a bit early few months ago, but glass core roadmaps have been speeding up like CPO. Disclosure: I do have positions. This are just my thoughts. People on X did their homework.
-
好奇你的大仓位
@frsinvesting 出于好奇，你最大的仓位是什么？
英文原文
@frsinvesting what were your biggest positions out of curiosity?
-
变压器瓶颈已成国家行动
哈哈，我之前说过变压器瓶颈和像 $HPS.A 这样的上游供应链吧？？ 今天新的总统行政行动： “国家设计、生产和部署大规模电网基础设施的能力，包括 TRANSFORMERS……已经非常受限” “根据第 14156 号行政令宣布国家紧急状态，以扩大本土电网基础设施的开发、制造和部署能力” 现在已经是美国为开发供应链、向这个领域倾注资金的国家紧急状态了。
英文原文
LOL what did i say about transformer bottlenecks and upstream supply chains like $HPS.A ??? New Presidential Action Today: "The Nation’s capacity to design, produce, and deploy large-scale grid infrastructure, including TRANSFORMERS ... is dangerously limited" "National emergency under Executive Order 14156 to expand the domestic capability to develop, manufacture, and deploy grid infrastructure" It's now a US National Emergency to develop supply chains and pour money into the sector for expansion.
-
ARM 因 CPU 瓶颈受益
在新的瓶颈重新回到 CPU 的背景下，我看好 $ARM。微软的材料里也显示，编排 / RAG 这类东西需要 CPU。 但我预测本地推理的一部分会越来越多地由 CPU 来处理……随着 Gemma 之类的模型未来变得更轻量。 不是每个机器人都需要解开宇宙的奥秘。 数据中心会需要海量的传统 CPU 算力（AWS Graviton、$GOOGL Axion 和 $MSFT Cobalt 都是 ARM 架构）。 $META 和 OpenAI 也是 AGI CPU 的买家。 而 AI 还会继续下沉到边缘端。 150 亿美元的年收入目标……现在看起来越来越合理了？
英文原文
Bullish on $ARM, given the new bottleneck shifting back to CPUs. MS shows stuff like Orchestration/RAG requiring CPUs. But I'm predicting parts of localized inference to be handled by CPUs more and more... as models like Gemma get lightweight in the future. Not every robot needs to be able to solve the mysteries of the universe. Data centers will need an astronomical amount of traditional CPU compute (AWS Graviton, $GOOGL Axion, and $MSFT Cobalt), which are all ARM based. $META + OpenAI are also buyers of the AGI CPU. And AI will flow down to edge. $15B annual revenue target.. Starting to look reasonable?
-
IREN 仍是差价
现在还有人继续看多像 $IREN 这种 F 级 neocloud，真是奇怪。 在 60 亿美元稀释悬压下？ 而 $NBIS 现在和 $ORCL、$MSFT 一起看起来都已经去风险了？ https://t.co/vIdzdk4Hqi
英文原文
How are there people still long on an F-tier Neocloud like $IREN. Amid a $6,000,000,000 dilution? When $NBIS is up there with $ORCL and $MSFT looking derisked? https://t.co/vIdzdk4Hqi
-
SIVE 可能明年就到 10B
@88magalhaes 对，我之前还觉得 $SIVE 两年内能到 100 亿美元……但也许明年就真的有可能了？ 市场通常会提前 12 个月定价，所以如果 2028 年有巨大的量产爬坡预期，那么 2027 年就有可能先反映出来。
英文原文
@88magalhaes Yeah I thought $SIVE could reach $10B in 2 years time... but maybe by next year it's a real possibility now? Markets usually look forward 12M, so if there's massive 2028 volume ramp projections, it's a possibility in 2027.
-
Win Semi 去风险 SIVE 扩产
@JerryOBrie16396 Win Semi 负责 $SIVE 的量产爬坡。 而 Win 早就是 $AVGO 和 SpaceX 供应链里一个很大的、成熟的代工厂，所以这部分扩产已经去风险了。
英文原文
@JerryOBrie16396 Win Semi handles volume ramp for $SIVE. And Win is a massive established foundry for $AVGO and SpaceX supply chains already. So that scaling is de-risked.
-
连续催化都对 SIVE 有利
这真的是巧合，催化剂一个接一个（$JBL）之后是纳斯达克上市，然后是（$GOOGL + $MRVL），再然后是（$GFS），再然后是（$GFS + $AMD）。 再加上 $NVDA 在到处投资 CPO。 我只是 OFC 之后才做多，没想到后面会有这么多事情，哈哈。
英文原文
It's legit coincidence that it's catalyst ( $JBL ) after catalyst (Nasdaq listing) after catalyst ( $GOOGL + $MRVL ), after catalyst ( $GFS ), after catalyst ( $GFS + $AMD ). Alongside $NVDA investing into CPO everywhere. I just timed my long after OFC but did not expect all of this lol.
-
AMD 进军 CPO 对 SIVE 很利好
什么疯狂时间线？ 今天，$AMD 宣布要进入 CPO，并和 $GFS 合作。 这对 $SIVE 非常利好，因为 CPO / SiPH 需要外部激光光源。 而且……猜猜 GlobalFoundries CPO 生态里那两家公开激光供应商是谁？ $SIVE（大约 9.35 亿美元）和 $LITE（大约 640 亿美元） Sivers 的激光现在看起来真的可能会出现在未来所有超大规模云厂商的 CPO / 1.6T 部署里。 所以……如果 Sivers 能拿下为 $AMD Instinct MI500 的 CPO 方案提供激光阵列的位置。 （这看起来很有可能，因为目前列出的激光名字只有三个，而且超大规模云厂商喜欢多来源） 这可能意味着 Sivers 会迎来市场还没定价进去的大规模高量产。
英文原文
What a crazy timeline? Today, $AMD announced it's pushing into CPO and collaborating with $GFS. This is highly bullish for $SIVE since for CPO/SiPH, it requires external laser light source. And... guess who are the two public laser suppliers in the GlobalFoundries CPO ecosystem? $SIVE (~$935m) and $LITE (~$64 billion) Sivers lasers feel like they could be everywhere now in upcoming hyperscaler CPO/1.6T deployments. So... if Sivers secures the spot to provide the laser arrays for the $AMD Instinct MI500's CPO solution in 2027. (Which seems highly possible given there’s only three laser names listed + hyperscalers like to multi-source) This might represent massive, high volume production for Sivers that markets haven't pricing in.
-
ARM 股价涨得不错
@beauty_oe $ARM の株価が大きく上がってよかったです！AI向けCPUの予測が凄まじかったので、いずれ株価もそれに追いつくだろうと思っていました。
英文原文
@beauty_oe $ARM の株価が大きく上がってよかったです！AI向けCPUの予測が凄まじかったので、いずれ株価もそれに追いつくだろうと思っていました。
-
媒体叙事忽略了背景
现在媒体基本就是这么运作的。 我在伯克利的时候就亲自做过 RISC-V 架构实现，而且现在还持有活跃专利。 frontier 实验室和 $NVDA 以前都找我去带一些 AI 团队。 结果“专家分析师”对我背景的唯一描述居然是“WSB meme trader”，哈哈。
英文原文
That’s basically how media works nowadays. I literally pioneered RISC-V architecture implementations back when I was at Berkeley and still hold active patents. Frontier labs and $NVDA tried recruiting me to lead some AI teams. Yet the only mention of my background is “WSB meme trader” by “expert analysts” lol.
-
与光子无关
@LepoulpePoulpo 深度学习领域，不是光子相关。
英文原文
@LepoulpePoulpo Deep learning field, not photonics related.
-
喜欢帮助救助猫狗
@ravietweet 我很喜欢 DPS rescue，他们会帮助那些原本要被安乐死的猫狗，并努力帮它们找家。 如果你真的去做，我会偷偷帮你把捐款配平。
英文原文
@ravietweet I like DPS rescue, they help cats/dogs that were set to be put down and try to find them homes. I’ll try and secretly match your donation if you end up doing anything.
-
不要只看 meme 标签
就在一个月前，记者和媒体还在淡化我对 $SOI 和 $RPI 的 thesis。 他们引用我的 WSB 标签，把它说成是“注定要崩的 meme 股”，却没有分析底层 thesis。 结果这两只都涨了 100%，而且把涨幅守住了。 $SIVE 现在也是同样的情况。 我从来不靠权威说话……但既然他们想走这条路： 我就很好奇……他们的叙事里为什么总把我在 Nature 之类的地方发表过、被引用上千次的基础 AI 论文背景漏掉？ 又或者为什么今天 X 上那些看多 $SIVE 的分析师里，有些人是 UC Berkeley 的 Ph.D.，而且本身就在光子学领域发文？ 反而总是英语系或非技术背景的毕业生跑去媒体上，对 CPO 超大规模云厂商认证周期和供应链映射发表最强硬的意见。 我总觉得机构最不喜欢的，就是散户投资者在 X 上真的懂自己在说什么。 一个 thesis 应该靠本身的价值生死，而不是发言人是什么权威。
英文原文
Just 1 month ago, journalists and media tried downplaying my $SOI and $RPI thesis. As “meme stocks that were set to crash” citing my WSB tag without analyzing the underlying thesis. They’re both up 100% and held their gains. Same is happening to companies like $SIVE. I never argue from authority… but since they’re going that route: I’m curious why… in their narratives they just leave out the fact I published fundamental AI papers in places like Nature with thousands of citations? Other the fact other analysts on X positive about $SIVE today have Ph.Ds from places like UC Berkeley and publish in the photonics space? It’s always the English or non-technical graduates that go out in the media and have the strongest opinion about CPO hyperscaler qualification cycles and supply chain mapping. It just feels like institutions hate it when a retail investors on X know what they’re talking about. A thesis should live and die based on merit, not the authority of who is comes from.
-
主题框架
CPO / 1.6T。 基板（玻璃核心、InP、SoI）。 CPU / ASIC。 变压器。
英文原文
CPO / 1.6T. Substrates (Glass Core, InP, SoI). CPUs / ASICs. Transformers.
-
GFS 光子/CPO 生态名单
Global Foundries 的光子 / CPO 生态名单： 1. $GFS - 305 亿美元 2. $CDNS - 859 亿美元 3. $SEIGY - 2172 亿美元 4. $SNPS - 840 亿美元 5. $KEYS - 573 亿美元 6. $ATEYY - 1304 亿美元 7. RoboTechnik（ficonTEC）- 114 亿美元 8. $GLW - 1412 亿美元 9. $LITE - 638 亿美元 10. $SIVE - 9.5 亿美元 11. $FN - 247 亿美元 12. $ASX - 633 亿美元 这些都是公开交易的名字。 按他们的演示材料做一个等权做多这个生态，可能也不是个坏主意？ 我还是觉得好笑，所有上市公司都在几十亿美元到几百亿美元市值，而这里却夹着一家小小的瑞典激光公司，和 $LITE 并排出现。
英文原文
Global Foundries photonics/CPO ecosystem list: 1. $GFS - $30.5B 2. $CDNS - $85.9B 3. $SEIGY - $217.2B 4. $SNPS - $84B 5. $KEYS - $57.3B 6. $ATEYY - $130.4B 7. RoboTechnik (ficonTEC) - $11.4B 8. $GLW - $141.2B 9. $LITE - $63.8B 10. $SIVE - $950m 11. $FN - $24.7B 12. $ASX - $63.3B For publicly traded names. Equal weighted long on the ecosystem from their presentation might not be a bad idea? I still find it funny how everything publicly listed is trading in the tens of billions. Then there's some small Swedish laser company there next to $LITE.
-
IBKR 业务也会有点帮助
@pennycheck 说实话，这对 $IBKR 来说也不算坏主意，我会去听他们的财报哈哈！ 但我确实认为，零售账户带来的增量相对机构成交量来说只是小幅提升……不过交易费 / 保证金使用这些长期可能会像 $HOOD 那样变得有意义。
英文原文
@pennycheck Honestly, not a bad idea for $IBKR, will tune into earnings lol! But I do think all the retail accounts only account for a slight boost relative to institutional volume... the trading fees / margin usage might be material like $HOOD over time.
-
TTD 价格也并非没考虑风险
@DudeWhoInvests 对啊……X 上当时有很多人都做多 $TTD 那些价位。 我确实觉得 22 美元左右已经有点把衰退 / 颠覆计进去了，但他们看起来还行。
英文原文
@DudeWhoInvests Yeah… there were a lot of people on X long on $TTD at those levels. I do think $22 is kinda priced for distress/disruption, but they seem to be doing okay.
-
ALMU 和量子点还早
@Lennr1chards 我觉得 $ALMU 和量子点现在还太早了，所以你大概得再等几年。 QD laser 是其他量子点名字，但它们似乎已经因为 Fujitsu 而去风险化了一些。
英文原文
@Lennr1chards I think $ALMU and quantum dot is too early, so you’ll probably need to wait a few years. QD laser were other quantum dot names, but they seem more derisked from Fujitsu.
-
TTD 的短线反弹空间
你能相信 $TTD 不久前还在 139 美元吗？ 我觉得他们转向广告之后，短期 ChatGPT 货币化这条线被低估了。 我确实在 22 美元附近有一些短线仓位，把它当作底部均值回归的机会，不过这绝对不是高 conviction。
英文原文
Can you believe $TTD was once $139 not too long ago? Think the short-term ChatGPT monetization angle is undervalued given they’re moving to ads. I do have short term positions at $22 as a potential mean reversion the bottom, but this is not high conviction by any means. https://t.co/7VMfVACrrl
-
AEHR 被重新定价
我说过了，$AEHR 正在起飞。 当你把市值做到一定程度，基本上就是机构在主导价格了。 如果你找到一家很多超大规模云厂商都在用的公司……它被重估几乎是注定的？
英文原文
Told ya. $AEHR is cooking. Once you reach a certain point with MC it’s mostly institutions driving price. If you find a name that a ton of hyperscalers use… it’s feels bound to get repriced? https://t.co/AHqpgbqHJK
-
已经去风险化
@NabQ321 没有。只是从积压订单和瓶颈需求来看，方向上已经去风险了。
英文原文
@NabQ321 Nope. Just directionally it looks derisked from backlog + bottleneck demand.
-
HPS.A 这种慢牛最能涨
像 $HPS.A 这种看起来很无聊的名字，最后反而涨得最猛？ 我在 186 美元时就说它看起来像一只稳定复利的名字，而现在两周已经到 241 美元了。 我觉得变压器 / 开关设备瓶颈在市场里被严重低估了。 https://t.co/XkI6kHdExh
英文原文
Boring names like $HPS.A end up cooking hardest? I said this looked like a steady compounder back at $186 and it’s already at $241 in 2 weeks. I think transformers/switchgear bottlenecks are underappreciated in markets. https://t.co/XkI6kHdExh
-
那个东西太怪了
@Himothee 那是个什么怪东西。
英文原文
@Himothee What is that abomination
-
HIMS 开始被注意到
看起来 $HIMS 终于开始被注意到了，今天涨了 +8.64%。 肽业务和收购都在起飞？ 让我有点意外的是，虽然基本面在改善，SI 还保持在 36%。 我很想看看多头和空头最后会怎么演。
英文原文
Looks like $HIMS is finally getting some attention, up +8.64% today. Peptide arc and acquisitions go brrr? Caught me by surprise why the SI is still 36% despite improving fundamentals. I’m very curious to see how this all plays out with the longs vs. shorts. https://t.co/CtpscjKIxl