供应链分析
产业链结构、上下游与瓶颈环节研究 · 共 1440 条 · 滚到底自动加载更多
-
强调他们做的是围绕 photonic memory 的机箱/外壳
@chucktheeegreat 就是所有人。 他们做的是围绕 photonic memory 公司的机箱/外壳。 你们全在胡闹。 https://t.co/kyJcrw2Hg9
英文原文
@chucktheeegreat Literally everyone. They make the chassis around photonic memory companies. You all are trolling. https://t.co/kyJcrw2Hg9
-
区分 PENG 的整机层与真正的光子内核 IP
说得很细。买 $PENG 去押 $MRVL 的“photonic memory”大概率会失望。 他们更像是 $SMCI 那一层的集成商，外加一些潜在的软件增值。 高毛利、真正的基础 IP 属于 Celestial 或 Lightmatter 这类公司。 他们做的是在机架里围绕 Celestial / Marvell 的 photonic memory IP 搭一个 2U 盒子。 这也还处在联合开发阶段。 散户只是把这些层级全搞混了，以为这家公司在做核心 photonic memory IP。 我不是在评论股价会不会涨，只是说它的基础业务确实有意义；如果能越过 sampling，向上的弹性会很大。 只是每次看到别人把“PHOTONIC MEMORY”挂在嘴边就很烦。 他们做的其实只是放核心 IP 的机箱，连 qualification 都还没到。
英文原文
Very nuanced. People buying $PENG for $MRVL "photonic memory" are likely to be disappointed. They're on the $SMCI integrator level with potential software add. The high margin, foundational IP belongs to companies like Celestial or others like Lightmatter. What they do is build a 2U box, inside the rack, around Celestial/Marvell's the photonic memory IP. It's also in the development collaboration stage. Retail just misunderstand the different layers and conflate them all as this company building core photonic memory IP. Not commenting on the potential price, but it's a legitimate idea for its base business. And upside would be material if they go past sampling. Just annoying when I get comments about "PHOTONIC MEMORY". Then they're building the chassis that the actual photonic memory IP sits inside and not even in qualification stage.
-
深度分析光子学供应链四只股票，强调TAM扩张和垂直整合逻辑，认为是长期核心持仓而非短期交易
当我看到像这样（而且很多）的来自散户投资者的评论时，我立刻认为他们缺乏技术深度。 我来逐一分析从$SIVE到$LPK的每家公司： 1. 光子学TAM（总可寻址市场）从$140亿增长到$1540亿，仅仅两年时间，而且它很可能会在2030年之后继续增长，因为它是下一代架构的首选。 它不会在1年内消失。它不会在3年内消失，这就是为什么$LITE的溢价越来越高，因为他们已经排产到2028年了。 $SIVE供应CW（连续波）激光器，高度绑定CPO（共封装光学）和现在的可插拔收发器，用于1.6T和3.2T... 适用于像$JBL、Ayar、Lightmatter、Lightelligence、$POET、$MRVL、Celestial和$AMD这样的预期公司。 这不是一个"交易"，它是下一代光子学的核心瓶颈和IP持有者。 而且作为未来几年的舒适持仓是很合适的，因为他们正在扩展成为下一个$LITE。 我个人看到的风险（既然他们已经与这么多玩家完成认证），主要是他们能从整体光学超级周期中捕获多少TAM（市场空间）。（以及Win Semi产能爬坡的潜在风险，但Win体量很大所以我可以放心睡觉）。 因为仅仅供应激光器不足以支撑估值。 是TAM向下扩展到制造整个ELS（外部激光源）或整个可插拔收发器，才让这些激光器公司如此有价值。 之后，他们可以向上垂直整合以扩大毛利率，像$COHR那样做激光代工甚至衬底级别的业务。 在我看来，这是非常不对称的风险/回报比，因为我们已经在$LITE身上看到过——他们从$20亿增长到$800亿。 2. $LPK——是下一代先进封装转向玻璃基板的最纯粹敞口，没有SKC Absolics混乱的财务问题。 几乎每家主要半导体公司从$INTC到三星都在采用玻璃基板。 $LPK基本上就是该瓶颈的$ASML，因为他们目前供应给全球约80%的玩家。 是的，设备供应商有"贸易周期"，比如$ASML，如果有更多晶圆厂资本支出，ASML就会增长。但如果出现低迷，这些往往表现不佳，也无法捕获之后发生的所有量产出货。 然而，如果市值是$6.5亿，他们正在实现$1-2亿收入，随着每个客户的量产出货增长，他们从玻璃基板周期中获得的收益可能会超过当前估值。 而且他们会有基本面支撑（因为越来越多的公司采用封装转型），这会维持他们的估值。 现在只是等待产能爬坡的时机而已。 3. $AAOI——这简直就是美国版的光子学$INTC。可以说Intel不是长期投资。 猜猜你们所有的光收发器是在哪里制造的？ 中国、泰国、马来西亚。如果你看看Innolight、Eoptolink、$FN和其他公司。 AOI正在建设最大的美国本土供应链，包括CW激光器代工以及800G、1.6T组装。 是的，这也有可插拔周期起伏。明年会有1.6T的浪潮，然后CPO会蚕食可插拔市场。 但由于他们在内部制造整个供应链，他们对其他细分市场有极大的选择性。而且像$NVDA的老一代GPU一样，会有主权数据中心对老一代可插拔收发器的需求，来自像$AAOI这样的厂商。 它很可能会继续上涨，因为它在2026年下半年达到每月$4亿+收入的目标。 一路上有很多不同的短期波动，比如$6亿的稀释。 4. $IQE——？？？这是西方世界最重要的外延晶圆玩家之一。 $MTSI不遗余力地偿还了IQE的债务，因为他们不能让IQE倒闭。$IQE还向$LITE供货。 目前世界在 Landmark评论提到的外延晶圆水平和InP衬底水平都存在瓶颈。 他们的财务数据虽然糟糕，但从原始账面价值和他们在整个西方供应链中的价值来看……完全支撑他们的估值。其他光学公司不会让他们核心的上游供应链倒闭。 因为这些价值数千万美元的材料出问题，会搞砸下游价值数十亿美元的产品。 再次强调，光子学是扩展AI所需的下一代架构。不是量子计算那种"还在开发中"的东西。 它就在这里，而且是$NVDA选择的架构。 如果几年后这些都涨了很多，我一点也不会惊讶。 那些认为"只是因为我提到了"就在3个月内一波流的 people，根本不知道他们在说什么。 机构最终会买入这些标的（就像Point 72买入$IQE那样），散户只会在估值已经涨了600%之后才知道。 真的应该在发表评论前做好研究： 这些都是需要深入供应链知识的前瞻性增长公司。
英文原文
When I see comments like this (and there are a lot) from retail investors: I immediately think they lack the technical depth. I'll walk through each one from $SIVE to $LPK: 1. Photonics TAM goes from $14B -> $154B In just two years time, and it's likely going to keep scaling past 2030 as it's the next generation architecture of choice. It's not going away in 1 year. It's not going away in 3 years, which is why $LITE premiums keep going higher since they're backlogged into 2028. $SIVE supplies CW lasers and is highly tethered to CPO and now pluggable transcivers for 1.6T and 3.2... For expected companies like $JBL, Ayar, Lightmatter, Lightelligence, $POET, $MRVL Celestial, and $AMD. This isn't a "trade", it's the core chokepoint and IP holder for the next generation of photonics. And it's a comfortable hold for the next few years as they scale to become the next $LITE. The risk I personally see (since they're already qualified with so many players), it's mainly how much TAM they can capture of the overall optical supercycle. (And potential risks with Win Semi volume ramp, but Win is massive so I can sleep tightly there). As just supplying lasers isn't enough to justify valuation. It's TAM expansion downward into making the entire ELS or entire pluggable transceiver that makes these laser companies so valuable. Then afterward, they can vertically integrating upward for gross margin expansion upward like $COHR into doing the laser fabs or even substrate level. And that in my view is a very asymmetric risk/reward ratio as we've already seen this done with $LITE as they went from $2B to $80B. 2. $LPK - Is the purest exposure, without the messy financials of SKC Absolics, as the next advanced packaging shift for glass substrates. Almost every single major semi company from $INTC to Samsung are adopting glass substrates. $LPK is basically $ASML of this chokepoint, since they supply to ~80% of the global players currently. Yes, there's "trade cycles" for equipment suppliers like $ASML, where if there's more foundry capex, ASML scales up. But if there's downturns, these tend to perform poorly, and don't capture all the volume ramp that happens after. However, if the MC is $650m and they're making $100-200M, revenue per costumer volume ramped, the amount they make from the glass substrate cycle will likely exceed current valuations. And they'll have baseline fundamentals (as more companies adopt the packaging shift), that keeps their valuation up. It's just a waiting game for volume ramp at this point. 3. $AAOI - This is literally $INTC but for America + Photonics. It's like saying Intel is not a long term investment. Guess where all your optical transcivers are made? China. Thailand. Malaysia. If you look at Innolight, Eoptolink, $FN, and others. AOI is building the largest Made in America supply chains for both CW laser fab, as well as 800g, 1.6T assembly. Yes, there are pluggable cycle ups and downs to this as well. There's going to be a wave for 1.6T next year, then CPO cannibalizes pluggables down the road. But since they make the entire supply chain in house, they have extreme optionality for other segments. And like $NVDA older gen-GPUs, there's going to be sovereign DC requirements for older gen pluggables from names like $AAOI. It's likely going to keep rising as it hits that $400m+/month revenue target H2 2026. There's just a lot of different short term volatility along the way like the $600m dilution. 4. $IQE - ??? It's one of the most important players in the Western word for epiwafers. $MTSI went out of their way to pay off IQE's debt because they can't have them going under. $IQE is also supplying to $LITE. The world is currently bottlenecked both on the epiwafer level from Landmark comments and InP substrate levels. Their financials were track but the raw book value, and value they hold to the entire Western supply chain... completely justifies their valuation. And other optical companies will not let their core upstream supply chain go under. As these tens of millions worth of materials would screw up tens of billions worth of downstream products. Again photonics is the next generation architecture required to scale AI. It's not Quantum where it's just "In development". It's literally here and the architecture of choice by $NVDA. I would not be surprised if all of these are a lot higher in 3-4 years time. People who think it's one and done in 3 months time "only because I mentioned it" don't know what they're talking about. Institutions would have bought up the name eventually (like Point 72 on $IQE) and retail would only find out after their valuations are 600% higher. Should really do the research before adding comments like these: These are all forward growth companies that require in-depth supply chain knowledge.
-
分析对华加征关税效果有限，需切断上游供应才有效。
@lorddaws 我认为中国不会太在意美国对廉价$LU LU仿品或袜子进口征收45%关税这种事。 如果你能通过切断台湾、欧洲、韩国、日本等地区向上游供应商的出口，来完全扼住他们整个人工智能/机器人/太空项目，那确实会有效果。
英文原文
@lorddaws I don't think China cares so much about US slapping 45% tariffs on cheap $LULU ripoffs or sock imports. If you completely freeze their entire AI/robotics/space programs by choking off upstream vendors over in Taiwan, Europe, Korea, Japan, and other exports to China, sure.
-
用 InP substrate 说明供应链控制应该瞄准上游
这次 $AXTI / InP substrate 出口管制正好说明： 特朗普的 America First 政策应该聚焦在保护欧洲、日本、韩国这些地方的供应链。 并且把它们武器化来对付中国。 单靠美国本身当然有筹码，但把日本、欧盟、韩国一起用起来，筹码会大得多。 如果让日本切断中国的人形机器人项目 / 上游化学品供应，再让欧盟切断中国的 photonics 项目： 他们会为了 substrates 和稀土让步。 你真正需要做的是掐住他们前沿产业发展的良率，或者直接卡制造这些组件的能力。 而不是只拿 40% 关税吓唬他们别再出口廉价商品。
英文原文
This $AXTI / InP substrate export controls is EXACTLY why: Trump's America 1st policies should focus around securing its supply chains over in EU/Japan/KR. And weaponizing it against China. US alone has leverage, but everyone together has extreme leverage. If you use Japan to cut off China's humanoid program / upstream chemicals and leverage EU to cut off China's photonics program: They'll bend for substrates and rare earths. You literally need to break their frontier industry development over yields or actually manufacturing the components. Not just threaten 40% tariffs to stop them from shipping cheap goods.
-
把 AXTI / InP substrates 牵扯进中美谈判
天啊，聊天室里不会吧……真不敢信 $AXTI / InP substrates 真的成了中美和特朗普、习近平贸易谈判的一部分？ 我这个 thesis 完全押对了。 但说真的，日本应该开始切断像 Ulvac 这样的上游设备供应，优先针对中国的人形机器人产业；红磷供应也该像 Rasa / NCI 那样限制。 而美国应该施压欧盟，把那些被中国公司像 Robotechnik 买走的 photonic 垄断资产，比如 ficonTEC，收回来。
英文原文
LOL chat no way... Can't believe $AXTI / InP substrates really a major part of the China-US Donald Trump's trade talks with Xi? I got this thesis completely right. But really, Japan should start to cut off upstream machine supply like Ulvac for humanoids. And red phosphorus like Rasa/NCI to China, if they're not shipping. And US should pressure EU to take back their their photonic monopolies like ficonTEC that Chinese companies like Robotechnik managed to buy.
-
说 SIVE 正在变成下一家美国 photonics 龙头
$SIVE 现在正朝着下一个 $LITE、也就是美国 photonics 巨头的方向前进。 他们正在把董事会重新调整为以美国高管和美国 photonics 为中心。 所以现在核心董事会成员变成了：来自美国的 $GFS 高管和 $CITI 高管，再加上由 UC Berkeley 毕业生和 $LITE 前高管管理公司。 原来离开的 3 位成员来自瑞典 / 欧盟本地。 这只是战略从“发展本地瑞典半导体生态”转向“主导美国 / 全球 photonics 市场”。 我不是想贬低他们的履历。 只是按我的看法，如果目标是聚焦美国 / 全球 photonics 市场，更多美国高管会更优。 但无论如何，他们都值得为把 $SIVE 带到今天的样子而自豪。
英文原文
$SIVE is now aiming to become the next $LITE, a US photonics giant. They're re-centering their board around US executives + US photonics. So the core board are now: US $GFS Executives and $CITI Executives, with the company run by UC Berkeley grads and $LITE executives. The 3 members leaving were local Swedish/EU. This is just a shift in strategy from focusing on developing local Swedish Semi environments: To dominating the US/global photonics market. I'm not trying to discredit their service/background. But in my view to focus around US/global photonics markets, it's likely optimal to have more US executives. But they should all be proud for helping make $SIVE what it is today.
-
批评台湾金融监管的荒唐规则
台湾金融行业的规则真是离谱。 他们竟然把联发科这样的大公司都卖掉了，笑死。 这就像美国因为 Google 涨太多，就不让人再买 $GOOGL 一样。 然后还导致 Google 股价下跌。 他们难道不希望自己公司的估值上涨、让本国经济受益吗？ 尤其在半导体大牛市的这个阶段，明明每一周都很关键，这种规则其实是在拖自己高科技经济的后腿。 真有意思，自己的规则反而在伤害自己的高科技经济。
英文原文
Taiwan financial sector rules are hilariously stupid. They literally dispositioned Mediatek, one of their largest companies lmfao. It's like if the US stopped letting people buy $GOOGL because the stock went up. Then that caused Google's stock dropped after. Wouldn't they want their companies to appreciate in value, so their economy benefits? Especially during one of their largest bull-runs from the semi sector, it's actually hindering their own growth where every week matters. Interesting to see such backward rules hurting their own high-tech economy.
-
把 SIVE 和 Ayar / AlChip / Amazon Trainium 链条连起来
@DanilSer33 新的 $SIVE / Ayar / AlChip / $AMZN Trainium 供应链，比最近那个 $POET 5000 万到 5 亿美元采购订单更值得看多。 市场大概率不会立刻看出 Ayar、AlChip 和 $SIVE 之间的联系，除非有人把它点出来。
英文原文
@DanilSer33 The new $SIVE / Ayar / AlChip / $AMZN trainium supply chain is more bullish than the recent $POET $50m -> $500m purchase order. Markets probably won’t see the connection between Ayar AlChip and $SIVE unless someone points it out though.
-
补充 Amazon / AlChip / Ayar 对 SIVE 的利好链条
再补一句关于很正面的 $SIVE 消息，以及 MSCI 未来两周可能带来数千万资金流入： $AMZN 刚和 AlChip 做了新的私募配售。 这大概率意味着未来 Trainium 还有设计赢单。 如果你还记得的话……AlChip 是 Ayar 的主客户。而 $SIVE 是 Ayar 的主要激光供应商。 所以对 $SIVE 的含义会非常大，它是在搭 Amazon 生态增长的顺风车。
英文原文
Just to add to the very positive $SIVE news amid MSCI likely tens of millions of inflow in 2 weeks: $AMZN has a new private placement with AlChip. Probably implying design wins with future Trainium. If you don’t remember… AlChip was Ayar’s lead customer. And $SIVE is the primary laser supplier to Ayar. So implications for $SIVE, is enormous piggybacking off of Amazon’s ecosystem growth.
-
质疑 POET 估值高过 FOCI 的合理性
为什么 $POET（31.4 亿美元）的估值会比 FOCI（3363，31 亿美元）还高？？？ FOCI 明明就是 CPO 量产爬坡的瓶颈，也是 $TSM 和 $NVDA 的主要供应商。 我非常看好 FOCI，等机构发现这个名字后，它会跑赢。 另外，FOCI 管理层能不能像 $HIMX 那样去做 NASDAQ ADR？谢谢。
英文原文
HOW DOES $POET ($3.14B) HAVE A HIGHER VALUATION THAN FOCI (3363, $3.1B)??? FOCI IS LITERALLY THE BOTTLENECK FOR CPO VOLUME RAMP AND MAIN SUPPLIER FOR $TSM AND $NVDA. High conviction Foci outperforms once institutions find this name. Also, can Foci management please pursue NASDAQ ADR like $HIMX? Thank you.
-
解释 POET 订单激励其实也利好 SIVE
这些 warrants 确实有助于激励 $POET 再增加 5 亿美元的采购订单。 不过我不在意，实际上它对 $SIVE 更有利，而不是对 Poet 更有利，因为 Sivers 会拿到那 5 亿美元 $POET BOM 激光订单里很大一部分，而且不需要承担同样的激励抵押。 只是市场现在还没看懂这个细节。
英文原文
The warrants do help incentivize the $500m in extra purchase orders for $POET though. But I'm not complaining, it's actually more beneficial to $SIVE than it is to Poet, since Sivers gets all the large percent of that $500M $POET BOM of laser orders without that incentive as collateral. Just don't think markets know that nuance yet.
-
评论CPO供应链分析文章，指出Sivers与多家光互联芯片公司存在映射关系。
很有意思的阅读！觉得你覆盖了关于整体CPO增长曲线的大部分重要观点，写了一个很好的牛市案例。 特别是关于$SIVE如何服务于Ayar、$JBL和$POET，并通过与Win合作采用资本支出轻量的晶圆厂轻资产模式实现规模化。 实际上还有大量与$MRVL Celestial、Lightelligence、Lightmatter以及其他参与者的供应链映射关系，以及$AAPL等用于相邻SiPH垂直领域的玩家。 但不包括更多尚未100%确认的参与者是合理的。不过O-Net/enablence ELS面向亚洲超大规模业者这一点已得到确认（文章中没有看到这一点）。 但这也说明了Sivers映射到的玩家数量众多。
英文原文
Fun read! Think you covered majority of the important points regarding the overall CPO growth curve and wrote a great bull case. Especially how $SIVE feeds into Ayar, $JBL, and $POET + scales using capex light fab-lite model with Win. There's actually a ton of supply chain mapping with $MRVL Celestial, Lightelligence, Lightmatter, and others of players like $AAPL for adjacent SiPH verticals. But makes sense not to include more not 100% confirmed players. O-Net/enablence ELS for Asian hyperscalers was confirmed though (didnt see that there) But just goes to show many players Sivers maps to.
-
把 POET 的大单和 SIVE 的激光瓶颈联系起来
如果你还不知道，$POET 因为 Lumilens 的一笔 5000 万美元采购，盘前涨了 22%。 后续还会按更长期框架扩到 5 亿美元。 正如 Sivers 2025 年报里提到的：$POET 会在 2026 年下半年和 $SIVE 一起放量。 这些中游玩家往上反向拉动 $SIVE 的需求，而激光就是一切的 chokepoint。 所以，是的，这对 Sivers 极度利好。
英文原文
If you didn't know $POET is up 22% premarket on a $50m purchase from Lumilens. Scaling to $500m with a longer term framework. As mentioned in Sivers 2025 annual report: $POET is ramping up H2 2026 with $SIVE. Any demand from these midstream players back to $SIVE as the laser chokepoint for everything. So yes, this is highly bullish for Sivers.
-
认为 SIVE 的光学侧信息全都很牛
我的意思是，$SIVE 光学侧给出的信息几乎全是我想听到的？ -> H2 开始放量 -> TAM 扩展到 pluggable transceivers -> 除了 $JBL 之外，还有多家新的潜在 hyperscaler 供应商在测试/认证他们的激光器，用于 pluggable。 -> Win Semi 产能已经锁定。
英文原文
I mean $SIVE delivered everything I wanted to hear on their optical side? -> Volume ramp starting H2 -> TAM expansion into pluggable transceivers -> Multiple new likely hyperscaler suppliers testing/qualifying their lasers for pluggable aside from $JBL. -> Win Semi capacity secured.
-
$SIVE 年报显示 CPO 与 pluggable 双线扩张
$SIVE 2025 年报分析。 TLDR：非常看多。 Sivers 的主增长向量是 CPO，但他们已经把 TAM 扩展到 pluggable transceivers，以及多项新的 qualification / development。 1. “We are currently seeing great interest... testing our DFB lasers across multiple manufacturers in pluggable transceivers” 从 pluggable 角度看，我们已经在 $JBL 的 1.6T LRO 上看到类似情况，而年报暗示他们还在和更多 hyperscaler 供应商开发 / 认证。 “Our serviceable markets have now been expanded to include pluggable optical interconnects as well as scale-up and scale-out architectures for co-packaged” （TAM 扩张） 2. “Discussions with hyperscalers and pluggable transceiver suppliers indicate a shortage of CW lasers in the coming years” $LITE 已经提示过 CW laser 的瓶颈，他们甚至不得不外购竞品。我们之前其实就猜到 CW laser 会是瓶颈。 而这次年报确认了这一点，所以我也在看 Win Semi。 “The partnership announced with high-volume supplier Win Semiconductor in March 2025 now gives us a strong position to meet growing demand” $SIVE 很可能已经通过 Win 锁定了产能，这正是我最想确认的。 这把 Sivers 放在了 CW laser 既是瓶颈又是 CPO 激光架构领导者的位置。 H2 量产信号（看多）： 3. “The collaboration positions both companies to address the rapidly growing market for optical AI connectivity, with prototypes to be demonstrated to customers during the first half of 2026 and with the goal of scaling up production by the end of 2026” H1 更偏 preproduction，H2 才是量产启动，像 $POET 那样。 4. “We are pleased that our largest LIDAR customer will increase production starting in the fourth quarter of 2026” $AEVA 在 2026 年 Q4 开始量产，对 $SIVE 也是利好。 LIDAR 提供了 revenue floor，而他们的 CPO 正在放量。 5. Sivers 还宣布和 LIGHTIUM AG 合作，把 CW laser 直接整合到 TFLN wafers 上。3.2T+ 周期，属于未来防守。 顺带说一下，去年 2025 的财报本身，除了瑞典本地媒体/散户外，真没几个像样的投资者会在意。 尤其当你看的是 2027-2028 的 CPO 超级周期时。 但从财报和地理布局里你能提取到的一个高置信客户信号是：$NOK 现在已经是 $SIVE 的高置信客户。 TLDR： -> Win Semi 暗示在 CW laser 瓶颈下锁定了产能 -> 新一批 hyperscaler 供应商在测试/认证 pluggable transceivers，TAM 巨大扩张 -> 新的 CW laser 客户 -> photonics 和 lidar 都在 H2 开始量产爬坡。
英文原文
$SIVE 2025 annual report analysis. TLDR: Extremely Bullish. Sivers main growth vector is CPO, but they've TAM expansioned to pluggable transcivers + multiple new qualifications/development. 1. "We are currently seeing great interest... testing our DFB lasers across multiple manufacturers in pluggable transceivers" For pluggable angle, we've seen this with $JBL 1.6T LRO already, but annual report hinted they're developing/qualifying with more hyperscaler suppliers. "Our serviceable markets have now been expanded to include pluggable optical interconnects as well as scale-up and scale-out architectures for co-packaged" (TAM expansion) 2. "Discussions with hyperscalers and pluggable transceiver suppliers indicate a shortage of CW lasers in the coming years" $LITE already signaled CW laser bottlenecks, and they had to buy externally from competitors. So we kinda guessed CW Laser was a bottleneck. And this confirmed it, so was wondering about Win semi. "The partnership announced with high-volume supplier Win Semiconductor in March 2025 now gives us a strong position to meet growing demand" $SIVE likely has capacity locked in with Win from this nuance, which is exactly what I wanted to know. This positions Sivers in the CW laser as both a bottleneck and CPO laser architectural leader. VOLUME PRODUCTION H2 INDICATIONS (BULLISH): 3. "The collaboration positions both companies to address the rapidly growing market for optical AI connectivity, with prototypes to be demonstrated to customers during the first half of 2026 and with the goal of scaling up production by the end of 2026" H1 is more preproduction, H2 production signaled starting with names like $POET. 4. "We are pleased that our largest LIDAR customer will increase production starting in the fourth quarter of 2026" $AEVA start of volume production Q4 with $SIVE = bullish for both. Revenue floor from LIDAR as their CPO scales. 5. Sivers announced a partnership with LIGHTIUM AG to integrate their CW lasers directly onto TFLN wafers. 3.2T+ cycle. (future proofing) FYI no decent investor cares about last year's 2025 financials from development contracts aside from Swedish Media/Locals. Especially when you're forward looking for the 2027-2028 CPO supercycle. But the hint from you can take away from financials + geography that is $NOK is now the high confidence customer of $SIVE. TLDR: -> Win Semi implied capacity lock in during CW laser bottleneck -> Hints of new group of hyperscaler suppliers testing/qualification for pluggable transcivers, which is massive TAM expansion. -> New customers for CW lasers -> Volume production scaling starting H2 for both photonics and lidar.
-
爆料 Shunsin 拿下 AVGO 的 CPO/SiPH 订单
我错过了一篇很有意思的报道：Shunsin（6451）拿到了 $AVGO 的 CPO / SiPH 订单…… 而且他们的 EPS 增长率可能会超过 1240%。 当然这还是保密内容，媒体也只是猜测。 但这也说得通：你们那些顶级玩家都用 Foxconn 做光学封装 / 测试。 https://t.co/aqnUQEQdYS
英文原文
Interesting report that I missed that Shunsin (6451) landed $AVGO CPO/SiPH orders… And their EPS growth rate could reach over 1240%. Of course this is confidential and is media speculation. But makes sense that all your top players use Foxconn for optical packaging/test. https://t.co/aqnUQEQdYS
-
发现 Nextronics 也是机器人供应链里被忽视的供应商
现在再想想…… Nextronics（8147）其实也是 Amazon 机器人供应链里一个很被低估的供应商。 所以 Amazon 越扩张，他们的收入也会越大（市值才约 2 亿美元，如果毛利率真能做到 38-40%，那影响应该会很明显）。 我之前只盯着 $NVDA 的 CPO 供应链，因为那是它们最大的增长向量。 但我也认为 Amazon 的机器人计划是最适合大规模生产的路线，因为每造一个内部机器人，就会降低 opex、削减 headcount、提升利润率。 而且整个生态都会受益。
英文原文
Now that I think about it more… Nextronics (8147) is a pretty undiscovered supplier to robotics supply chains like $AMZN too. So as Amazon scales up, so does their revenue (it’s only ~$200m mc so it should be material if they’re hitting 38-40% gross margins). I was only focusing on $NVDA CPO supply chains as their largest growth vector earlier. But I do think Amazon’s robotics program have the simplest route for mass production since each one they make internally lowers opex, cuts headcount, and improves profitability. And their whole ecosystem should benefit.
-
主张美国应扩产本土稀土，而不是只盯少数项目
@beroccan 我之前写过一篇文章，讲为什么美国需要扩张更多本土稀土项目，而不是只盯着像 $MP 这样的少数几个。 但对，稀土确实对压低成本很关键。 https://t.co/fLN7SdBegt
英文原文
@beroccan I wrote on article on why the US needs to scale more domestic rare earths projects rather than focusing on a few like $MP. But yes they’re critical to bring costs down https://t.co/fLN7SdBegt
-
从 CAD CEO 的说法看好机器人，但强调去中国供应链化
“机器人可能会成为史上最大的产品类别”——$CDNS CEO。 “预测是 25 万亿美元。全球 GDP 才 110 万亿美元。所以如果这真发生了，那就太大了。” 我方向上对机器人 / 人形机器人非常看多。 但也许现在是时候让 $TSLA 和美国真正开始优先考虑，如何在中国供应链之外去建设它了？
英文原文
“Robotics may be the biggest product category of all time” - $CDNS CEO. “The projection is $25 trillion. The whole GDP of the world is $110 trillion. So this is huge if this happens.” Extremely bullish on robotics/humanoids directionally. But maybe it’s time for $TSLA and America to really start prioritizing how we build it outside Chinese supply chains?
-
用 TSEM 的预付款解释上游 substrate 需求
顺便提醒一下，$TSEM 报了额外的预付款（2.9 亿美元）来锁产能。 下游晶圆厂产能紧张，会直接转化为上游原始 substrate 的巨大需求，尤其是 $SOI（也意味着更多收入）。 我的光学持仓彼此之间关联非常强。 https://t.co/4vIAvS1jOl
英文原文
Just in case you were wondering $TSEM reported extra prepayment ($290M) to secure capacity. Downstream foundry capacity squeeze directly translates to an immense demand for raw substrates upstream with $SOI (more revenue). My optical picks are very interconnected. https://t.co/4vIAvS1jOl
-
欢迎瑞典对冲基金面对全球机构资金流入
@JinYu762 瑞典本地对冲基金，准备好迎接全球金融机构对 $SIVE 的资金流入吧。
英文原文
@JinYu762 Local Swedish hedge funds, get ready to meet global financial institution inflow with $SIVE.
-
计算 MSCI 被动资金流入对 SIVE 的拉动
@Platypus_55 我算了一下，大概会有 2500 万到 3200 万美元的新资金流入买盘压力？ 这对 $SIVE 来说是结构性的巨大顺风。
英文原文
@Platypus_55 I did some calculations and it might be around $25-$32m worth of new inflow buying pressure? Which is structurally a large tailwind for $SIVE.
-
解释 SIVE 的 CW laser 为什么适配多家 hyperscaler 供应链
他们特别之处在于拥有非常先进、架构上独特的 CW lasers，所以能适配 Ayar、$MRVL Celestial、Lightmatter、Lightelligence 以及 hyperscaler 供应链里的这些规格。 像 Furukawa、Sumitomo 这种公司，做的更多是大规模生产通用 / 标准化 CW lasers。 这也是为什么很多中游光学公司最后都会用 $SIVE 这样小但关键的 laser 公司，而随着它们一起成长，Sivers 也会跟着放大。
英文原文
What they do special is really advanced/architecturally unique CW lasers, so they fit Ayar, $MRVL Celestial, Lightmatter, Lightelligence, and these types of specifications in hyperscaler supply chains. Companies like Furukawa, Sumitomo just mass produce generic/standardized CW lasers. That’s a reason why so many of these midstream optical companies ended up using such small laser company in $SIVE and as they all grow, Sivers scales with them.
-
用历史市值路径说明 laser chokepoint 的威力
回答这个问题，我就拿一些 laser chokepoint 的历史数据出来： $LITE 从 30 亿 -> 150 亿 -> 800 亿美元，只用了从 2024 年开始的 2 年。 $AAOI 从 7.7 亿 -> 150 亿美元，只用了 2025 年开始的 1 年。 $SIVE 现在在 2026 年是 16 亿美元。 世界上这类公司没多少，而 laser 又是 photonics 的绝对中心。 每家公司都拥有一个特定的 chokepoint，对应一次光学架构的变化。 $LITE 对应 EML。 $AAOI 对应 cw 800g/1.6T pluggable。 而现在 $SIVE 对应 cw CPO。
英文原文
To answer that question. I’ll just float some historical data out there with laser chokepoints: $LITE went from $3B -> $15B -> $80B in 2 years from 2024. $AAOI went from $770M -> $15B in 1 year from 2025. $SIVE is $1.6B today in 2026. There’s not many in the world and lasers are the absolute center of photonics. Each owned a specific chokepoint from optical architectural shifts. $LITE for EML. $AAOI for cw 800g/1.6T pluggable and now $SIVE for cw CPO.
-
CPO市场将从零增长至910亿美元，欧洲、台湾、日本新玩家将主导CPO供应链变革。
人们好奇我为什么最近在关注非美国市场。 为什么？因为共封装光学(CPO)是我排名第一的主题性看多方向。 市场尚未意识到，光子学领域正经历一场突如其来的范式转变... 2025年，我是少数几个提前预判当前超级周期的人之一，当时买入的$AAOI约在$30、$LITE约在$300多、$AXTI约在$13...这些都是基于真实的业绩验证和其他人无法展示的thesis。 CPO市场将从约0增长到910亿美元的TAM机会。 高盛研究预计在未来1.5年内实现。 与此同时，整体光学市场将达到1540亿美元。 许多之前对当前光子学周期几乎没有任何敞口的企业： -> 欧洲企业拥有高端激光设计，如$SIVE或$SOI涉及衬底 -> 台湾企业如Foci (3363)、Nextronics (8147)、Shunsin (6451)等提供光学元件和代工服务 -> 日本企业涉及激光量产、衬底和化学品 突然之间成为了CPO的新主导力量。 至于美国玩家，目前敞口不大。但现有企业如$LITE、$COHR仍能从CPO中获得上涨空间，因为这是它们新的增长向量。 我对现有玩家的逆向思考： 他们的大部分估值已经反映了巨大的传统可插拔光学收入，而这些收入将不可避免地随时间被 cannibalize（ cannibalization蚕食），因此除非有人能利用杠杆，否则重新估值的空间有限。 许多新的纯CPO概念股会在H2 2026开始量产时从0到100极速增长（作为收入桥梁），然后进入H2 2027的规模化增长阶段（成为大规模增长驱动力）。 市场通常会在事件发生前8-12个月定价... 尽管在此期间市场可能出现波动，但我对自己的供应链研究在主题性方向上有高度信心。
英文原文
People wonder why I'm focusing on non-US markets recently. Why? CPO is my #1 thematic long. Markets don't know yet, the sudden paradigm shift in photonics... I was one of the only to frontrun the current supercycle in 2025 w/ $AAOI @ ~$30, $LITE ~$300s, and $AXTI at ~$13 on X.... With the actual receipts and thesis that others can't show. CPO goes from ~$0. To $91 Billion TAM opportunity. In the next 1 1/2 years from GS research. While overall optical market reaches $154B. Many players that had little exposure to the current photonics cycle at all: -> In Europe with high-end lasers design like $SIVE or $SOI with substrates. -> In Taiwan with Foci (3363), Nextronics (8147), Shunsin (6451) and others for optical components and foundries. -> In Japan with laser mass production, substrates, and chemicals. Are suddenly the new dominant players for CPO. As for US players, there's not much exposure. But the existing ones like $LITE, $COHR still get upside from CPO as that's their new growth vector. My contrarian thought process on current players: Is that most of their valuation is priced in huge legacy pluggable revenue that will inevitably face cannibalization over time, so re-rating potential is less unless someone uses leverage. A lot of these new purer play CPO names go from 0 to 100 extremely quickly one mass production starts H2 2026 for scale out (as a revenue bridge) into H2 2027 for scale up (massive growth driver). Markets usually price things in 8-12 months ahead of time too... I have high conviction thematically in my supply chain research despite any market volatility leading up until then.
-
用历史说明 SIVE 已站在新的 laser chokepoint 上
$LITE 用 2 年从 20 亿涨到 800 亿，因为它掌握了 EML 的 laser chokepoint。 $AAOI 用 1 年从 14 亿涨到 140 亿，因为它掌握了 800g / 1.6T transceivers 和组装的 laser chokepoint。 $SIVE 今天大概是 14 亿美元，在 CPO 之前就已经掌握了 CW laser chokepoint。 Sivers 是我最喜欢的一个。
英文原文
$LITE went from $2B to $80B in two years by owning the laser chokepoint for EML $AAOI went from $1.4B to $14B in one year by owning laser chokepoint for 800g/1.6T transceivers + assembly $SIVE is $1.4B today before CPO and owns the CW laser chokepoint. Sivers is my favorite out of everything.
-
对 CPO 里两个不同 bottleneck 都很看好
CPO 供应链里有两个不同的 chokepoints / bottlenecks，而且风险结构不一样。我真的、真的都很喜欢。 MSScorps 有潜在的 yield monopoly，市场份额最高可到 90%（基本就是 100%，只是我不想说成反垄断）。这对规模化来说绝对必需。 FOCI 则是 FAU 和光学组件的 bottleneck。 FOCI 可能是一个增长很快、很有吸引力的敞口，而且需求可见度很高。 如果它的竞争对手真被专利打掉，Msscorp 就像一艘从 0 到 100 的火箭。
英文原文
Two different chokepoints/bottlenecks in CPO supply chain with different risk profiles. I really, really like both. MSScorps is potential monopoly over yields with up to 90% market share (basically 100% just don’t want antitrust), which is an absolute must for scaling. Foci is a bottleneck over FAU and optical components. Foci is probably a really high growth, compelling exposure with a lot of demand visibility. Msscorp is like a 0 to 100 rocket ship if all their competitors get successfully slapped by patents.
-
说明 FOCI 在 CPO 暴露里非常有吸引力
FOCI（3363）在现在这个位置上，作为 CPO 敞口看起来极具吸引力，市值大概只有 33.5 亿美元。 1. $TSM COUPE 先进封装负责人暗示，FOCI 提供的 FAU 供给会是量产中的大瓶颈。 2. Morgan Stanley note 预计他们会成为 $NVDA 和 $TSM 的领先供应商，市占率最高可到 50%。 3. $HIMX 释放了创纪录需求的信号，而且表示 Foci 应该扩产，这意味着中期需求可见度很高。 4. 根据 Goldman Sachs 的研究，FAU 和光学组件占 CPO 相关 BOM 的很大比例。 5. 总体 CPO TAM 按 GS 的说法，未来两年会从接近 0 增长到 910 亿美元。 方向上我对这个主题非常有信心（NFA），尽管短期市场波动很大。 风险在于后续 generation 被设计替代。 但在未来 1-2 年里，我认为它有很高的重估潜力，相比其他名字也许需要持续跟踪。 只是顺手分享一下我持有的更纯 CPO 敞口思路。
英文原文
FOCI (3363) is looks extremely compelling around now at ~$3.35B MC for CPO exposure. 1. $TSM COUPE advanced packaging director hinted that FAU supplies by FOCI be a pretty big bottleneck for mass production. 2. Leading supplier for $NVDA and $TSM expected with up to 50% market share from Morgan Stanley note. 3. $HIMX signaled record demand and that Foci should scale up capacity (meaning high medium term demand visibility) 4. FAU and optical components make a large % of CPO related BOM from Goldman Sachs research note. 5. Overarching CPO tam basically goes from near 0 to $91B in the next two years from the GS note. I’m very confident about this theme directionally over time (NFA), despite any recent market volatility. Risks include getting designed out for later generations. But over the next 1-2 years, I think it has high potential to be re-rated compared to other names but might need to be actively monitored. Just throwing out ideas over long positions I hold, for more purer play CPO exposure.
-
众包人形机器人投资标的清单，汇总出现频率最高的六只股票。
这是人形机器人敞口众包列表： - $OUST - Rainbow Robotics (277810) - $AMBA - Ubtech Robotics - $MKA - Nextronics - $SYM - Harmonic Drive (6324) - $VPG - 北京极智嘉 - $MBLY - $ARBE - Nabtesco (6268) - $SERV - $HSYDF - Robotstrategy - $ZBRA - $CATL - $ABB - $BOT - Unitree (尚未上市) - $LSCC - 亿sunny Robot (300024) - $NOVT - $RR - $PDY - 禾赛科技 (2525) - $SHA.DE - $XBOT - $XPEV - $BAM - $ALNT - 6268.T - $AMBQ - $ATOM - $MRAM - $ISRG - $HLIT - Robosense (2498) - $HG - $ACUVI - $CGNX - $KLIC - $BSL - $AEVA - $AUR - $CTH.V - $IMSR - $NEO - $KDK - $MRLN - $KITT - $INDI - $NOVT 凭记忆： Harmonic Drive、$OUST、$BOT、$VPG、$MBLY和Ubtech出现频率最高。 接下来会对这些提及最多的标的开始做尽职调查。
英文原文
Here's the humanoid exposure crowdsourced list: - $OUST - Rainbow Robotics (277810) - $AMBA - Ubtech Robotics - $MKA - Nextronics - $SYM - Harmonic Drive (6324) - $VPG - Beijing Geekplus - $MBLY - $ARBE - Nabtesco (6268) - $SERV - $HSYDF - Robotstrategy - $ZBRA - $CATL - $ABB - $BOT - Unitree (not public yet) - $LSCC - Esunny Robot (300024) - $NOVT - $RR - $PDY - Hesai (2525) - $SHA.DE - $XBOT - $XPEV - $BAM - $ALNT - 6268.T - $AMBQ - $ATOM - $MRAM - $ISRG - $HLIT - Robosense (2498) - $HG - $ACUVI - $CGNX - $KLIC - $BSL - $AEVA - $AUR - $CTH.V - $IMSR - $NEO - $KDK - $MRLN - $KITT - $INDI - $NOVT Off the top of my head: Harmonic Drive, $OUST, $BOT, $VPG, $MBLY, and Ubtech showed up the most. Will start doing DD into mentions.