白发股神：2026年6月推文归档 8

📝 推文原文：

$NVDA Networking Senior Vice President refuting recent analyst reports on delays: “the most exciting stuff is co-packaged optics.” There is no delay in H2 CPO product delivery schedule. CPO switch will enter mass production and begin ramping up customer deliveries as planned in the second half of 2026. This was a media article, original interview source credit should have been credited to Tae Kim / Computex. Something fun to note too was this quote “Gilad was VERY enthusiastic about the CPO ramp from Nvidia.” Both near term and long term. Yeah… I’m extremely bullish on CPO alongside Nvidia.

$NVDA 网络业务高级副总裁反驳近期关于延迟的分析师报告：“最令人兴奋的是 co-packaged optics（CPO，共封装光学）。”H2 CPO 产品交付节奏没有延迟，CPO 交换机将按计划在 2026 年下半年进入量产并开始向客户爬坡交付。原始采访来源应归功于 Tae Kim / Computex。另一个有趣细节是：“Gilad 对 Nvidia 的 CPO 爬坡非常热情。”无论短期还是长期。我对 CPO 以及 Nvidia 相关链条都极度看多。

📝 推文原文：

CPO scale out earlier than expected: > Foxconn: est. units register upward and optical switches shipped early to $NVDA. CPO scale up timelines from $LITE Mizuho Technology Conference today: “The company expects to start shipping Scale-Up optical products in the second half of 2027, with formal volume ramp-up in 2028”. SVP $NVDA networking: “We’re going to ramp up CPO second half of this year”. No delay indications. I’m gonna go ahead and trust industry projections. Where they all reiterate faster timelines for scale out CPO H2 onward. And scale up CPO H2 2027 onward (with main growth happening 2028) Over a questionable motive analyst firm that said $MU had no share of HBM4 Rubin (causing a selloff) Where micron went out shortly later to into enter mass production. (Triple digit return shortly after) I think people going long on temporary bridge architectures from this incorrect report won’t be too happy. Appreciate the buying opportunity though. CPO scale-out 比预期更早：富士康预计单量登记上修，并已提前向 $NVDA 交付光交换机。$LITE 在今天的瑞穗科技会议上给出的 CPO scale-up 时间线是：“公司预计 2027 年下半年开始出货 Scale-Up 光学产品，2028 年正式放量爬坡。”$NVDA 网络业务高级副总裁也表示：“我们会在今年下半年开始 CPO 爬坡。”没有延迟迹象。我选择相信产业界预测：它们都重申 scale-out CPO 从 H2 起时间线更快，scale-up CPO 从 2027H2 起推进（主要增长在 2028）。相比之下，某家动机可疑的分析机构曾称 $MU 没有 HBM4 Rubin 份额并引发抛售，但 Micron 随后很快宣布进入量产，之后带来三位数回报。我认为基于这份错误报告去做多临时桥接架构的人不会太开心。不过感谢这个买入机会。

📝 推文原文：

CPO scale out earlier than expected: > Foxconn: est. units register upward and optical switches shipped early to $NVDA. CPO scale up timelines from $LITE Mizuho Technology Conference today: “The company expects to start shipping Scale-Up optical products in the second half of 2027, with formal volume ramp-up in 2028”. SVP $NVDA networking: “We’re going to ramp up CPO second half of this year”. No delay indications. I’m gonna go ahead and trust industry projections. Where they all reiterate faster timelines for scale out CPO H2 onward. And scale up CPO H2 2027 onward (with main growth happening 2028) Over a questionable motive analyst firm that said $MU had no share of HBM4 Rubin (causing a selloff) Where micron went out shortly later to into enter mass production. (Triple digit return shortly after) I think people going long on temporary bridge architectures from this incorrect report won’t be too happy. Appreciate the buying opportunity though.

CPO scale-out 比预期更早：富士康预计单量登记上修，并已提前向 $NVDA 交付光交换机。$LITE 在今天的瑞穗科技会议上给出的 CPO scale-up 时间线是：“公司预计 2027 年下半年开始出货 Scale-Up 光学产品，2028 年正式放量爬坡。”$NVDA 网络业务高级副总裁也表示：“我们会在今年下半年开始 CPO 爬坡。”没有延迟迹象。我选择相信产业界预测：它们都重申 scale-out CPO 从 H2 起时间线更快，scale-up CPO 从 2027H2 起推进（主要增长在 2028）。相比之下，某家动机可疑的分析机构曾称 $MU 没有 HBM4 Rubin 份额并引发抛售，但 Micron 随后很快宣布进入量产，之后带来三位数回报。我认为基于这份错误报告去做多临时桥接架构的人不会太开心。不过感谢这个买入机会。

📝 推文原文：

I don’t quite think photonics from $AAOI to $LITE or $SIVE are disappearing anytime soon… Just extremely volatile. Anyway, curious what other people are buying today? 我不太认为从 $AAOI 到 $LITE 或 $SIVE 的光子学标的会很快消失……只是波动极端剧烈。顺便问问，大家今天还在买什么？

📝 推文原文：

I don’t quite think photonics from $AAOI to $LITE or $SIVE are disappearing anytime soon… Just extremely volatile. Anyway, curious what other people are buying today?

我不太认为从 $AAOI 到 $LITE 或 $SIVE 的光子学标的会很快消失……只是波动极端剧烈。顺便问问，大家今天还在买什么？

📝 推文原文：

$EWY 32% IV into 58% IV expansion trade. Into underlying SK Hynix / Samsung increase way ITM. Was such a goated call? (~383% return) Had way too many great ideas this year… Still super proud of predicting South Korea index volatility increase due to memory concentration. $EWY 从 32% 隐含波动率扩张到 58% 的交易。底层 SK Hynix / Samsung 上涨后期权大幅实值。这个判断是不是太强了？（约 383% 回报）今年好点子太多了……我仍然非常自豪于提前预测韩国指数会因为记忆体集中度而出现波动率上升。

📝 推文原文：

$EWY 32% IV into 58% IV expansion trade. Into underlying SK Hynix / Samsung increase way ITM. Was such a goated call? (~383% return) Had way too many great ideas this year… Still super proud of predicting South Korea index volatility increase due to memory concentration.

$EWY 从 32% 隐含波动率扩张到 58% 的交易。底层 SK Hynix / Samsung 上涨后期权大幅实值。这个判断是不是太强了？（约 383% 回报）今年好点子太多了……我仍然非常自豪于提前预测韩国指数会因为记忆体集中度而出现波动率上升。

📝 推文原文：

Names like: - $ASX - Sumitomo Electric - $JBL - $VICR - $GFS - $AAOI - AlChip - $TSEM - $FN - Furukawa Electric - $CLS - $NBIS - $NOK - $AMKR - $LITE - $COHR. Off the top of my head. So basically, AI exposure trading in the $10-100B range. Likely have compelling ROI right now compared to indexes or $ARM to $MRVL that ran quite a bit? (Just a disclosure, only have financial interest in NBIS/TSEM/AAOI above) I mention a lot of smaller ideas, but that’s just to chase outsized returns. Still feels like many of these have room to go. 比如：$ASX、住友电工、$JBL、$VICR、$GFS、$AAOI、AlChip、$TSEM、$FN、古河电工、$CLS、$NBIS、$NOK、$AMKR、$LITE、$COHR。临时想到这些。基本上就是市值在 $100亿-$1000亿区间的 AI 暴露标的。相比指数，或已经上涨很多的 $ARM 到 $MRVL，这些现在可能有更有吸引力的 ROI？（披露：上述名单中我只对 NBIS/TSEM/AAOI 有财务利益）我提很多小一点的想法，是为了追求超额回报。感觉其中许多仍有上涨空间。

📝 推文原文：

Names like: - $ASX - Sumitomo Electric - $JBL - $VICR - $GFS - $AAOI - AlChip - $TSEM - $FN - Furukawa Electric - $CLS - $NBIS - $NOK - $AMKR - $LITE - $COHR. Off the top of my head. So basically, AI exposure trading in the $10-100B range. Likely have compelling ROI right now compared to indexes or $ARM to $MRVL that ran quite a bit? (Just a disclosure, only have financial interest in NBIS/TSEM/AAOI above) I mention a lot of smaller ideas, but that’s just to chase outsized returns. Still feels like many of these have room to go.

比如：$ASX、住友电工、$JBL、$VICR、$GFS、$AAOI、AlChip、$TSEM、$FN、古河电工、$CLS、$NBIS、$NOK、$AMKR、$LITE、$COHR。临时想到这些。基本上就是市值在 $100亿-$1000亿区间的 AI 暴露标的。相比指数，或已经上涨很多的 $ARM 到 $MRVL，这些现在可能有更有吸引力的 ROI？（披露：上述名单中我只对 NBIS/TSEM/AAOI 有财务利益）我提很多小一点的想法，是为了追求超额回报。感觉其中许多仍有上涨空间。

📝 推文原文：

A massive catalyst arrived today with $SIVE: Sivers announced $8.2M volume orders starting for Space applications (ALL.SPACE). This is for Beamforming ICs powering Space LEO/multi-orbit satellite communication. The bigger implication is not the contract size: But that Sivers now powers a larger defense prime in $YSS following their ALL.SPACE acquisition (similar to $MRVL design-in with Celestial). Which typically leads to more follow-up orders + volume contracts for Sivers, rather than just this specific contract. Turns out Sivers is also a Space/Defense supply chain chokepoint (ahead of SpaceX IPO) on top of their photonics AI DC sector lasers… This win aside, I’m expecting more volume ramps to be coming soon as well from their photonics side (looking at you Jabil + other pluggable makers). 今天 $SIVE 来了一个重大催化剂：Sivers 宣布获得 $820万 ALL.SPACE 太空应用批量订单。这是用于 LEO/多轨道卫星通信的波束赋形 IC。更大的意义不在于合同规模，而在于 Sivers 通过 ALL.SPACE 收购进入了更大的防务主承包商体系（类似 $MRVL 与 Celestial 的设计合作）。这类合作通常会带来更多后续订单和批量合同。事实证明 Sivers 在 AI 数据中心光子学激光器之上，还是太空/国防供应链的瓶颈（赶在 SpaceX IPO 之前）。除此之外，我预计光子学方面（Jabil 和其他可插拔光模块厂商）也很快会有更多量产爬坡。

📝 推文原文：

A massive catalyst arrived today with $SIVE: Sivers announced $8.2M volume orders starting for Space applications (ALL.SPACE). This is for Beamforming ICs powering Space LEO/multi-orbit satellite communication. The bigger implication is not the contract size: But that Sivers now powers a larger defense prime in $YSS following their ALL.SPACE acquisition (similar to $MRVL design-in with Celestial). Which typically leads to more follow-up orders + volume contracts for Sivers, rather than just this specific contract. Turns out Sivers is also a Space/Defense supply chain chokepoint (ahead of SpaceX IPO) on top of their photonics AI DC sector lasers… This win aside, I’m expecting more volume ramps to be coming soon as well from their photonics side (looking at you Jabil + other pluggable makers).

今天 $SIVE 来了一个重大催化剂：Sivers 宣布获得 $820万 ALL.SPACE 太空应用批量订单。这是用于 LEO/多轨道卫星通信的波束赋形 IC。更大的意义不在于合同规模，而在于 Sivers 通过 ALL.SPACE 收购进入了更大的防务主承包商体系（类似 $MRVL 与 Celestial 的设计合作）。这类合作通常会带来更多后续订单和批量合同。事实证明 Sivers 在 AI 数据中心光子学激光器之上，还是太空/国防供应链的瓶颈（赶在 SpaceX IPO 之前）。除此之外，我预计光子学方面（Jabil 和其他可插拔光模块厂商）也很快会有更多量产爬坡。

📝 推文原文：

Just a random thought: $JBL seems highkey compelling long idea at $38B. Don’t really think markets have priced in their 1.6T LRO pluggable transceiver business yet. Especially if it’s “how much can you make” with $SIVE as the bottleneck H1 2027. They already have the massive supply chains setup… and took over $INTC pluggable lines. Seems more scalable than $AAOI capex ATMs for laser fabs, if you have $SIVE + tons of different fabs like Win Semi + others mass producing lasers and $JBL doing the rest. So you’re getting that Innolight style setup for free (with US premiums), with an already validated hyperscaler supply chain. Prob H1 2027 is when everyone starts realizing. Maybe 40% rerating seems plausible? (dont have any open positions, just a thought). 随便想了一下：$JBL 在 $380亿市值看起来是很有吸引力的长线标的。我觉得市场还没有给他们的 1.6T LRO 可插拔光模块业务定价。尤其是如果 $SIVE 是 2027H1 的瓶颈，“你能造多少”才是关键。他们已经有庞大的供应链体系……还接手了 $INTC 的可插拔产线。如果你有 $SIVE 激光器 + Win Semi 等多家代工厂量产激光器 + $JBL 做其余部分，这比 $AAOI 不断 ATM 融资建激光晶圆厂更具可扩展性。相当于免费获得一个旭创式的业务模式（还带美国溢价），供应链已经被超大规模客户验证。可能 2027H1 大家才会意识到。40% 重估似乎合理？（不持有任何仓位，纯思考）。

📝 推文原文：

Just a random thought: $JBL seems highkey compelling long idea at $38B. Don’t really think markets have priced in their 1.6T LRO pluggable transceiver business yet. Especially if it’s “how much can you make” with $SIVE as the bottleneck H1 2027. They already have the massive supply chains setup… and took over $INTC pluggable lines. Seems more scalable than $AAOI capex ATMs for laser fabs, if you have $SIVE + tons of different fabs like Win Semi + others mass producing lasers and $JBL doing the rest. So you’re getting that Innolight style setup for free (with US premiums), with an already validated hyperscaler supply chain. Prob H1 2027 is when everyone starts realizing. Maybe 40% rerating seems plausible? (dont have any open positions, just a thought).

随便想了一下：$JBL 在 $380亿市值看起来是很有吸引力的长线标的。我觉得市场还没有给他们的 1.6T LRO 可插拔光模块业务定价。尤其是如果 $SIVE 是 2027H1 的瓶颈，“你能造多少”才是关键。他们已经有庞大的供应链体系……还接手了 $INTC 的可插拔产线。如果你有 $SIVE 激光器 + Win Semi 等多家代工厂量产激光器 + $JBL 做其余部分，这比 $AAOI 不断 ATM 融资建激光晶圆厂更具可扩展性。相当于免费获得一个旭创式的业务模式（还带美国溢价），供应链已经被超大规模客户验证。可能 2027H1 大家才会意识到。40% 重估似乎合理？（不持有任何仓位，纯思考）。

📝 推文原文：

🇨🇳 中文翻译： 核心交易研判与决策指引。

📝 推文原文（摘要）：

Re-sharing my “goated” list of 30 US-available stocks with brief explanations. These are not short-term trades — most are 3-5 year conviction plays across AI supply chain, space, fintech, and energy.

再次分享我精心挑选的30只美股清单及简要说明。这些不是短线交易——大多数是跨越AI供应链、太空、金融科技和能源的3-5年信念投资。

📝 推文原文：

🇨🇳 中文翻译： 核心交易研判与决策指引。

📝 推文原文（长推文摘要）：

I think my personal style of investing is a bit different, just some reflection: It’s inherently discretionary, based on stuff markets don’t know yet. A lot of it is guessing on unstructured relationships then seeing if it’s right or not down the line. Not big on sum-of-the-parts DCF models. Just figuring out what people might want/need, then what happens when the demand hits.

我觉得我的投资风格有点不同，只是一些反思：本质上是主观判断，基于市场还不知道的信息。很大程度上是对非结构化关系的猜测，然后看后续是否被验证。不太喜欢DCF模型。就是搞清楚人们可能想要/需要什么，然后看需求到来时会发生什么。

On toxic financing: “If they’re legitimately a good company, then it might be a good idea to go long after all the existing holders get diluted to oblivion. But if you care about your equity appreciation, it’s a good idea to stay far away from toxic financing structures.”

关于有毒融资： “如果它们确实是一家好公司，那等现有股东被稀释到极致后再做多可能是个好主意。但如果你在意股权增值，最好远离有毒融资结构。”

Red flags listed: $IREN ($6B ATM, 无限稀释), $SLNH ($500M ATM on $250M MC), $BKKT (稀释用于高管薪酬), $SNAP (隐藏的股权激励)

📝 推文原文：

🇨🇳 中文翻译： 核心交易研判与决策指引。

📝 推文原文（核心摘要）：

Glass substrate is NOT a concept play — it’s a confirmed capacity supercycle for 2026-2028. Ramping faster than SEMI forecasts. H2 2026 mass production starts. High growth in H1 2027. Core players: SKC/Absolics (first to MP for AMD), Samsung EM (2027 H2 for Apple/Broadcom), TSMC CoPoS (2-3 years out). Equipment “shovel seller”: LPKF.

玻璃基板不是概念炒作——这是2026-2028年确定的产能超级周期。起量比SEMI预测更快。2026下半年开始量产，2027上半年高增长。核心玩家：SKC/Absolics（首个量产，供AMD）、三星电机（2027H2，供苹果/博通）、台积电CoPoS（还需2-3年）。设备卖铲人：LPKF。

📝 推文原文：

I keep telling retail + Swedish Hedge Funds how important $SIVE is to CPO, but people don’t listen. JP Morgan increased ownership from 0.4% to 5%+ in ONE month. They get it. 我一直告诉散户和瑞典对冲基金 $SIVE 对 CPO 有多重要，但没人听。摩根大通在一个月内从0.4%增持到5%+。他们懂了。

📝 推文原文：

I keep telling retail + Swedish Hedge Funds how important $SIVE is to CPO, but people don’t listen. JP Morgan increased ownership from 0.4% to 5%+ in ONE month. They get it.

我一直告诉散户和瑞典对冲基金 $SIVE 对 CPO 有多重要，但没人听。摩根大通在一个月内从0.4%增持到5%+。他们懂了。