觅信聚合

我的关注

2026-09-05 06:26

原文：
我觉得这种思路很好：借鉴我对 ESMT DDR2/DDR3 等老一代存储芯片的研究, 举一反三.

或者把我研究 $AAOI 时关于 CW 激光器供需失衡的思路, 应用到 A 股.

很开心看到大家学习我的思考方式，而不是照抄具体的股票, 并用这些思路去寻找自己看好的做多标的！

翻译：
我觉得这种思路很好：借鉴我对 ESMT DDR2/DDR3 等老一代存储芯片的研究，举一反三。

或者把我研究 $AAOI 时关于 CW 激光器供需失衡的思路，应用到 A 股。

很开心看到大家学习我的思考方式，而不是照抄具体的股票，并用这些思路去寻找自己看好的做多标的！

2026-09-04 20:39

原文：
There's actually a whole new universe I haven't discovered with the Indian stock market...

Apparently there's some company called Sterlite Technologies (~$4.5B MC) that's up 630%, so I'm always curious why:

- upstream glass preforms (Used in optical fiber), so part of the fiber bottleneck
- preform -> fiber -> cable -> connectivity integration
- Optical fiber and cables, connectors, etc.

So maybe reminds me of India's alternative to Shin-Etsu/Furukawa?

And they got some $1.11 billion hyperscaler award, another $288m hyperscaler agreement and EBITDA up +184%.

So then i look into related companies like:

-West Coast Paper (india)
- FINCABLES

to see if markets missed anything or if there's any nuances I missed on why Sterlite is so highly valued. And why hyperscalers gave that award.

They also cited Germanium and helium inflation are headwinds, so you also get to see what other companies are citing too.

Regardless I'm always curious about the implications of random stuff I know nothing about.

So just having a good time looking through these things and doing research now (obviously never heard of this before until today, so don't own anything above)

翻译：
其实我在印度股市里还有一个全新的领域未曾涉足…

据说有家叫 Sterlite Technologies（约 $4.5B 市值）的公司涨幅达 630%，所以我一直很好奇原因：

- 上游玻璃预制棒（用于光纤），是光纤瓶颈的一环
- 预制棒 → 光纤 → 电缆 → 连接集成
- 光纤、电缆、连接器等

于是让我想到它可能是印度版的 Shin‑Etsu / Furukawa？

他们还获得了约 $1.11 billion 的 hyperscaler 奖项，另有 $288 m 的 hyperscaler 合同，EBITDA 增长 +184%。

于是我去查了相关公司，例如：

- West Coast Paper（印度）
- FINCABLES

看看市场是否遗漏了什么，或者我是否错过了 Sterlite 估值如此高的细节，以及 hyperscaler 为何会授予该奖。

他们还提到锗和氦价上涨是逆风，这也让人看到其他公司提到的因素。

总之，我总是对自己不熟悉的随机事物的影响感到好奇。

现在就只是抱着兴趣浏览这些信息并进行研究（显然今天之前从未听说过，所以并未持有任何相关股票）。

2026-09-04 19:28

原文：
OpenAI's Astra is now available. https://t.co/foqBHhPGkH

翻译：
OpenAI 的 Astra 现已可用。 https://t.co/foqBHhPGkH

2026-09-04 18:55

原文：
RT @MichaelDell: Thanks

翻译：
RT @MichaelDell: 谢谢

2026-09-04 14:32

原文：
Let me get this straight chat… you have:

$NVDA 70% supply constrained revenue growth projections (would be 100%+)

$AVGO AI revenue growing 100%+ Y/Y for next 2 years from mid ~$50B -> $230B

OpenAI Astra blowing away AGI benchmarks and starting to train itself.

$MSFT / $GOOGL / $AMZN stating compute demand imbalances into next year, with Nvidia projecting $1.3T spend for 2027.

You’re seeing this flow through supply chains from Samsung 70% capacity LTAs into 2031 to Powertech advanced packaging capacity 100% booked into 2030.

Even your legacy players like Winbond are starting 2029-2030 allocations.

Markets always have temporary drawdowns and macro scares… but…

How can anyone not think AI stock go brrr?

翻译：
让我理清一下，你们有：

$NVDA 供应受限70%导致收入增长预测（本应超过100%）

$AVGO AI收入在未来两年同比增长超过100%，从约500亿美元中期增长到2300亿美元

OpenAI Astra 超越 AGI 基准并开始自我训练。

$MSFT / $GOOGL / $AMZN 表示计算需求失衡将持续到明年，Nvidia 预测2027年支出达 1.3 万亿美元。

你可以在供应链中看到这种趋势，从三星 70% 容量的长期协议到 2031 年，再到 Powertech 高级封装容量 100% 已预订至 2030 年。

连像 Winbond 这样的老牌厂商也在开始 2029‑2030 年的分配。

市场总会有短暂的回调和宏观恐慌……但……

谁还能不认为 AI 股会一路飙升？

2026-09-04 12:30

原文：
From my channel checks, it appears Korean retail is firmly siding with $HOOD over the $AMC CEO.

Korean sources find the prospect of 40x leveraged Samsung / $SKHY a powerful enough argument to win the tokenization debate.

翻译：
从我的渠道调查来看，韩国散户明显倾向于支持 $HOOD，而不是 $AMC 的 CEO。

韩国消息人士认为，40 倍杠杆的三星 / $SKHY 这一前景足以成为赢得代币化辩论的有力论点。

2026-09-04 10:37

going down the list to eliminate competitor chokepoints internally by throwing subsidized spend into R&D. Or by acqusition as seen with China's acquisition of EU leaders like Ficontec.

America... they had all the cards initially, but I think they got too comfortable, and took too long to focus heavily on rare earths (recent funding is a good thing).

The earlier tariffs went the wrong way (socks, furniture, other exports), and pissed off allies in EU/Canada too.

Which could have been used for leverage for major chokepoints. But maybe they'll realize soon enough why allies are important.

Anyway, we'll see what happens, just my two cents about ongoing dynamics.

In an ideal world, everyone works together...

TLDR: Just some shower thoughts on how there's an unspoken race on who achieves supply chain independence first to gain leverage over the other.

翻译：
所以……我对美中紧张局势的看法是：

你有欧洲的垄断企业，例如：
- $ASML (EUV)
- $SOI (Photonics‑SOI)
- Zeiss SMT (ASML EUV 供应商)，Trumpf (ASML EUV 激光供应商)

日本拥有许多近乎垄断的企业，例如：
- TOK、JSR、ShinEstu、Fujifilm 主导 EUV 光刻胶，市场份额 90‑100%
- 与 Tokyo Electron 合作的涂布/显影设备，份额 90% 以上
- Lasertec 主导的光刻掩模检测，份额 100%
- HOYA、AGC 提供的 EUV 掩模基片，约 93% 市场份额
- arf 光刻胶
- Nittobo T‑Glass
- Ajinomoto 生产的 ABF 薄膜
- 以及其他特种玻璃、AGC、Disco 等公司的晶圆薄化/研磨/切割业务

还有美国的盟友国家，例如台湾的 $TSM，韩国的内存厂商 $SKHY / Samsung。

美国拥有 Synopsys、Cadence 等 EDA 公司，$LCRX、$KLA，$NVDA，以及众多下游巨头。

中国则面临诸多瓶颈，如镓、石墨、精炼锂等，同时在许多未来供应链（如机器人）上受益于规模化成本。

如果要谈美中供应链战争的看法：

- 美国正努力实现西方对中国/俄罗斯稀土及材料供应的独立，但进展太慢（本应在上个十年就优先推进）。
- 中国则在消除关键瓶颈依赖，并迫使美国依赖中国供应链（如 Wf6 所示）。

这是一场谁先实现供应链独立、从而对对方获得杠杆的竞争。

有趣的是，AI 加速在一定程度上为独立带来了未知变量。

以及开源努力（如 RISC‑V），中国正大力投入。因此出现了一个奇怪的悖论：美国理论上应支持开源硬件和开源 AI 的发展，却也被这些技术利用来对抗自身。

从中国的角度看，他们通过针对日本来施压美国供应链，进而提升西方对中国的供应链依赖。

并且通过向研发投入补贴或收购（如中国收购 EU 领头企业 Ficontec）来内部消除竞争对手的瓶颈。

美国……最初拥有所有牌，但我认为他们过于自满，且在稀土方面投入太慢（近期的资金投入是好事）。

早期的关税政策走向错误（袜子、家具等出口），也激怒了欧盟/加拿大的盟友。

本可以用来在关键瓶颈上获取杠杆，但也许他们很快就会意识到盟友的重要性。

2026-09-04 10:37

总之，未来会怎样我们拭目以待，这只是我对当前动态的两句随想。

理想情况下，大家可以合作……

TLDR：只是一点灵感，关于谁

2026-09-04 10:37

原文：
So... here's my take on US / China tensions:

You have European monopolies like:
- $ASML (EUV)
- $SOI (Photonics-SOI)
- Zeiss SMT (ASML EUV supplier), Trumpf (ASML EUV laser supplier)

Japan has many near/complete monoplies like:
- TOK, JSR, ShinEstu, Fujifilm for EUV photoresist - 90-100% share
- coat/develop equiment with Tokyo Electron -90%+
- actinic EUV patterned mask inspection (Lasertec) -100% share
- EUV mask blanks (HOYA, AGC) ~93% share
- arf photoresist
- Nittobo T-Glass
- ABF film (Ajinomoto)
- lot of misc from specialty glassAGC, wafer thinning/grinding/dicing with Disco and others.

Then there's US allied countries like Taiwan eg. $TSM, Korean memory with $SKHY / Samsung.

US has EDA with Synopsys/Cadence, $LCRX / $KLA, $NVDA, and many downstream giants.

China has many chokepoints such as gallium, graphite, refined lithium, and others + benefits from cost mass production over many future supply chains (eg. robotics).

If I had to give my opinion on US / China supply chain wars:

- US is trying to achieve Western independence from China/Russia rare earths + materials supply, but took too long (should have been a priority last decade).

- China is trying to eliminate strategic chokepoint dependence and forcing US reliance on Chinese supply chains (as seen with Wf6)

It's a race on who achieves supply chain independence first to gain leverage over the other.

And fun thing is, AI acceleration kinda throws an unknown variable in terms of speeding up independence.

As well as open source efforts (eg. RISC-V), which China is heavily focusing on. So it's a weird paradox where US should theoretically support open source hardware development + open source AI, but it's also being exploited against them.

From a Chinese perspective, they're pressuring US supply chains by targeting Japan, which increases Western supply chain reliance on China.

And

2026-09-04 09:14

原文：
China reportedly halts some rare earth shipments to US.

Not a new blanket ban, or any mention of specific rare earths.

But… if I had to guess, Japan is usually a decent indicator of what’s actively important since Japan:

- “sources virtually all of its dysprosium and terbium from China.”
- gallium, dysprosium, terbium and yttrium from China to Japan all stood at “zero” in June.
- Shin-Etsu Chemical has reportedly suspended acceptance of new orders for magnets containing dysprosium.
- Global magnet producer Proterial had received no export licenses for Chinese dysprosium as of last month

The Reuters piece mentioned yttrium, gallium, terbium, indium phosphide, and tungsten as historical examples (doesn’t say what’s being targeted now).

On a side note, $LYC.AX is becoming one of the most important alternatives, as Japan has locked up 75% of Lynas’s medium/heavy rare-earth output. (8 tons of dysprosium + terbium in Q1)

Something fun I also read is that Japan’s deep sea resource is enormous (est. at 16 million tons of rare earth rich mud, but commercialization is post-2028)

But in the meantime, China seem to be choking off feedstock that would materially affect Japan.

Since you’ve already seen the WF6 Japanese industry stressed from tungsten related halts.

And indium/others earlier this year that is used to make InP substrates from Sumitomo/JX.

But still gradually allowing exports directly eg. $AXTI -> $LITE to allow US/China trade to continue, kinda disrupting that from time to time, then using that as leverage.

It’s interesting since Japan has so many different monopolies like ABF, so we’ll see if they do anything back.

But if I had to give a prediction, it does look like the US has a countdown timer until China becomes self-sufficient on semiconductors.

TLDR:
- rare earth disruption to US
- China actively targeting Japan for upstream feedstock (which

2026-09-04 09:14

goes to Us supply chains), probably want to focus on that
- many many second/third order effects from disruptions
- we’ll see what US/Japan wants to do back, but supply chain wars are being fought upstream.

翻译：
据报道，中国暂停向美国发运部分稀土。

这不是新的全面禁令，也没有提到具体是哪种稀土。

但……如果让我猜，日本通常能很好地反映当前真正重要的东西，因为日本：

- “几乎所有的镝和铽都来自中国。”
- 6月中国向日本出口镓、镝、铽和钇都为“零”。
- 据报道，信越化学已暂停接受含镝磁体的新订单。
- 全球磁体生产商 Proterial 截至上月仍未获得中国镝的出口许可证

路透那篇文章提到了钇、镓、铽、磷化铟和钨作为历史案例（但没说现在具体瞄准什么）。

顺带一提，$LYC.AX 正在成为最重要的替代来源之一，因为日本已经锁定了 Lynas 75% 的中重稀土产量。（第一季度镝 + 铽共 8 吨）

我还读到一件有趣的事：日本的深海资源非常庞大（估计有 1600 万吨富含稀土的泥，但商业化要到 2028 年以后）

但在此期间，中国似乎正在切断会实质影响日本的原料供应。

你已经看到 WF6 因钨相关停供而给日本产业带来的压力。

今年早些时候，三菱/新日本制铁等也出现了用于制造 InP 基板的铟等材料短缺。

但与此同时，中国仍在逐步允许直接出口，例如 $AXTI -> $LITE，以便维持美中贸易继续进行，某种程度上时不时打断一下，再把这当作筹码。

这很有意思，因为日本有很多不同的垄断环节，比如 ABF，所以我们看看他们会不会采取什么反制。

但如果非要我给个预测，看起来美国距离中国在半导体上实现自给自足，确实还有个倒计时。

总之：
- 稀土供应受扰，影响美国
- 中国正主动针对日本的上游原料（这些会流向美国供应链），大概应该重点关注这一点
- 供应中断会引发很多二、三阶连锁影响
- 之后看看美日会怎么反制，但供应链战争是在上游打的

2026-09-03 19:12

原文：
OpenAI GPT-6 Astra benchmarks got released. Looks both AGI and the Singularity are almost here.

The leap in eval performance:

From 78.5% -> 100% on ExploitBench to 22.4% -> 64.6% on the Science benchmark is enormous.

And the most striking part is the VP of Research comment around:

"Astra was also OpenAI's first model that involved previous AI models very heavily in the training process, with the older generations helping train the new one."

So we also get the beginnings of recursive self-improvement as a bonus.

I don't think Bernie Sanders will be too happy about this?

翻译：
OpenAI GPT-6 Astra 基准已发布。看起来通用人工智能和奇点几乎已经到来。

评估性能的跃迁：

从 ExploitBench 的 78.5% 提升到 100%，在 Science 基准上从 22.4% 提升到 64.6%，幅度巨大。

最引人注目的是研究副总裁的评论：

“Astra 也是 OpenAI 第一个在训练过程中大量使用前代 AI 模型的模型，旧一代帮助训练新一代。”

这也意味着我们获得了递归自我改进的雏形。

我觉得伯尼·桑德斯可能不会对此感到高兴。

2026-09-03 18:30

年代用于相机、打印机、汽车等场景的老旧内存，现在已经没多少人生产了。）

甚至像 Winbond 这种老 DDR2 供应商，也在逐步退出市场……

所以 ESMT 其实是 DDR3/DDR2 缺货的受益者。

TLDR：

ESMT 8 月营收约 2.49 亿美元，高于上月的约 2.14 亿美元。

2026-09-03 18:30

原文：
Just an update to my DDR2/DDR3 memory idea with ESMT. They reported August revenue (Taiwan sometimes does monthly reporting):

June revenue: ~$153M
July: ~$214M
August (new): ~$249M vs. $35.3m (last year)

Monthly revenue has increased ~63% in two months despite reportedly similar 2026 wafer allocations to 2025 (MoneyDJ report).

The profitability is already really high as seen with July's monthly revenue (where they reported ~$109M net income).

So if pricing holds, and revenue increases, that would be very nice...

Now a lot of US people don't really care since it's in Taiwan, but if you're in the $MU / $SKHY, $SNDK trade.

Just educational purposes, it's a second order effect where your leading memory players prioritize AI/DC higher end memory and reduce allocations to older generations (causing shortages).

But... DDR4 since prices going up too much, so older gen memory products (DDR4) are getting redesigned with older gen memory (DDR3). And DDR3 products with DDR2.

And so on. Causing cascading supply chain shortages.

(DDR2 for example is boring/old memory used for cameras, printers, automotive back in the 2000s and not many ppl make them anymore).

And even your old DDR2 suppliers like Winbond are gradually withdrawing from the market too...

So ESMT is the beneficiary of DDR3/DDR2 shortages.

TLDR:

ESMT August revenue was ~$249M, up from ~$214M last month.

翻译：
只是更新一下我关于 ESMT 的 DDR2/DDR3 内存想法。他们公布了 8 月营收（台湾有时会按月披露）：

6 月营收：约 1.53 亿美元
7 月：约 2.14 亿美元
8 月（最新）：约 2.49 亿美元，去年同期为 3530 万美元

尽管据报 2026 年晶圆分配与 2025 年大致相近（MoneyDJ 报道），月营收在两个月内仍增长了约 63%。

从 7 月月营收就能看出，盈利能力已经非常高了（当时他们披露的净利润约为 1.09 亿美元）。

所以如果价格能稳住，营收还能继续增长，那就会非常不错……

现在很多美国人其实不太在意，因为这是台湾的公司，但如果你在做 $MU / $SKHY、$SNDK 这类交易。

纯属科普，这是一个二阶传导效应：头部存储厂商会把 AI/DC 更高端的存储优先级提得更高，减少对老一代产品的分配（从而造成短缺）。

但……DDR4 由于价格涨得太多，老一代内存产品（DDR4）又开始改回使用更老一代内存（DDR3）。而 DDR3 产品又转向 DDR2。

以此类推。于是就会引发连锁式的供应链短缺。

（比如 DDR2，就是 2000

2026-09-03 17:31

原文：
$DELL founder is a legend.

The Dell family pledged $6.25B / $250 each to 25M children through Trump Accounts, with $HOOD (up 14.83%+ today) as the initial sole brokerage.

Little off topic but getting the public invested early is very meaningful compared to the dollar amount. https://t.co/Xwrp7lMPyS

翻译：
$DELL 创始人是传奇。

Dell 家族通过 Trump Accounts 向 2500 万儿童承诺 62.5 亿美元 / 每人 250 美元，$HOOD（今日涨幅 14.83%+）作为唯一初始经纪商。

稍微离题，但让公众提前参与的意义远超金额本身。 https://t.co/Xwrp7lMPyS

2026-09-03 16:19

原文：
This is the correct response when looking at $SIVE >100m CW DFB extraordinary capacity disclosure.

To give some context, Trendforce July data cites global CW/EML capacity is roughly 608.4M/year (~50.7M monthly units annualized).

With $AVGO, $LITE, and Sumitomo making up ~335m/year combined.

By Q4 2027, it's likely a lot of these players will significantly expand capacity again, with broadcom citing 3x Y/Y growth. (also capacity not being 1:1 since many are allocated to EML or UHP CWs, etc).

But this capacity target would place Sivers in global Tier 1 laser supplier category.

With possibly low double digits global laser market share if you factor in their hybrid 2:1 manufacturing model.

This new disclosure is very exciting coming from a conservative company.

And sheds more light on why the $MRVL SVP cited Europe as a major geography when talking about laser supply chain sourcing.

翻译：
这是在查看 $SIVE >100m CW DFB 超常容量披露时的正确回应。

提供一些背景，Trendforce 七月数据指出全球 CW/EML 产能约为 608.4M/年（约合年化 50.7M 月产量）。

其中 $AVGO、$LITE 和 Sumitomo 合计约为 ~335m/年。

到 2027 年第四季度，这些厂商很可能再次大幅扩产，Broadcom 提到同比增长 3 倍。（此外产能并非 1:1，因为许多被分配给 EML 或 UHP CW 等）

但该产能目标将使 Sivers 进入全球一级激光供应商行列。

若考虑其 2:1 混合制造模式，可能在全球激光市场占有低双位数的份额。

这份来自保守公司的新披露令人非常振奋。

也进一步说明了 $MRVL SVP 在谈及激光供应链采购时为何将欧洲列为重要地区。

2026-09-03 12:35

原文：
$NVDA acquires hugging face for $12,930,300,000.

Jensen stated the reason being:
- 18M+ devs use Hugging Face
- 3M+ models, 500k+ datasets, 1M+ apps
- 200k+ companies use the platformm
- support of open weight models

this almost reminds me of $MSFT buying Github? https://t.co/8FhE7b9mU0

翻译：
$NVDA 收购 Hugging Face，金额为 $12,930,300,000。

Jensen 表示原因是：
- 1800 万以上开发者使用 Hugging Face
- 300 万以上模型，50 万以上数据集，100 万以上应用
- 20 万以上公司使用该平台
- 支持开源权重模型

这几乎让我想起 $MSFT 收购 Github？ https://t.co/8FhE7b9mU0

2026-09-03 10:35

原文：
$IQE is actually a quantum dot foundry leader believe it or not.

Today IQE announced a purchase agreement with Quintessent (top player in the QD space).

“As quantum dot lasers move to customer sampling.”

Your QD bros like $ALMU, QD Laser (6613), and others might finally start to see the light of day, with broader commercialization in the next few years.

Second order effects are QD machine chokepoints like $ALRIB that might capture more attention.

But TAM is typically pretty small on the QD material epi side of things. (No disclosure over $ figure), but good news nevertheless.

翻译：
$IQE 实际上是量子点代工龙头，信不信由你。

今天，IQE 宣布与 Quintessent（QD 领域的顶级玩家）达成收购协议。

“随着量子点激光器进入客户采样阶段。”

你的 QD 兄弟们如 $ALMU、QD Laser（6613）等，可能终于迎来曙光，未来几年将实现更广泛的商业化。

第二层效应是像 $ALRIB 这样的量子点机器瓶颈，可能会受到更多关注。

但在量子点材料外延方面，TAM 通常相当小。（未披露具体金额），但仍是好消息。

2026-09-03 10:07

原文：
$SIVE announced today that it is expanding InP manufacturing in Glasgow.

The new capacity target is:

~100 million CW DFB laser annual production.

Using Sivers’ historical $50-$100 per 8-laser array pricing, the implied revenue capacity might be ~$625M-$1.25B/year (not guidance, just illustrative modeling)

I’m personally very surprised by the amount of capacity coming online from their new hybrid manufacturing model, targeted Q4 2027.

Especially during an industry shortage, mentioned by $AVGO earnings call today.

翻译：
$SIVE 今日宣布将在格拉斯哥扩大 InP 生产。

新的产能目标是：

~1亿 CW DFB 激光器年产量。

按照 Sivers 过去每 8 激光阵列 $50-$100 的定价，暗示的收入规模可能在约 $625M-$1.25B/年（非指引，仅为示例模型）

我个人对他们新混合制造模式将在 Q4 2027 投产的产能规模感到非常惊讶。

尤其是在今天 $AVGO 财报电话会议提到的行业短缺期间。

2026-09-02 22:39

原文：
$AVGO CEO: "Demand for lasers, whether it is EML lasers, CW lasers.

Is far surpassing supply out there in the industry"

Pretty material coming from one of the largest laser suppliers in the world... Even while they 3x laser capacity.

Exciting validation for laser companies. https://t.co/l4alSsexMT

翻译：
$AVGO CEO：“无论是EML激光器还是连续波激光器，需求
远超行业供应。”

来自全球最大激光供应商之一的精彩内容……即使他们将激光产能提升了3倍。

对激光公司来说是令人振奋的验证。 https://t.co/l4alSsexMT

Loading...