觅信聚合

我的关注

2026-09-30 22:33

原文：
$GOOGL Gemini 4 Argon benchmarks got released.

Against GPT-6 Astra + Anthropic Fable 5.1 / Opus 5.5.

Still coughing nonstop, so will comment on $MU / $JBL earnings later.

I've always thought Google's internal models were stellar, but anecdotally, they seem to get hard nerfed shortly after public release... Maybe because of compute constraints?

But yeah, looks like Google is back.

翻译：
$GOOGL Gemini 4 Argon 基准已发布。

对比 GPT-6 Astra + Anthropic Fable 5.1 / Opus 5.5。

仍在不停咳嗽，所以稍后会评论 $MU / $JBL 财报。

我一直认为 Google 的内部模型很出色，但据非正式经验，它们在公开发布后不久就会被大幅削弱……可能是算力限制？

不过，看起来 Google 已经回来了。

2026-09-30 04:33

原文：
Bro, $AMZN semi ETF portfolio is probably going to end up more compelling than what $NVDA owns...

Today Amazon signed a $49.8M private placement into PCB maker Gold Circuit Electronics (2368). So now Amazon has ownership/warrants in:

- $AAOI
- Alchip
- $MRVL (Celestial)
- $ALAB
- $CRDO
- $JBL
- $FLEX
- $STM
- $CBRS
- $GNRC
- $FN
- TD SYNNEX
- $QCOM
- Gold Circuit Electronics

And others..

If we look at the private placement amount for GCE in speciifc, it's like .36% of the company... Kinda small financially, meaningful strategically. Esp. since there were claims that GCE supplies ASIC-server PCBs to the four major hyperscaler CSPs...

Regardless it's pretty funny to see Amazon to Nvidia start to own all the worlds leading companies.

Since Amazon has exposure to the entire supply chain like PCBs now, optical fabric with Celestial, lasers with $AAOI, assembly from $FN to $JBL, ASIC partners like $QCOM / ALchip.

翻译：
兄弟，$AMZN 半导体ETF组合可能最终比$NVDA的更具吸引力...

今天亚马逊向PCB制造商 Gold Circuit Electronics (2368) 进行了一笔 4980 万美元的私募。于是亚马逊现在拥有以下股份/认股权：

- $AAOI
- Alchip
- $MRVL (Celestial)
- $ALAB
- $CRDO
- $JBL
- $FLEX
- $STM
- $CBRS
- $GNRC
- $FN
- TD SYNNEX
- $QCOM
- Gold Circuit Electronics

以及其他...

如果具体看 GCE 的私募金额，仅占公司约 0.36%……在财务上虽小，但在战略上意义重大。尤其是因为有人称 GCE 为四大超大规模云服务提供商提供 ASIC‑服务器 PCB...

不管怎么说，看到亚马逊开始拥有所有全球领先公司的情况，真是有趣。

因为亚马逊现在已经涉及整个供应链，如 PCB、Celestial 的光学织物、$AAOI 的激光、从 $FN 到 $JBL 的组装、以及 $QCOM / Alchip 等 ASIC 合作伙伴。

2026-09-28 22:32

World Labs 的消息……但现在身体很不舒服，先休息一下。

我记得很早就说过物理 AI + 世界模型会带来重大变革，或许当时应该投资私营公司……

除此之外……还有其他有趣的新闻：

- 来自 ECOC 的一些渠道调查显示，$NVDA 似乎已经锁定了 $LITE Greensboro UHP 激光产能的 50%？并且据 Barclays 称可能还有合同让他们锁定剩余产能。

这相当于 50亿美元的激光（年产能）……看到 Nvidia 通过这种方式限制整个行业的产能真是滑稽。但这也提升了其他商用激光供应商的战略价值，并让 Lumentum 更受关注。

这有点像分配的博弈论，因为如果你把所有 HBM、激光等都抢走，竞争对手的项目规模就会变得更难扩展。

- 还有 Nvidia 规模达 1500亿美元的授权回购（只要看看 $AAPL 回购时的情况……）他们在印太多钱。

- 同时 Samsung Electro-Mechanics 正在扩展其 FC-BGA 基板产能，但资金来自客户。

我还记得 $MRVL 在 SEC 文件中提到大尺寸基板是其项目的核心瓶颈之一（以及 $TSM 的单一来源

2026-09-28 22:32

原文：
Everyone is asking "Where is Serenity" or "Who is Serenity". But nobody is asking...

"How is Serenity"?

I do read things from $AMD $8.2B acquisition of World Labs... but feeling pretty sick right now, so taking a little break.

I remember saying physical AI + world models would major shift way back when, so maybe I should have invested in private companies at the time...

Apart from that... other interesting news:

- Some channel checks from ECOC was that $NVDA apparently secured 50% of $LITE Greensboro UHP laser capacity? And possibly contracts that allow them to secure the remaining capacity per Barclays.

That's $5B in lasers (annual capacity)... so it's hilarious to see Nvidia go out and bottleneck the rest of the industry by doing this. But this just puts more strategic value on other merchant laser suppliers + more visibility for Lumentum.

It's kinda game theory over allocation, since if you secure all the HBM, lasers, and others it inadvertently makes it much harder for competitors to scale their programs.

- Then there's your massive $150B Nvidia authorized buybacks (just look at $AAPL when they did buybacks..) They're printing too much.

- What's also interesting is that Samsung Electro-Mechanics is expanding their FC-BGA substrate capacity today... but customers are the one funding this.

I also remembered $MRVL in SEC filings cited large body substrates as one of the core bottlenecks of their program (alongside $TSM sole source dependencies). So they also went out and did some other expansion agreements with related companies like AT&S recently.

- OpenAI delaying Astra-6.1 October release over safety concerns just now. Nvidia also happened to launch an AI safety platform.

I don't think this is meaningfully bearish that the models are just getting 2 good.

翻译：
大家都在问“Serenity 在哪儿”或“Serenity 是谁”。但没人问……

“Serenity 怎么样？”

我确实在关注 $AMD 82亿美元收购

2026-09-25 16:06

原文：
People really need to stop thinking about 5% yields over 10 years or 30 years.

Even the inflation of my $SUBWAY sandwich has gone up like 12% a year.

Today it was like $20 after tax, when it used to be like $5… 12 years ago.

The best way to keep up with inflation of the things that matter, like Sandwiches, is equities like $SPY.

翻译：
人们真的该停止只盯着 10 年或 30 年的 5% 收益。
甚至我那份 $SUBWAY 三明治的通胀率也每年涨到约 12%。
今天税后要价约 $20，而 12 年前才只有 $5…
跟上重要事物（比如三明治）通胀的最佳方式是持有像 $SPY 这样的股票。

2026-09-25 10:06

原文：
我突然意识到, 原来我有这么多粉丝.

第一次从第三方的视角看了一下自己的主页…感觉好像还有点吓人?

如果我们办一场派对, 而且所有人都来了, 那应该不是塞满一个小房间的问题, 而是能塞满一整座城市吧?

总之, 中秋节快乐！

翻译：
我突然意识到，原来我有这么多粉丝。

第一次从第三方的视角看了一下自己的主页…感觉好像还有点吓人？

如果我们办一场派对，而且所有人都来了，那应该不是塞满一个小房间的问题，而是能塞满一整座城市吧？

总之，中秋节快乐！

2026-09-25 02:08

原文：
$NVDA Groq3 sole source wafer test supplier looks like Doosan Tesna (~$1.3B MC, KR 131970)

Groq looks like it’s mass production now, so it should be material to this company’s revenue growth moving forward.

They also do test historically for Samsung, sk Hynix, Tesla, Apple, and others.

Just something interesting I came across (no positions)

翻译：
$NVDA Groq3 唯一来源的晶圆测试供应商看起来像 Doosan Tesna（约 $1.3B 市值，KR 131970）

Groq 看起来已经进入量产，因此对该公司未来的收入增长应是重要因素。

他们过去也为 Samsung、SK Hynix、Tesla、Apple 等进行测试。

只是我偶然发现的有趣信息（不构成持仓）

2026-09-24 21:17

原文：
$AKAM signs $11.6B compute deal with Anthropic for accelerated CPU workloads demands over 7 years.

$AMD to $MU should be happy to hear this…

Akamai ests. ~$5.5B of capex to service the $11.6B commitment. Including +$1.7B of 2026 capex specifically to pre-purchase supply chain components like memory.

So much for Anthropic/OpenAI calling for an AI slowdown?

翻译：
$AKAM 与 Anthropic 签署 116亿美元计算协议，以满足未来 7 年加速的 CPU 工作负载需求。

$AMD 和 $MU 听到这个应该会很高兴……

Akamai 估计约 55亿美元资本支出用于支持 116亿美元的承诺。其中包括 2026 年额外 17亿美元的资本支出，专门用于预先采购供应链组件，如内存。

这下可真是对 Anthropic/OpenAI 呼吁 AI 放缓的回应啊？

2026-09-24 19:38

原文：
If you're curious about the random $SPY / $QQQ candle:

Just now, “US and Iran discuss phased deal to reopen Hormuz and end US blockade" - Reuters

- "A deal before the midterms could benefit Trump, as reopening Hormuz could ease tensions in the Gulf, contain oil prices and help bring down politically sensitive US gasoline prices."

- "Trump has said he thinks the US and Iran could reach a deal after midterm elections to Congress on November.”

Would make sense for Trump to want markets as strong as possible during Midterms elections…

翻译：
如果你对随机的 $SPY / $QQQ K线感到好奇：

刚刚，“美国和伊朗讨论分阶段协议以重新开放霍尔木兹海峡并结束美国封锁” - Reuters

- “在中期选举前达成协议可能有利于特朗普，因为重新开放霍尔木兹海峡可以缓解海湾紧张局势，抑制油价，并帮助降低政治敏感的美国汽油价格。”

- “特朗普表示，他认为美国和伊朗可以在十一月的中期选举后达成协议。”

让特朗普希望在中期选举期间让市场尽可能强劲是合情合理的。

2026-09-24 00:57

原文：
I’ve actually entered positions in many tiny Korean companies recently…

That actively supply memory, test, machines, etc. to $MU, $SKHY, Samsung, and others like $SMTC + US companies.

It’s pretty interesting what you find in the $30-$250m range for optical components, CPUs, memory, and others that are inflecting due to AI.

I don’t share these names publicly since risks/volatility is enormous.

But one or two going from $40M -> $400m makes the risk/reward worth it for me personally.

翻译：
我最近实际上在许多小型韩国公司建仓……

这些公司积极向 $MU、$SKHY、三星以及类似 $SMTC 的美国公司等供应内存、测试设备等。

在 $30-$250m 规模的光学元件、CPU、内存等因 AI 而出现拐点的领域，发现的东西相当有趣。

我不会公开这些公司名称，因为风险/波动极大。

但有一两只从 $40M -> $400m 的公司，使得对我个人而言风险/回报值得。

2026-09-23 05:07

原文：
As for a potential Muse trade, $META CPU designer was Socionext apparently from MS channel checks back in 2024.

They're heavily tied to $ARM AGI CPU programs and actually going out to secure $TSM + $INTC capacity (this part is very valuable).

Small Japanese chip designer ~$2.2B with 2 hyperscaler connections, which is rare.

But for high sensitivity $META CPU + Muse exposure, this company is probably ideal to look at.

They also have ongoing programs with $GOOGL Quantum (SoC) + Waymo. I'd treat Google's program as noise, since Quantum + Robotaxis sounds cool but doesn't scale much near term.

As for why it's in the dumps, I don't quite think markets realize they're tied to Meta yet since it's not publicly disclosed.

Anyway, personally no positions, just sharing. Probably a better M&A target by $AVGO / GUC and others though...

There were some external GUC projections for $GOOGL CPU to reach $2.7B or so, so if the Meta program takes off, it could cook (TAM wise from a hyperscaler CPU program).

TLDR: $2.2B Japanese company likely higher sensitivity to $ARM CPU ramp + $META CPU as a potential Muse trade.

Economic exposure is really hard to quantify since projections don't go very far out. I'd normally take a bet but earnings visibility from other companies just seem a bit more compelling.

翻译：
关于潜在的 Muse 交易，$META 的 CPU 设计公司似乎是 Socionext，来自 2024 年的 MS 渠道调查。

他们与 $ARM 的 AGI CPU 项目关系密切，并且正在争取 $TSM + $INTC 的产能（这部分价值很高）。

这是一家约 22 亿美元的日本小型芯片设计公司，拥有 2 家超大规模云服务商的连接，这很罕见。

但对于高度敏感的 $META CPU + Muse 曝光来说，这家公司可能是理想的关注对象。

他们还与 $GOOGL Quantum（SoC）+ Waymo 有在研项目。我会把 Google 的项目视为噪音，因为 Quantum + 自动驾驶出租车听起来很酷，但短期内难以规模化。

至于为何股价跌至低位，我认为市场尚未意识到他们已与 Meta 绑定，因为尚未公开披露。

总之，我个人没有持仓，仅作分享。可能更适合作为 $AVGO / GUC 等的并购目标……

有外部 GUC 预测 $GOOGL 的 CPU 业务将达到约 27 亿美元，如果 Meta 项目起飞，市场规模将大幅增长（来自超大规模云服务商的 CPU 项目）。

TLDR：这家约 22 亿美元的日本公司对 $ARM CPU 增长和 $META CPU 的敏感度更高，可能是潜在的 Muse 交易。

2026-09-23 05:07

经济敞口很难量化，因为预测的时间跨度不长。我本会下注，但其他公司的

2026-09-23 01:10

原文：
Elon Musk: "AI will beat [humans] in all fields by the end of next year or maybe 2028, at the latest."

Kind of maps to Anthropic's economic implications paper where 2026-2028 was the buildout.

Then 2028-2030 was unprecedented economic growth.

So we're still early H2 2026. https://t.co/LpGJkHcgn2

翻译：
Elon Musk：“AI将在明年年底前或最迟到2028年在所有领域击败[humans]。”

这与Anthropic的经济影响论文相吻合，其中2026-2028年是建设阶段。

随后2028-2030年出现前所未有的经济增长。

所以我们仍处于2026年下半年早期。 https://t.co/LpGJkHcgn2

2026-09-22 16:00

原文：
I guess I need to start calling $SNDK or $LITE Super Intelligence stocks.

Thanks Trump. https://t.co/x7tJcKjfsc

翻译：
我想我应该开始把 $SNDK 或 $LITE 称为超级智能股票。

谢谢 Trump。 https://t.co/x7tJcKjfsc

2026-09-22 13:37

原文：
I feel like traditional "value investing" went into hibernation.

Before it was finding a 10x -12x P/E company like $UPWK or $PYPL, Dominoes, Pepsi. Or even a cheap Japanese company below book value.

Then just enjoying slow growth but buyback/dividend appreciation from their cashflow.

Now you can find 1.8x runrate P/E company like $ESMT that grew revenue 605.24% Y/Y.

And the main thing you need to argue about is duration.

Or a 16.2x trailing P/E Sumitomo, despite having way too many hyperscaler LTAs in AI from fiber to lasers.

AI created much bigger medium-term anomaly, where "value investing" went towards high growth, "deep value" companies.

翻译：
我觉得传统的“价值投资”已经进入休眠。

过去它会寻找市盈率 10‑12 倍的公司，比如 $UPWK、$PYPL、Dominoes、Pepsi，甚至是账面价值以下的日本廉价公司。

然后只享受慢速增长以及来自现金流的回购/股息增值。

现在你可以找到市盈率 1.8 倍的 $ESMT，收入同比增长 605.24%。

而你需要争论的核心是持有期限。

或者是市盈率 16.2 倍的 Sumitomo，尽管它在 AI 领域从光纤到激光拥有过多的超大规模长期协议。

AI 产生了更大的中期异常，使得“价值投资”转向高增长的“深度价值”公司。

2026-09-22 12:18

原文：
ASIC supply chain from $MSFT Maia to $GOOGl TPUs + timeline mapping with $AVGO, Mediatek, GUC, Alchip $MRVL, and others.

Along with your assembly players (Accton, $FN, $CLS) + Wiwynn, Foxconn.

Haven't really seen much presented in a nice format so just wanted to share.

I did find Alchip to be particularly interesting since they have Trainium 3 ramping now and Trainium 4 next year. (source: Macquarie)

翻译：
ASIC 供应链从 $MSFT Maia 到 $GOOGl TPU + 时间线映射，涉及 $AVGO、Mediatek、GUC、Alchip、$MRVL 等。

以及你的组装厂商（Accton、$FN、$CLS）+ Wiwynn、Foxconn。

目前还没有看到以良好格式呈现的内容，所以想分享一下。

我发现 Alchip 特别有意思，因为他们的 Trainium 3 正在 ramp up，而 Trainium 4 明年推出。（来源：Macquarie）

2026-09-22 05:21

原文：
Wow Bitcoin + Ethereum pulled pretty extreme recoveries.

Looks like $HOOD / $COIN / Crypto is back?

On-chain equities trading + regulator exemptions probably put a lot of fuel into the fire for the sector.

That aside... crypto is usually a great indicator for high-beta AI equities believe it or not (they tend to frontrun broader market drops or rallies).

So overall market risk appetite might be coming back despite the rate hikes.

翻译：
哇，Bitcoin + Ethereum 实现了相当剧烈的反弹。

看起来 $HOOD / $COIN / Crypto 正在回暖？

链上股票交易 + 监管豁免可能为该行业注入了大量燃料。

除此之外…… crypto 通常是高贝塔 AI 股票的极佳指标，信不信由你（它们往往领先于更广泛的市场跌宕起伏）。

因此，尽管加息，整体市场风险偏好可能正在回升。

2026-09-21 19:14

原文：
We’re in the modern agricultural revolution.

In terms of civilization-level impact of AI.

So… not sure how anyone can be bearish so early on into the singularity and before mass deployment of physical AI. https://t.co/uJAXpZm4MP

翻译：
我们正处于现代农业革命中。

从文明层面的 AI 影响来看。

所以……我不明白为什么有人会在奇点初期、实体 AI 大规模部署之前就持悲观态度。 https://t.co/uJAXpZm4MP

2026-09-21 15:49

原文：
I'm not having a fun day since I got sick today. But the bottlenecks are having a good time:

$AMD +9.13% | $INTC +14.16% | $ARM +14.33%

CPUs go brrr

$SIVE +10.32% | $AAOI +3.39% | AMS OSRAM +24.61%

Lasers go brrr

$EWY +4.08% | $DRAM +2.83% | $MU +2.23%

Memory go brrr

$NBIS +4.7% | $CRWV +5.03% | $IREN +1.73%

Compute go brrr

Probably most material developments were CPU:GPU ratio expectations + CPU demand (from new AI releases).

As well as optical developments from ECOC 2026 (as seen with $MRVL PR + others)

Still a lot to catch up on.

翻译：
我今天生病了，今天过得不太开心。不过瓶颈们倒是玩得很开心：

$AMD +9.13% | $INTC +14.16% | $ARM +14.33%

CPU 轰轰作响

$SIVE +10.32% | $AAOI +3.39% | AMS OSRAM +24.61%

激光轰轰作响

$EWY +4.08% | $DRAM +2.83% | $MU +2.23%

内存轰轰作响

$NBIS +4.7% | $CRWV +5.03% | $IREN +1.73%

计算轰轰作响

可能大多数材料发展都是 CPU:GPU 比例预期 + CPU 需求（来自新 AI 发布）。

以及来自 ECOC 2026 的光学发展（如 $MRVL PR 等）

仍有很多需要跟进。

2026-09-21 05:27

原文：
The main duo $SKHY and Samsung are having a fun day.

I'm still long memory (eg. $EWY / $DRAM ) since if capacity agreements extend 3-5 years... and your analysts model 2.8-3.3x 2027E. Risk/reward is attractive over time.

But I think legacy memory a very interesting trade currently due to ASP hike potential..

It also just so happens these same companies procure memory from even smaller memory companies (that might have higher sensitivity)...

So if you want to, you can track $PSMC wafer allocations, and find some interesting companies that likely replace $MU discontinued legacy memory through Alliance or supply directly to Samsung.

Won't mention them on X, but just throwing out some research ideas.

翻译：
主要的两大玩家 $SKHY 和 Samsung 正在玩得很开心。

我仍然看好内存（例如 $EWY / $DRAM），因为如果产能协议延长 3-5 年……而且你的分析师给出 2027E 2.8-3.3 倍的模型。风险/回报随时间具有吸引力。

但我认为传统内存目前是一个非常有趣的交易，因为其可能的 ASP 上涨。

恰好这些公司也从更小的内存公司采购内存（可能更敏感）……

所以如果你愿意，可以跟踪 $PSMC 的晶圆分配，找到一些可能通过 Alliance 替代 $MU 已停产的传统内存，或直接供应给 Samsung 的有趣公司。

不会在 X 上提及它们，只是抛出一些研究思路。

Loading...