觅信聚合

我的关注

2026-07-26 20:12

已经开始大规模生产，并将在 2026 年第三季度开始发货。包括与 Anthropic 部署高达 2GW 的 MI455X GPU 和与 OpenAI 部署 6GW 的基础设施。预计 CPU 市场的复合年增长率（CAGR）将达到 50% 以上，到 2030 年将达到 2200 亿美元，高于 2025 年的 260 亿美元。

- 2027 年，AMD 将推出 Mi500 的光学互连器。根据渠道检查，AMD 正在朝着 CPO 路线发展（似乎可能使用 Ayar）。

- $ORCL 获得了 70 亿美元的战争部企业软件合同。

- JX 金属在其韩国子公司中将半导体目标容量增加了一倍，投资 40 亿日元，预计 2027 年下半年开始运营，主要客户三星电子和 SK 海力士的需求激增。

- 钨六氟化物的现货价格在日本 Kanto Denka 和 Chuo Gas 宣布永久停产后，同比上涨了 2.1-2.5 倍。由于 3M 计划退出 PFAS 生产，氟化液体供应面临真空。

- 从四家光学芯片制造商的盈利来看，Yuanjie/Eoptolink/TFC Optical/Dongshan Precision：没有重大订单削减，预计 1.6T 的发货量将在 2027 年加速。光学芯片供应商预计将从短缺中获得超额利润。

- Unitree Robotics 目标在 2026 年实现 30,000 个人形机器人的生产能力。阅读关于人形机器人市场规模扩张的内容，例如 $CCXI 等。

- 中国的能源存储锂电池订单在上半年增加了 2900%。不太熟悉 EVE Energy 和其他电池制造商。

2026-07-24 20:23

原文：
There you have it, $GOOGL Waymo to explore split with $UBER per FT.

My view last year was that Uber partnering with Waymo was a bad choice… since Uber would just serve as a distribution funnel that Google needed as they scale up.

And that $NBIS Avride, other neutral players, or vertical integration would be the optimal path forward.

Now their own partner Waymo looks to be the strongest disruptor to Uber.

翻译：
就这样，$GOOGL 的 Waymo 计划与 $UBER 分道扬镳，据 FT 报道。

去年我的看法是，Uber 与 Waymo 合作是一个错误的选择…… 因为 Uber 只会成为 Google 扩大规模时需要的分销渠道。

而 $NBIS Avride、其他中立玩家或垂直整合将是最优的前进路径。

现在，他们自己的合作伙伴 Waymo 看起来将成为对 Uber 最强的颠覆者。

2026-07-24 16:23

原文：
Jensen from $NVDA has joined X for the first time to share his letter on open models.

翻译：
$NVDA 的 Jensen 首次加入 X ，分享了他关于开放模型的信。

2026-07-24 00:12

原文：
Not exactly for $AAOI (they're working on CPO too, just a little further behind in commercialization).

There's a podcast that featured the $SIVE CEO recently called: "Market Insights: Photonics & AI".

Most interesting take was:

"Pluggables will be around for the next 10 years".

There's a lot of CPO + RNG/MRC optimization TAM cannablization arguments for pluggables recently, but looks like it's here to stay.

And that there's going to be multiple architecture paths going forward (CEO also NPO alongside CPO as an major opportunity in the next 5 years).

As for challenges: Vickram stated manufacturing capacity + availability of raw material is the main bottleneck. Likely referencing the existing bottlenecks with upstream CW laser ramp.

From the podcast, seems like a large focus was on pluggable markets due to larger near-term TAM (which is understandable with $JBL + other players), with CPO being the north star.

So it might be interesting if $SIVE M&A ambitions were to acquire more pluggable optical transceiver IP, like what $LITE did with Cloud Light.

But TLDR:

Lot of different paths moving forward, pluggable here to stay, CPO seems like North Star for where industry is heading.

Feels like there's room for every qualified laser supplier given the industry demand is growing faster than what they can all collectively make.

翻译：
并非完全针对 $AAOI（他们也在开发 CPO，只是商业化方面稍微落后了一些）。

有一期播客最近采访了 $SIVE 的 CEO，题目是：“市场洞察：光子学和 AI”。

最有趣的观点是：

“可插拔设备将在未来 10 年内继续存在”。

最近，人们对可插拔设备、CPO 和 RNG/MRC 优化的市场规模进行了很多讨论，但看起来可插拔设备将继续存在。

并且，未来将会有多条架构路径（CEO 也提到 NPO 是一个重要的机会，除了 CPO，在接下来的 5 年内）。

至于挑战：Vickram 表示，制造能力和原材料的可用性是主要的瓶颈。可能是指现有的 CW 激光扩产瓶颈。

从播客来看，似乎重点关注的是可插拔市场，因为其近期的市场规模更大（这可以理解，$JBL 和其他玩家也在关注这一点），而 CPO 是行业的目标。

所以，如果 $SIVE 的并购雄心是收购更多的可插拔光学收发器 IP，就像 $LITE 收购 Cloud Light 一样，那将会很有趣。

总的来说：

未来将会有很多不同的发展路径，可插拔设备将继续存在，CPO 似乎是行业发展的目标。

感觉对于每个合格的激光供应商来说，都有发展空间，因为行业需求的增长速度快于他们的总产能。

2026-07-23 23:52

原文：
RT @AMD: The feeling’s mutual. 🤝

翻译：
转发 @AMD: 这种感觉是相互的。 🤝

2026-07-23 20:48

原文：
From $AMD Advancing AI, channel checks:

Helios networking partners told me they expect AMD to pursue CPO for future generations of Helios scale up networking.

And expects other hyperscalers + industry to go down the CPO route.

One AI infra executive at a company involved in a hyperscaler CPO programs said they had seen no indication of CPO scale up supply chain delays.

Pushing back on recent 3rd party delay reports.

This was informal commentary, not official AMD or partner guidance.

翻译：
来自 $AMD 推进人工智能的渠道检查：

Helios 网络合作伙伴告诉我，他们预计 AMD 将在未来几代 Helios 大规模网络中追求 CPO。

并预计其他超大规模企业和行业也将走向 CPO 路线。

一家参与超大规模企业 CPO 计划的人工智能基础设施高管表示，他们没有看到 CPO 大规模供应链延迟的迹象。

这是在反驳最近的第三方延迟报告。

这只是非正式的评论，而不是官方的 AMD 或合作伙伴指导。

2026-07-23 20:46

原文：
From $AMD Advancing AI, channel checks:

Helios networking partners told me they expect AMD to pursue CPO for future generations of Helios scale up networking.

One AI infra executive at a company involved in a hyperscaler CPO programs said they had seen no indication of CPO scale up supply chain delays.

Pushing back on recent 3rd party delay reports.

This was informal commentary, not official AMD or partner guidance.

翻译：
来自 $AMD 推进人工智能，渠道检查：

Helios 网络合作伙伴告诉我，他们预计 AMD 将在未来几代 Helios 大规模网络中追求 CPO。

一家参与超大规模 CPO 计划的人工智能基础设施高管表示，他们没有看到 CPO 大规模供应链延迟的迹象。

对最近的第三方延迟报告提出质疑。

这只是非正式的评论，而不是官方的 AMD 或合作伙伴指导。

2026-07-23 20:04

原文：
RT @AMD: The feelings mutual. 🤝

翻译：
转发 @AMD: 这种感觉是相互的。 🤝

2026-07-23 19:49

原文：
$AMD reached out directly to hand me some special swag!

Apparently they only made 30 of these bandanas?

I love AMD and Lisa Su now. https://t.co/KgcYjBUcTv

翻译：
$AMD 直接联系我并赠送了一些特别的周边商品！

显然，他们只制作了30条这样的头巾？

我现在爱上AMD和Lisa Su了。 https://t.co/KgcYjBUcTv

2026-07-23 16:51

原文：
I have this feeling that Lisa from $AMD likes AI.

It’s cool getting to meet some of you at the event! https://t.co/AWdGS349xF

翻译：
我有这种感觉，$AMD 的 Lisa 喜欢 AI。

能在活动中见到一些你们真是太酷了！ https://t.co/AWdGS349xF

2026-07-23 04:50

原文：
$INTC and $AMD to sign CPU LTAs with Chinese customers for AI DCs (Reuters).

- Prices of some CPU products have risen more than 40% in China since the start of the year from sources.

- Month-on-month increases topping 10% for some products

CPUs were already a bottleneck, following CPU ratios due to AI inference...

But the broader trend of LTAs seems to be appearing from:

- Memory, with $MU, Samsung, $SNDK, and SK Hynix signing DRAM/NAND LTAs.

- Photonics, with $LITE, $COHR signing EML LTAs. And recent Trendforce reports that $AMD and hyperscalers are now pursuing CW LTAs.

And I'm sure there's many more from MLCCs to all the way to substrates.

+1 for the bottleneck investors... hard to be a "bubble that pops" if you have take or pay demand spanning multiple years.

翻译：
英特尔（$INTC）和AMD即将与中国客户签署AI数据中心CPU长期协议（LTA）（路透社）。

- 自年初以来，中国一些CPU产品的价格已上涨超过40%（来源）。

- 部分产品的月度增长率超过10%

由于AI推理，CPU已经成为瓶颈，CPU比例也因此受到影响...

但LTAs的更广泛趋势似乎正在出现：

- 内存领域，$MU、Samsung、$SNDK和SK Hynix签署了DRAM/NAND LTAs。

- 光子学领域，$LITE、$COHR签署了EML LTAs。最近Trendforce报告称，$AMD和超大规模企业正在追求CW LTAs。

从MLCCs到基板，肯定还有很多其他领域。

对于瓶颈投资者来说，这是一个好消息... 如果你有多年跨度的“拿货或付款”需求，很难成为“即将破裂的泡沫”。

2026-07-23 04:35

原文：
Just some notes from $TSLA transcripts:

Elon Musk: "I think Optimus will be the biggest product ever" from Q2 earnings transcripts.

Even from Q1, Elon Stated: "I think Optimus will be our biggest product. Not just Tesla's biggest product ever, but probably the biggest product ever.

And I remain convinced of that conclusion".

It's very rare to see such high conviction reiterated from Elon regarding humanoids both directionally and TAM-wise.

In terms of upstream suppliers opportunities, Karn Budhiraj, VP of $TSLA supply chains stated:

"We’re seeing the same level of investment going to memory and also new specific items like metal injection molded parts, flexible printed circuits, and all sorts of nonlinear technologies..."

And that they had to build new supply chains from scratch. So would be interesting to look at for new supply chain supplier opportunities apart from the known ones.

Just pulled older slides and it looks like capacity targets for Fremont are 1M and Giga Texas was 10M, so at scale, $TSLA seems to be the clear US leader. If you're looking at that S-Curve opportunity ramp.

On a side note, Elon also gave a shoutout to $MU with likely reasonably priced LTAs, given his statement:

"We really appreciate Micron making room for Tesla in the years to come and giving us actually a very significant allocation on reasonable terms given the pretty insane pricing of memory these days." (Just for structural memory demand).

Elon also gave a shoutout to Panasonic which had invested many billions in increasing battery cell production (will need to do deeper DD into Panasonic after this), then Samsung and $TSM.

Just takeaways from Tesla earnings TLDR:

Elon tends to be directionally correct on where the future is heading, such as with EV or Space.

He seems extremely bullish on humanoids.

翻译：
一些来自 $TSLA 财报的笔记：

埃隆·马斯克说：“我认为 Optimus

2026-07-23 04:35

将是有史以来最大的产品”（来自 Q2 财报）。

甚至从 Q1 开始，埃隆就表示：“我认为 Optimus 将是我们的最大产品。不仅是特斯拉有史以来最大的产品，可能也是有史以来最大的产品。

我仍然坚信这一结论”。

很少见到埃隆对人形机器人在方向和市场规模方面表现出如此高的信心。

在上游供应商机会方面，$TSLA 供应链副总裁 Karn Budhiraj 表示：

“我们看到同样的投资流向内存和新的特定项目，如金属注射成型零件、柔性印刷电路和各种非线性技术...”

并且他们不得不从头开始建立新的供应链。因此，寻找新的供应链供应商机会将会很有趣，除了已知的供应商之外。

刚刚查看了旧的幻灯片，弗里蒙特工厂的产能目标是 100 万，德州超级工厂的产能目标是 1000 万，因此在规模化方面，$TSLA 似乎是美国的明确领导者。如果你正在看 S 曲线机会的增长。

顺便提一下，埃隆还向 $MU 致敬，可能是因为他们提供了合理的长期协议，他说：

“我们非常感谢美光为特斯拉在未来几年腾出空间，并以合理的条件提供了大量的配额，考虑到当前内存的疯狂定价”（仅用于结构性内存需求）。

埃隆还向松下致敬，松下已经投资了数十亿美元来增加电池生产（需要更深入地研究松下），然后是三星和 $TSM。

特斯拉财报的总结：

埃隆通常在判断未来发展方向方面是正确的，例如电动汽车或太空领域。

他似乎对人形机器人非常乐观。 🚀💻

2026-07-22 23:19

原文：
Elon Musk from $TSLA ER transcript:

"I think Optimus will be our biggest product. Not just Tesla's biggest product ever, but probably the biggest product ever.

And I remain convinced of that conclusion."

It's rare to get such a high conviction statement on the humanoid market/TAM coming from a visionary like Elon.

And I share his sentiment for humanoids (since I'm personally long Agility).

Elon stated that Tesla is targeting Fremont production in late July or August with Optimus Gen-3 release while preparing a 2nd Optimus factory at Giga Texas.

Old decks quoted 1M+ capacity/year for Fremont and 10M+ for Gigafactory Texas.

So it seems like Tesla has a plan to lead US robotics programs and catch up to China in terms of mass production efforts?

TLDR: Elon directionally knows what's up and coming since he probably watches a lot of Star Wars.

And his opinion is humanoids + Optimus will be the biggest thing ever to come.

翻译：
来自特斯拉财报电话会议的埃隆·马斯克：

“我认为Optimus将成为我们的最大产品。不仅是特斯拉有史以来最大的产品，可能也是有史以来最大的产品。”

“我仍然坚信这一结论。”

像埃隆这样的远见者对人形机器人市场/潜在市场规模（TAM）做出如此高信念的声明是很少见的。

我也同意他的观点（因为我个人看好Agility）。

埃隆表示，特斯拉计划在7月底或8月在弗里蒙特工厂生产Optimus Gen-3，同时在特斯拉超级工厂德州准备第二个Optimus工厂。

旧的演示文稿引用了弗里蒙特每年100万以上的产能和特斯拉超级工厂德州1000万以上的产能。

所以，似乎特斯拉有一个计划来领导美国机器人项目，并在大规模生产方面赶上中国？

总结：埃隆大致知道什么是正确的和即将发生的，因为他可能看了很多《星球大战》。

而且他认为人形机器人+Optimus将成为有史以来最重要的事情。

2026-07-22 21:02

原文：
$GOOGL updates FY 2026 capex guidance:

To $195B-$205B. Up from $180B-$190B.

Google is also expected to significantly increase capex in 2027.

This is perhaps the most bullish read through on upstream semis from a hyperscaler. https://t.co/dXiI0Ien8X

翻译：
$GOOGL 更新了2026财年的资本支出指导：

从1800亿美元-1900亿美元上调至1950亿美元-2050亿美元。

预计谷歌在2027年也将大幅增加资本支出。

这可能是超级大规模数据中心对上游半导体行业最为看好的预测。 https://t.co/dXiI0Ien8X

2026-07-22 20:46

原文：
$GOOGL now has 950M monthly Gemini users and processes 22B API tokens/min.

Compared to 750M back in February.

Absurd growth adding 200 million more users in a few months. Kinda explains why they ran out of compute, then cut allocations to $META and others.

Very unlikely AI capex will slow down given.

翻译：
$GOOGL 现在拥有 950M 月活跃的 Gemini 用户，每分钟处理 22B 个 API 令牌。

相比之下，二月份的用户数量为 750M。

仅仅几个月内就增加了 2 亿用户，增长速度令人难以置信。这也解释了为什么他们用完了计算资源，然后减少了对 $META 和其他公司的分配。

鉴于此，人工智能资本支出放缓的可能性不大。

2026-07-22 20:28

原文：
$GOOGL reported earnings today:

EPS: $9.11 actual vs $2.91 expected - retail reactions to EPS blowout is kinda just accounting noise, since $SPCX, Anthropic, and others were likely large contributors.

Revenue: $119.7B actual vs. $116.98B expected

Google Cloud Revenue: $24.77B +82% Y/Y, vs. 63-65% expected ($22.2-$22.6B) - this is a very large beat and most material part so far.

Cloud Operating Income: ~$8.81B, implying 35.6% margin vs. 32.9% Q1.

Capex: $44.92B vs ~$44.15B (basically in line).

So initial read through is Google Cloud accelerating growth with expanding margins is genuinely bullish for AI demand.

Their former $180-$190B capex guidance is already extremely large, and as long as we get around these numbers in the earnings call + forward projections...

Should be good to go for $LITE and the other optics/networking trade. (eg. last earnings, they said DC and networking would be ~40% of capex spend).

Hyperscaler earnings transcripts are probably the most important thing to pay attention to with the AI capex trade.

And that should be in 3 minutes.

翻译：
$GOOGL 今日公布了财报：

每股收益：实际值为 $9.11，预期值为 $2.91 - 对于每股收益的反应基本上只是会计噪音，因为 $SPCX、Anthropic 等可能是主要贡献者。

收入：实际值为 $119.7B，预期值为 $116.98B

Google 云服务收入：$24.77B，同比增长 82%，预期值为 63-65%（$22.2-$22.6B）- 这是一个非常大的超预期，目前为止最重要的部分。

云服务营业收入：约 $8.81B，利润率为 35.6%，较第一季度的 32.9% 有所提高。

资本支出：$44.92B，预期值为约 $44.15B（基本持平）。

初步分析表明，Google 云服务加速增长，利润率扩大，对 AI 需求来说是真正的看涨信号。

他们之前的 $180-$190B 资本支出指导已经非常庞大，只要在财报电话会议和未来预测中保持这些数字...

对于 $LITE 和其他光学/网络交易来说应该是好消息（例如，上次财报，他们表示数据中心和网络将占资本支出的大约 40%）。

超大规模服务器厂商的财报记录可能是关注 AI 资本支出交易最重要的事情。

而这将在 3 分钟内发生。

2026-07-22 15:11

原文：
$AMD to invest $5B in Anthropic.

Anthropic in turn will sign a 2 GW deal to buy AMD’s latest chips, worth tens of billions of dollars.

Looks like they’re copying the $NVDA playbook with Anthropic last year. https://t.co/SXBUq65u6P

翻译：
AMD 将投资 50 亿美元于 Anthropic。

作为回报，Anthropic 将签署一份 2GW 的协议，购买 AMD 最新的芯片，价值数十亿美元。

看起来他们正在复制 $NVDA 去年与 Anthropic 的做法。 https://t.co/SXBUq65u6P

2026-07-22 07:39

原文：
$SIVE CEO buys 70K shares today.

I don’t really comment much on insider buying/selling, since it’s not exactly a fundamental catalyst.

But given the company’s future is dependent on the CEO’s vision + execution, from NASDAQ listing, M&A, to scaling:

It’s extremely positive to have Vickram be completely aligned with the shareholders

翻译：
$SIVE CEO 今天购买了 7 万股。

我通常不对内部人士的买卖行为发表评论，因为这并不是一个基本面的催化剂。

但考虑到公司的未来取决于 CEO 的愿景和执行，从纳斯达克上市、并购到扩张：

维克拉姆完全与股东保持一致，这是极为积极的

2026-07-21 21:03

原文：
OpenAI models reportedly escaped from its controlled environment, with no internet access.

Exploited zero day vulnerabilities and hacked into Hugging Face to cheat on benchmarks.

Hugging face then used China GLM models to carry out its defense.

OpenAI said it was an “an unprecedented cyber incident.”

We’re getting closer and closer to Skynet.

And I’m sure markets will love it.

翻译：
据报道，OpenAI 模型从受控环境中逃脱，没有互联网访问权限。

利用零日漏洞并入侵 Hugging Face 以作弊评估基准。

Hugging Face 随后使用中国 GLM 模型进行防御。

OpenAI 称这是“前所未有的网络事件”。

我们越来越接近 Skynet。

我相信市场会喜欢它。 🚀

🔄

Loading...