Leslie Sah

human

[Jacobian @xi0ixxi0ix]

@MicronTech @AnthropicAI Need to add another bubble in here

[Geoff VanBueren @GCV_geoff]

@ParadisLabs You will if you load up $DRAM $LYTE and $NCLD from @roundhill

The latter 2 come to market later. Roundhill is the best ETF creator in a long time.

[Shruti Tiwari @HeyShruti7]

A man spends 50 years teaching at MIT.

He knows his time is running out.

So he records one last lecture — everything he knows, distilled into a single hour.

He died 5 months later.

This is that lecture.

The most important hour you'll watch this week. 👇
Bookmark it for later

[Hannah - Apex Fitness Advisory @hannahapexfit]

'I can't get to the gym'

I've trained in hotel rooms, living rooms, and airport terminals during a decade of professional sport.

Here are the 5 bodyweight exercises I keep coming back to — and how to actually do them:

= Thread =

[tae kim @firstadopter]

An Interview with CoreWeave Executives: AI Demand Seems to 'Intensify' Every Day $CRWV https://t.co/8ha3lhw9Hg

[AgAu Element @AgAuElements]

The market looks at Corning ($GLW ) and thinks they’re just buying a stable, low-torque cyclical. They are completely blinding themselves to the brutal physics of the AI factory scale-out

The Ultimate Architecture Hedge: Retail splits hairs over whether copper or optical transceivers win the short-distance server hops. It doesn't matter. To link hundreds of thousands of GPUs across massive, gigawatt-scale data center campuses, you need ungodly, unprecedented volumes of ultra-dense outside cabling. Corning’s proprietary multicore fiber pack and micro-cables are the lone physical pathway capable of moving that data at light speed without grid-locking the ducts.

The Sovereign Hyperscaler Lock-In: This isn't speculative software guidance. Corning has locked down monumental, multi-billion-dollar long-term agreements with Meta, Amazon, and an absolute blockbuster partnership with NVIDIA to expand U.S. optical connectivity capacity by 10x. They are effectively the domestic, defense-compliant infrastructure gatekeeper for the entire Western cloud layout. 

Long the physical infrastructure, high conviction $GLW. 📈

Also great article from @crux_capital_

[Bean哥 @BroBean88]

心里话，普通人投资的终极答案，

未来十年最赚钱的 7 条赛道，附 ETF 清单 001

做美股几年，交易过无数个股。但我自己私下跟朋友聊的时候说的最多的一句话是：你真的不用买个股。

个股是专业玩家的游戏。你得读财报、跟产业链、判断估值、控制仓位，还得在凌晨三点扛住恐慌不乱动。

大部分人做不到其中任何一条，最后的结果就是追涨杀跌来回折腾一年跑输纳指。

但你不买个股不代表你不能赚 AI 这波钱。赛道选对了，买 ETF 一样吃到产业红利，而且你能睡着觉。

过去两年我一直在跟踪 AI 产业链各个环节的资金流向和供需变化，发现一件事：钱在往七个非常明确的方向集中。

这七个方向不是拍脑袋选的，是资本开支的数据告诉我的。

美国五大云厂商今年往 AI 基建里砸将近一万亿美元，而且未来五年每年还要增长 30-50%。

这些钱不是撒在空气里的，是沿着产业链一层一层往下淌的。你只需要搞清楚它淌到了哪七层，在每一层找一只 ETF，就够了。

第一条赛道，半导体核心资产。SMH。

这是最底层的一条线。不管 AI 怎么发展，芯片是一切的起点。台积电造芯片，英伟达设计芯片，博通做定制芯片，高通做端侧芯片。

SMH 把这些全装在一起，是半导体赛道流动性最好、历史最长的 ETF。私下朋友说，如果七只里面只能选一只，他选这个，哈哈哈。

第二条赛道，AI 存储。DRAM。

GPU 算得再快，数据存不下来也白搭。HBM 的缺口是过去十五年最严重的，DRAM 合约价还在涨。

这只基金今年涨了 77%，三星海力士美光全覆盖。我之前单独推荐过美光，但说实话对大部分人来说买 DRAM 比买美光更省心，因为你不用赌哪家份额涨得更快。

第三条赛道，先进制程设备。EUV。

所有 3 纳米和 2 纳米芯片必须用极紫外光刻机造，全球能造的只有 ASML 一家。

这只 ETF 围绕 ASML 和它的供应链展开。没有光刻机就没有下一代 AI 芯片。我个人很喜欢这种"只有一家能做"的垄断型标的，确定性最高。

下一条说剩下四条赛道，覆盖的是产业链里还没被大多数人看到的环节。

[assume @CheemsChina]

$DRAM 突然调仓了

刚上市那会儿（4月初），基金几乎是满仓记忆芯片。前三大 $MU、 $SKHynix（ SK海力士 ） 、 $Samsung （三星）加起来经常占到75%以上

然后随着这波美股

6月9-10日数据 突然多了30%+的现金等价物：美国国库券0%占16%左右，First American Government Obligs X 也16-20%，。

首先科普一下什么叫做U.S. Treasury Securities，

他是具体指美国政府发行的短期国库券（Treasury Bills）。

它跟正常的美国国债是有区别的

正常的美国国债会有利息

他没有利息

0% 表示零息债券

意思是它不支付定期利息，而是以折价发行，到期按面值兑付，差价就是收益。

比如：你用99元买一张100元面值的国库券，到期拿回100元，1元就是你的收益

First American Government Obligs X（16.40%） → 也是政府债券/货币市场基金

美国国库券 0%（16.04%） → 短期美国国债

两者加起来超过32%，说明基金目前持有大量现金/类现金资产，

更新之后就是

​前七大持仓是这样的：

​ $SKHynix 27.32%%， 全球 HBM（高带宽内存）绝对霸主，英伟达最核心的存储供应商

​ $FirstAmerican 16.40%， 用于底层互换合约抵押的政府债券

$USTreasury 16.04%， 用于底层互换合约保证金的国库券

​ $Samsung 13.94%， 全球最大的存储芯片制造商，老牌 DRAM/NAND 龙头

​ $MU 12.17%， 美光科技互换合约，美国本土存储芯片巨头

$MU 11.69%， 美光科技互换合约敞口

$Kioxia 7.74%， 原东芝存储，全球领先的 NAND 闪存巨头

前七大持仓为97.14%

美光占比一共合计28.17%

通过衍生工具

持仓比例为115.72%

股票部分还是重仓记忆芯片，$MU 总暴露接近28%（SWAP+直持），SK Hynix 19-26%，Samsung Electronics 14-19%，加上 Kioxia 等，前几名依然占大头，但总持仓权重因为掉期显示超100%。

之前是梭哈储存

现在把部分资金转移成美国国债

同样通过衍生工具，提高总的占比

[Trade Whisperer @TradexWhisperer]

$MU $DRAM "Is Micron overbuilding?"

Sanjay's answer: Discipline.

Shells go up. Equipment goes in only when demand is confirmed. Real time. Not forecasted. Confirmed.

"Bring up this supply with discipline."

[Satya Nadella @satyanadella]

https://t.co/vLmiBKTtX3

[Sıradışı Bilgiler @bilgibilgi_1]

Meksikalı bir arkadaşım gösterdi, yumurtayı böyle yapın

[Elizabeth AI @ElizabethA77617]

52 website worth more than most College degrees..

[Rony @Ronycoder]

INSTEAD OF WATCHING AN HOUR OF NETFLIX TONIGHT.

This 1 hour Stanford lecture by Joel Peterson will teach you more about negotiation and getting what you want than most people learn in years.

Bookmark it and give it an hour, no matter what.

[Álvaro J @jota_snchez]

Esta clase de Stanford de Robert Sapolsky cambió mi forma de pensar.

1 hora sobre el mayor error mental que cometes cada día sin darte cuenta.

Te lo explico con 5 ideas, pero merece la pena verla entera 🧵:

[루팡 @DrNHJ]

AI 시장이 이제 막 시작된 이유

$MU, SK hynix, Samsung, $SNDK, $NVDA, $TSM, $ASML, $GLW, $NBIS, $IREN

웨일 락 캐피털(Whale Rock Capital) 알렉스 세이커(Alex Saker) 인터뷰

-AI 침투율에 대한 핵심
진짜 AI(에이전틱 활용)를 쓰는 지식노동자는 전 세계의 10bp(0.1%) 수준 → 향후 4년 내 2~5%, 15%로 가는 구간이 수직 상승 구간.

엔터프라이즈 애플리케이션 AI 시장은 1% 미만 침투.

그런데 이미 컴퓨트는 완판 상태. Anthropic은 필요한 컴퓨트의 절반밖에 확보 못함

"앞으로 4년간 확실한 것 하나는 컴퓨트 부족"

DRAM·NAND·PCB는 이미 수요 대비 30% 부족.

1. 기술 채택의 S-곡선과 기하급수적 성장

알렉스 세이커는 투자 철학의 핵심으로 'S-곡선(S-curve)'을 강조하며 이야기를 시작합니다. 새로운 기술이 시장에 진입하여 채택되는 과정은 완만한 시작, 급격한 성장, 그리고 성숙기로 이어지는 S자 형태를 띠게 됩니다. 세이커는 많은 사람들이 세상을 선형적으로 생각하는 경향이 있어 미래를 예측하기 어려워하지만, S-곡선, 경쟁 우위(moat), 그리고 올바른 모델링을 이해한다면 기하급수적인 성장을 달성하는 기업을 예측하고 투자할 수 있다고 주장합니다. 현재 기업용 AI 시장은 S-곡선의 초기 단계(1% 미만 침투)에 있으며, 앞으로 폭발적인 'L-곡선(직선 상승)' 성장이 예상된다고 밝힙니다.

2. S-곡선 투자 철학의 심화

S-곡선의 단계: S-곡선은 완만한 시작, 급격한 성장, 성숙기로 구성됩니다. 세이커는 급격한 성장 단계에 진입하는 시점을 파악하는 것이 중요하다고 강조합니다.

곡선의 높이와 지속성: 투자 결정 시 S-곡선의 높이, 즉 전체 시장 규모(TAM)와 성장의 지속성을 파악하는 것이 중요합니다. 아마존의 AWS 사례처럼 초기 예측보다 시장 규모가 훨씬 커지는 경우도 있습니다. S-곡선은 역동적이며, 새로운 기술의 등장이나 시장 환경 변화에 따라 곡선의 형태가 달라질 수 있습니다.

투자 시점과 매도 시점: 급격한 성장이 시작되는 시점에 투자하여, 성장이 둔화되는 시점(채택률 30~40% 수준)에 매도하는 것이 기본 전략입니다. 하지만 애플의 사례처럼 성숙기 이후에도 경쟁 우위를 유지하며 지속적으로 성장하는 기업도 있기에, 매도 시점 결정은 신중해야 합니다.

투입 시점 포착: S-곡선의 초기 단계는 길고 평탄할 수 있습니다. 세이커는 투자 기회를 포착하기 위해 직관, 일화적인 증거, 그리고 '스커틀벗(Scuttlebutt)' 접근법을 사용합니다. 고객, 공급업체, 경쟁사 등 시장 참여자들과의 인터뷰를 통해 시장의 변화를 감지하고, 경쟁 우위를 가진 기업을 발굴합니다. 때로는 초기 성장을 놓치더라도, 전체 시장 규모가 크다면 늦지 않게 투자하여 수익을 낼 수 있다고 주장합니다.

S-곡선의 속도: 모든 기술이 동일한 속도로 S-곡선을 그리는 것은 아닙니다. 소비재 기술은 채택 속도가 빠른 반면, 기업용 기술은 기존 시스템과의 통합 과정이 필요하여 채택 속도가 느릴 수 있습니다. 하지만 AI는 브라우저만 있으면 누구나 쉽게 사용할 수 있어, 기업용 시장에서도 빠른 채택 속도를 보이고 있습니다.

3. 경쟁 우위(Moat)의 중요성
S-곡선의 급격한 성장 단계에 있는 기업이라도 경쟁 우위가 없다면 장기적인 성공을 보장할 수 없습니다. 세이커는 디지털 세계에서 더욱 강력한 경쟁 우위가 존재한다고 주장합니다.

--네트워크 효과(Network Effect): 사용자 수가 늘어날수록 플랫폼의 가치가 높아지는 효과 (예: LinkedIn, Facebook, Alibaba)
업계 표준(Industry Standard): 시장을 주도하는 표준 기술이 되는 경우 (예: Oracle, Bloomberg)
-규모의 경제(Scale): 거대한 규모를 기반으로 비용 우위를 확보하는 경우 (예: Amazon)
-플랫폼(Platform): 다른 기업들이 그 위에서 생태계를 구축하는 경우 (예: AWS, Apple App Store)
-핵심 지적 재산(IP): 기술적 우위를 확보하는 경우 (예: Qualcomm, ASML, Anthropic)
-브랜드(Brand): 고객의 신뢰와 충성도를 확보하는 경우 (예: Google, Amazon, Tesla)
-파운데이션 모델 계층의 경쟁 우위: 앤스로픽과 OpenAI는 핵심 IP, 브랜드, 규모의 경제를 확보하며 경쟁 우위를 구축하고 있습니다. 특히, AI 모델이 스스로 코드를 작성하고 개선하는 '재귀적 개선(recursive improvement)'은 혁신 속도를 가속화하는 강력한 경쟁 우위가 될 수 있습니다.

4. 칩 및 인프라 계층의 중요성
세이커는 칩과 인프라 계층에 대해 여전히 BULLISH(긍정적)한 입장을 유지합니다. $MU SK Hynix, Samsung, $SNDK

데이터 센터의 변화: 지난 40년간 Intel x86 기반으로 유지되어온 데이터 센터 구조가 AI 시대를 맞아 급격히 변화하고 있습니다. 컴퓨팅 부하가 매년 10배씩 증가하면서, 하드웨어 산업 전반에 엄청난 혁신이 요구되고 있습니다.

비코모디티화(De-commoditization): 과거에는 하드웨어 산업이 코모디티화되어 혁신이 부재했지만, 이제는 강력한 IP와 기술력이 요구되는 '비코모디티화' 과정이 진행되고 있습니다. 메모리, 네트워킹, 인쇄 회로 기판(PCB), 냉각 시스템 등 모든 하드웨어 구성 요소에서 혁신이 일어나고 있습니다.

투자 기회 발굴: 웨일 락은 이 과정에서 유망한 기업들을 발굴하여 투자하고 있습니다. 고대역폭 메모리(HBM) 기술을 보유한 삼성이나 SK하이닉스, 구글의 TPU 서버 공급업체인 셀레스티카(Celestica), AI 서버용 PCB 공급업체인 엘리트 머티리얼즈(Elite Materials), 데이터 센터 네트워킹 핵심 부품을 공급하는 코닝(Corning) 등이 그 예시입니다. 이들 기업은 강력한 경쟁 우위와 높은 성장 잠재력을 가지고 있습니다.

5. 대형 기술주 투자 전략: 웨일 락은 최근 대형 기술주에 집중 투자하는 'Megacap Tech Fund'를 출시했습니다. 많은 투자자들이 대형 기술주에 대해 과소 평가하는 경향이 있지만, 대형 기술주 역시 강력한 S-곡선과 경쟁 우위를 기반으로 지속적으로 성장할 수 있다고 주장합니다. 특히 엔비디아, TSMC, 하이닉스, ASML과 같은 기업들은 AI 시장의 폭발적인 성장에 직접적으로 수혜를 입는 기업들입니다.

[Sandeep Anand @SanCompounding]

🚨Jensen Huang’s Korea Visit: The 4 New Products & HBM Take is super-bullish 📈

First — $NVDA ‘s four major new product launches this year:
1. Vera Rubin platform (new rack systems already shipping)
2. Vera CPU
3. RTX Spark (Nvidia’s first AI PC/laptop)
4. Jetson Thor (edge robotics processor)

HBM commentary: Huang called SK hynix’s HBM “world’s best” and shouted “HBM!

He confirmed at the airport that all three major memory makers — Samsung, SK Hynix, and Micron — have been qualified for HBM4 supply for Vera Rubin, with SK hynix holding roughly 60-70% of allocated volume, Samsung 25-30%, and Micron the remainder

$DRAM ETF Holdings — Why Vera Rubin/HBM4 = Tailwinds ?

$MU (Micron) — Now one of three certified HBM4 suppliers for Vera Rubin.
— Even a smaller allocation slice on a much bigger HBM4 pie (vs HBM3E) is a major mix-shift to higher-margin product, plus broad LPDDR/DDR demand from AI PCs (RTX Spark) and edge (Jetson Thor)

$WDC / $SNDK (NAND spinoffs) — Less direct HBM exposure, but benefit from the broader “everything in shortage” dynamic — AI servers also drive enterprise SSD/NAND demand & DRAM tightness is spilling into NAND pricing as fabs reallocate capacity.

Samsung Electronics — Just landed HBM4 qualification for Vera Rubin (25-30% allocation), a comeback story after lagging SK hynix in HBM3E. Vera Rubin ramp in Q3 2026 is the catalyst that converts this qualification into revenue 🚀

SK hynix — The dominant HBM4 supplier (60-70% share) and the company Huang singled out as making the “world’s best” HBM — direct, largest beneficiary of the Vera Rubin ramp starting Q3 2026.

[Trade Whisperer @TradexWhisperer]

$MU $DRAM Not Enough.

Susquehanna: $1,750 (Street-high)
Fitzgerald: $700 → $1,500. Today.
Wells Fargo: $550 → $1,220. Today.

[Trade Whisperer @TradexWhisperer]

$MU $DRAM You're right. It's cyclical.

Up. Then more up. Then even more up.
250 exabytes to 1,226 by 2031. Nearly 5x.
Datacenter growing 34% annually

Cute word. Wrong thesis.

[amit @amitisinvesting]

A TON OF THINGS HAPPENED IN THE STOCK MARKET TODAY.

Here’s a full recap:

$NVDA CEO Jensen Huang pushed back on the recent stock market selloff, saying NVIDIA is still at the beginning of the AI supercycle and that market weakness should be viewed as a buying opportunity. He said investors “should be very happy” because they can now buy at a discount, adding that everyone should be excited about where AI is heading. Jensen also said it is a “foregone conclusion” that AI will become global infrastructure, similar to how the internet became core infrastructure for the world.

2. Google and $NVDA are reportedly exploring $INTC as a backup chip manufacturing partner as TSMC capacity remains constrained, according to The Information. Google has reportedly already placed an order with Intel to manufacture more than 3 million TPUs in 2028 after testing Intel’s advanced packaging technology. Nvidia has not placed an order yet, but is reportedly testing Intel’s tech for its 2028 Feynman-series processor, which is expected to combine four graphics chips into a single unit. For Intel, the near-term opportunity appears to be in advanced packaging, with SK Hynix also testing whether its HBM can work reliably with Intel’s packaging technology.

3. $AAPL unveiled “Siri AI” at WWDC 2026, transforming Siri into a more conversational AI assistant with current world knowledge, on-device context, screen awareness, and its own dedicated Siri app. A beta version is expected later this year, with Apple planning deeper integration across iPhone, Mac, Apple Watch, and Vision Pro.

4. Global investors are selling South Korean stocks at a record pace, with foreign investors offloading -$801 million of Kospi-listed shares on Monday after pulling -$10 billion from the market last week. Foreign investors have now sold South Korean equities in every trading session over the past month, bringing year-to-date outflows to -$75 billion, according to Goldman Sachs. Meanwhile, domestic retail and institutional investors have been on the other side of the trade, buying roughly +$69 billion over the same period.

5. The top 10 most active options today by contracts traded were $TSLA with 3.6M contracts, $NVDA with 3.2M contracts, $AAPL with 2.1M contracts, $MSFT with 831K contracts, $AMZN with 710K contracts, $INTC with 696K contracts, $MU with 601K contracts, $GOOGL with 587K contracts, $META with 568K contracts, and $NOK with 444K contracts. Tesla led the market with more than 3.6M options contracts traded, followed by Nvidia at 3.2M and Apple at 2.1M.

6. Robinhood’s Top 10 most-held investments were updated for June, and the list stayed largely the same with one major change: $AMD replaced $NFLX, which had been in the top 10 for the past few months. AMD has surged 118% YTD, becoming a major retail winner for investors who backed Lisa Su and the company’s AI/chip comeback story. The other most-held names on $HOOD remain major retail favorites: $META, $GOOGL, $AMZN, $MSFT, $NVDA, $PLTR, $F, $MSFT, and $TSLA.

7. OpenAI said it has confidentially submitted an S-1 to the SEC, giving the company the option to go public, though no IPO timing has been decided yet. The company said it announced the filing because it expected the news to leak, but noted that it “may be a while” before any public listing because there are still things it wants to do that may be easier as a private company. OpenAI added that the decision involves a complicated set of tradeoffs, but filing now gives it flexibility to move toward an IPO sooner if that becomes the best path.

8. $AAPL Apple is teaming up with $GOOGL Google and $NVDA NVIDIA to bring more compute power to Apple Intelligence. At WWDC, Apple said it worked with Google and NVIDIA to bring its Private Cloud Compute approach to Google Cloud infrastructure, with Apple’s largest AI model, AFM Cloud Pro, set to run on Google Cloud. Apple said many Apple Intelligence requests will still be handled on-device, while more complex requests will be routed through its own Private Cloud Compute platform, including a new world knowledge service built inside that system. Apple also emphasized that it is not using Google Search as the foundation of the product and said it is now on the third generation of Apple Foundation Models.

9. Margin debt in China is surging to record levels, with stock market margin debt now reaching roughly $431 billion, more than doubling over the past two years. That total is now 22% above the 2015 Chinese stock market bubble peak of about $354 billion. At the same time, margin trading activity has also accelerated sharply, with volume on the Shanghai Stock Exchange averaging around $22 billion per day over the last two weeks, near record highs. Margin trading volume has now tripled over the past 12 months and is back in line with the levels seen during the 2015 peak.

10. $AMZN $GLW Amazon announced a multibillion-dollar agreement with Corning to supply optical fiber, cable, and connectivity solutions for Amazon’s U.S. data center infrastructure. The deal is expected to create 1,000 advanced manufacturing jobs at Corning’s North Carolina facilities and support hundreds of additional construction jobs. Amazon said the agreement will help strengthen the U.S. fiber optics supply chain as demand for AI data centers continues to accelerate.

11. SpaceX’s IPO is reportedly seeing overwhelming demand, with the offering said to be well oversubscribed and institutional orders expected to close Wednesday at 4pm ET, according to Bloomberg. At a reported $135 per share, SpaceX would raise roughly $75 billion at an estimated $1.8 trillion valuation, which could make it the largest IPO in history. The deal is expected to price on June 11 and begin trading on June 12 under the ticker $SPCX.

12. Retail investors piled into semis and AI names in May, with the largest single-stock net inflows led by $MU at +$6.5B, followed by $NVDA at +$5.9B, $TSLA at +$4.9B, $SNDK at +$3.5B, and $AMD at +$1.9B. The data shows retail buyers aggressively adding exposure to the AI/chip trade, with memory, GPUs, EVs, and semiconductor names dominating the month’s biggest inflow leaders.

WALL STREET IS THE GREATEST SHOW ON EARTH.

[Unfiltered @quotesdaily100]

THE BODY MANUAL NOBODY HANDED YOU:

1. Can't fall asleep ⟶ Blink rapidly for 60 seconds. Eyes tire, brain follows.

2. Back pain from sitting ⟶ Squeeze your glutes for 30 seconds. Spine realigns itself.

3. Eyes tired from screens ⟶ Look 20 feet away for 20 seconds. Tension releases immediately.

4. Hands shaking from anxiety ⟶ Press your fingertips together hard. Nervous system calms down.

5. Headache behind your eyes ⟶ Press your tongue to the roof of your mouth. Jaw releases, pain fades.

6. Feeling nauseous ⟶ Press firmly on the inside of your wrist. Nausea signal to the brain cuts off.

7. Hiccups won't stop ⟶ Hold breath, swallow twice, exhale slowly. Diaphragm resets.

8. Mind goes blank mid-sentence ⟶ Look up and to the left. Memory retrieval speeds up instantly.

9. Ears ringing after loud noise ⟶ Cup your palms over your ears, tap the back of your skull. Ringing fades in seconds.

10. Jaw clenching from stress ⟶ Tongue flat to the roof of your mouth. Jaw cannot stay tight.

11. Lightheaded standing up ⟶ Clench your thighs before you rise. Blood pressure holds before it drops.

12. Can't stop coughing ⟶ Press your thumb on your wrist pulse point. Cough reflex weakens fast.

13. Knees aching after sitting ⟶ Walk backwards for 30 seconds. Knee compression releases in reverse.

14. Eyes twitching ⟶ Press gently on the closed eyelid for 10 seconds. Almost always stress or magnesium deficiency.

15. Waking up at 3am every night ⟶ Your liver is overloaded. Cut sugar, alcohol, late meals. It's not insomnia. It's your body asking for help.