creatiff 🔎

[creatiff 🔎 @_creatiff_]

@blake_wilson @bot I'm stuck too, its very annoying

[creatiff 🔎 @_creatiff_]

@Merridew__ What if OAI wants , but it can’t?

[Global Bagholding Opportunities IV L.P. @justbrosef]

Semis.

[LEI @TheMarketMemo]

https://t.co/QEiEPItyKc

[Jason Luongo @JasonL_Capital]

BREAKING: AI can now analyze options trades like a $500/hr options strategist (for free)

Here are 10 Claude prompts I use to sell puts, buy LEAPs, and run the wheel without second-guessing every trade

(Save this for later)

[Origo @origoinvest]

What is a Super Company according to Chris Hohn?

Not only has $SAF.PA delivered a 20% IRR to TCI over 13 years (!)

They recently INCREASED their stake by 22%

So, what makes Safran such a great compounder?

And why is one of the GOATS now doubling down?

Let's dive in ⬇️

(HINT: cashflows are now de-risked for another >13 years)

[Dr_Gingerballs @Dr_Gingerballs]

This is 100% what an alien hiding on earth would say in November.

[₿ΞΞnThereDoneThat Capital 賢い @BeenThereCap]

My chart of the day.
Wow. Sure, I knew Korean retail investors had become *a* factor in pumping the meme stocks and no-revenue wishco stocks. But not *this big* of a factor. Korean trading volumes in U.S. markets have exploded.
1/3

[Rihard Jarc @RihardJarc]

A really insightful interview with a Former $GOOGL Cloud employee on TPUs:

1. In the short term, he expects that TPUs will continue to be used by $GOOGL's own internal needs (Gemini, Search, etc.), but sooner or later, he expects $GOOGL to start selling them externally.

2. He thinks the strategy will be similar to $NVDA, where $GOOGL would apply TPUs to other hosters (not hyperscalers). In his view, out of all the options on the market today, TPUs are the closest alternative to $NVDA GPUs.

3. TPUs and GPUs are very different; they are a substitute for around 20% of the workloads today, and over time, he expects that to increase. $GOOGL is trying to make it as generic as possible.

4. TPUs are specialized for ML and AI workloads, which require particular tensor operations. In the right application, they can deliver significantly better performance per dollar compared to GPUs, requiring much less energy and producing less heat. They are also more energy efficient and have a smaller environmental footprint. For a specific application, they can offer as much as 1.4x better performance per dollar vs a GPU.

5. He mentions that some clients already use TPUs on $GOOGL Cloud, but those are mostly super sophisticated and super large. For an average customer to use them, $GOOGL will have to do a lot of work so that a third party can host it.

6. He thinks $GOOGL is at least 2 years away, as it is not just about the developers but also building a whole ecosystem around it. He believes that in the meantime, they will continue to onboard more customers.

7. He is not worried about the argument that $NVDA is getting better each year, as he believes that the TPUs are getting better at a faster rate than GPUs each year: »The amount of performance per dollar that a TPU can generate from a new generation versus the old generation is a much significant jump than $NVDA «. Currently, the $NVDA vs TPU gap is quite large, but he is convinced that the gap will be significantly smaller in two years.

8. There is a misconception that TPUs are only good for very specific text use cases designed for optimizing $GOOGL Search. It was true in the past, but not anymore, according to him. They are multimodal now, not as strong in video and audio, but otherwise, they're very good. TPUs are also not used only for training, but are effective at both training and inference.

9. He also mentions that the ambition at $GOOGL with TPUs always existed, but now the management team is pushing super hard.

found on @AlphaSenseInc

[fin @fi56622380]

今天Amazon裁员3万人，比2022年行业收缩期裁的还厉害，原因很简单

买GPU的钱capex不够，导致AWS增速下降，被市场严厉的惩罚了，只能砍工资来省钱买GPU，让财报好看一点，营造一个AWS增速触底的故事

每一个互联网公司的SDE打工人，都应该买入Nvidia/amd作为风险对冲，弥补自己被GPU挤出价值链的风险

-------------------------
进入2024~2025年，美国SDE们就业差的主要因素已经不再是21年的大幅过度扩张，也不再是薪资更低的海外技术中心抢饭碗，暂时还不是AI提效造成需求降低，而是来了一个新的大boss：GPU capex

GPU capex在互联网公司正在创造一种诡异的繁荣性萧条：

公司业绩增速很好看，股价也节节高升，但是员工的工资开支成了管理层腾挪不开的问题，每个人都要担心自己的工作，同时因为持续性裁人，留守的员工的工作压力越来越大，人心惶惶，和大萧条时代无异

这不是传统的行业萧条，而是资本在人力与算力之间的激进再分配

Amazon这次裁员3万人，大概从两个月前就有风声了，年中perf review往年都是7月，这次到8月中下才开始，RTO的政策，也是裁员的重要考虑因素(借口)。AGI组肯定是一个都不动，PXT, device and service, 运营，都会是重灾区，按惯例AWS应该会晚一点再裁，等到AWS reinvent之后再裁，人尽其用

而在故事的另一端，AWS Q2的未完成订单（backlog）达到1950亿美元，同比增长25%。 这说明客户想买但AWS交付不了，需求持续火爆，GPU买的不够快不够多

在AI服务器供应跟不上需求爆炸性增长的时代，opex(工资)换capex就能带来公司的业绩提升，资本会毫不留情的惩罚一切不把这条路线贯彻到底的CSP/Hyperscaler

Meta开启每半年5%悄悄裁员模式基础之上，刚刚裁掉AI org的600人，而且还悄悄裁掉了很多部门的director，也是同一个逻辑：AI data center不够用，一年之内对未来18个月的capacity计划上修了三次，每次以为高估了需求，都会在几个月之后才痛苦的发现原来是低估了

互联网公司内部真的不需要人了吗？当然不是，招人的budget砍掉之后，只能倒逼公司内部增效来弥补

目前最激进的互联网大公司各种手段都用上了，内部tool耗费大量人力做各种agent功能，一键部署鼓励vibe，定了KPI每个部门AI使用率要达标，各级汇报AI使用进度和案例，还规定互相定期交流学习

一顿操作猛如虎，目标仅仅只是希望把大家的工作效率提升~20%，而已

接下来可能会发生什么？

为了增速，为了竞争，当公司提效遇到瓶颈，裁员也裁到只剩大动脉了，opex已经榨不出油水了，公司下一步只会继续牺牲现金流，甚至像ORCL一样，开始为了增速而不惜冒风险举债

Nvidia/AMD也会继续用手里的巨额现金推动投资，支持合作伙伴投入AI capex，就像openAI做的那样

这一切最大的受益者都会是半导体全产业链公司，未来可能会出现一个新常态：半导体公司的利润率会反超互联网公司

但他们也承担了最大的风险：当VC/Hyperscaler的前期投资看到token需求减缓时，甚至只是需求增长速度减缓(不再是一年数倍)，一定也会毫不留情的砍单，这个传导的速度会非常快，会比半导体公司的投产周期要快的多

这一步什么时候会出现，一个参考指标是，企业渗透率接近50%的时候。

2000年3月互联网泡沫破灭时候，美国互联网渗透率大概在52%左右(另有一组数据时43%)

目前互联网大厂的genAI daily user渗透率正在从50%向90%迈进，而广大行业公司的AI采用率不到10%，增速暂时是有保障的，每一代科技革命增速最快的时候就是全社会公司采用率10%迈向50%的阶段

Cisco泡沫的历史不会简单重现，我相信这一轮信息比上一轮要充足太多太多，一直会有足够多的泡沫怀疑论者警醒，至少会让泡沫破灭的时候不会那么惨烈

[Amanda Goodall @Amanda_Goodall]

Why did ALL Mag 7 slow US hiring at the EXACT same time?

(Hint: It's not the economy)

The hiring drop is synchronized across Apple, Google, Meta, Microsoft, Amazon, Tesla, AND NVIDIA.

All slammed the brakes in late-2022/early-2023... and never let go.

We’re not seeing a reversion to old patterns.

CEOs AI-wash layoffs to pump stock - real reason? $420B capex redirected to offshore GCCs (70% savings).

Mag7 bombshells:
- Microsoft: 15K US cuts → India (especially Hyderabad/Bengaluru GCCs) was explicitly spared layoffs and saw massive expansion via $3B+ cloud investments, data centers, and hiring in strategic areas.
- Google: Thousands laid off in 2025 (HR/Cloud/Android teams) while Pichai pours $15B into Visakhapatnam hub - briefed PM Modi Oct 14
- Amazon: 14K managers gone → GCCs (Chennai, Hyderabad) are expanding with AWS/offshore boom
- Meta: 20K+ cuts → India hiring up (Blind: "massive outsourcing")

US tech postings -36% below pre-COVID.
India? +47%.

We all feel the pressure in the labor market for US workers right now.

Bookmark if you are pissed at the gaslighting. Let's start watching these levels every single month.

[Ozzy @ozzy_livin]

#Bitcoin time to start selling the rips ✍️

will start buying the dips around October 2026

[Andrew Hiesinger @AndrewHiesinger]

There is a service charging $700/mo to view dealer positioning models (GEX/VEX).

We believe these tools should be accessible, not limited to a select few.

So we built our own and made it accessible.

🧵 (1/4)

[🐧 @Pentosh1]

Before getting into this, would like to mention this is for INVESTING, for the most part. Not TRADING. I think, most people won't be interested. Again, as I previously said. I am learning and being taken a little under the wings of others. This is just sort of a journal currently if you will.

This portion of my portfolio is essentially my "save game" point and more passive. Every few months, I reset my crypto trading account and have moved that over to tradfi. I've historically viewed it as a barbell approach where crypto = make money and tradfi = preserve it. So this specific bucket is a low risk bucket that protects the majority of my wealth. It also, in doing so smooths out my overall returns as a whole.

Most people think the goal is to make the highest return. But really, what we all should strive for is something that is sustainable, and can keep compounding without blowing up.

So this portfolio design is suppose to survive every environment as opposed to being a bet on any one outcome. Keep in mind, this is it's own bucket, there are a few buckets. I'll cover all of them.

There is the foundation, or the anchor. Which is bonds, credit funds, box spreads. These can be rotated into stocks on a huge correction or at least my plan but don't have to be.

The purpose of them is to have steady, predictable returns. They certainly won't make you wealthy, but they can help keep your wealth regardless of that level.

Orthogonal Bets:
It just means, uncorrelated or things that move differently from stocks or bonds. These are strategies like systematic quant funds, or multi-strategy platforms.

They smooth the overall return curve and keep the portfolio more stable.

Then you have your long /short strategies. This is where the highest returns are possible. This is more skill based, but think of it as long value and short junk. It doesn't necessarily rely on the market going up. The same trades many put on here in crypto.

Then you have convexity sleeves. They are the smaller positions, but can be explosive. Macro, momentum, or venture style bets. They don't cost as much but they can move a lot. Discovery, is one one example. A macro convex fund. When something breaks, these can shoot up and offset the pain elsewhere. And sometimes, but not always can create some pretty huge wins.

Anchor = Stability
Orthogonal = Diverisification
Alpha Engine = Growth
Convex = optionality

Big capital in low volatility sleeves
Small capital in high vol sleeves (convexity)

And this helps shape the return path into something smoother that can compound, and be resilient despite market conditions.

In short, you survive the crashes.

I currently in this bucket for this portfolio have the following:

40% in anchor atm. Bonds, Alt credit funds like helix, delphi, and legacy, and box spreads. These are just examples.

45% in the alpha engine. Fundamental and systematic long/short equities. Some of this in similar liquid funds. Some is self managed like on my brokerage accounts.

10% in orthogonal: Uncorrelated asset streams

5% in convex/opportunistic. This is more asssyemtric bets, smaller capital but more optionality.

It's balanced between offense and defense. The result is high sharpe, moderate returns, and low drawdown risks. If the markets blow up as they do time to time, I may only see 5-10% drawdown.

Everything is counter balanced, the beta ends up being around .20-.25, with low volatility. Drawdowns are shallower, rebalacing is more effective. So one part is maybe losing money, another is flat, and then a different part of the port is making a killing when turmoil hits. But the part that is making a killing is the smallest part of the port bc its the convex sleeve. So you end up with a smooth equity curve overall

I will add more onto this later. Bc it's already a long post.

[Kris Sidial🇺🇸 @Ksidiii]

And although I truly believe we will continue to see these cyclical waves of volatility, a part of me is starting to think that the real tail event might be one where volatility stays elevated, similar to 2008, but less extreme. Imagine a VIX that holds between 40 and 60 in recurring waves instead of collapsing back down quickly.

Whenever we speak to potential investors, it is very clear that many market participants are conditioned to want long volatility programs that monetize gains almost immediately. Everyone is always focused on fighting the last war and optimizing around it, as usual.

It has become a running joke to “always short the VIX whenever it goes up.” But I think a prolonged period of elevated volatility would break a lot of models and hedging frameworks that rely on quick normalization. That path, where volatility remains sticky and refuses to die down feels increasingly possible, and I believe the market is underestimating that risk when they think about how vol can react whenever equities do eventually take a substantial leg lower.

[Campbell @abcampbell]

https://t.co/gEM1YugpNk

[Invest In Assets 📈 @InvestInAssets]

When to sell, by Peter Lynch:

[Julian Komar @BlogJulianKomar]

Why Position Sizing Decides Your Future 👇

[TraderLion @TraderLion]

William O'Neil analyzed 7,500+ stocks over 25 years.

These are the biggest factors that determine if a cup with handle will succeed or fail.

[Citrini @citrini]

Rick Rescorla was the director of security for Morgan Stanley. His preparation and relentless drilling of evacuation techniques (due to his strong suspicion that the WTC was a target for terrorism since 1988) resulted in nearly all of MS’ employees being successfully evacuated.

When AA11 struck WTC1, the port authority made announcements urging people to stay at their desks. Rick ignored the announcement, got on his bullhorn and began the process of evacuating MS’ 2700 employees. In a call to his best friend Dan Hill, he relayed,

“The dumb sons of bitches told me not to evacuate. They said it's just Building One. I told them I'm getting my people the fuck out of here."

During his tour of duty in Vietnam, Rescorla had boosted morale among his men by singing Cornish songs from his youth. He did the same in the stairwells of the world trade center, directing people down stairwells and away from elevators while singing:

“Men of Cornwall stand ye steady;
It cannot be ever said ye
for the battle were not ready;
Stand and never yield!”

Between songs, after successfully evacuating almost all of the 2700 employees, he called his wife Susan and told her,

“Stop crying. I have to get these people out safely. If something should happen to me, I want you to know I've never been happier. You made my life.”

Susan replied, "You made my life, too”.

Despite successfully evacuating everyone, Rescorla then went back into the building to continue getting everyone out. When the message came through on his walkie-talkie that he, too, had to evacuate he said, simply, “As soon as I make sure everyone else is out.”

Rick was last seen heading up the stairs of the South Tower on the 10th floor, shortly before the collapse. Just 13 of Morgan Stanley’s 2700 employees at WTC perished on 9/11 - Including Rick and his deputies Wesley Mercer, Jorge Velazquez and Godwin Forde who all went back into the building to help anyone who needed it.

We don’t all need to be heroes, defying death with Cornish battle songs and saving thousands of innocent people. But if we could all internalize just a small portion of the sense of duty, preparedness and selflessness that Rick possessed, the world would be a much better place. I share this story on each anniversary with the hopes it gets us just a little bit closer to that.