3li.Mo7d

[Wall St Engine @wallstengine]

ZOOM $ZM Q2’27 EARNINGS HIGHLIGHTS

🔹 Revenue: $1.28B (Est. $1.27B) 🟢; +4.9% YoY
🔹 Adj. EPS: $1.55 (Est. $1.48) 🟢; +1% YoY
🔹 Enterprise Revenue: $787.5M; +7.8% YoY
🔹 Non-GAAP Op Margin: 40.0%

FY27 Guide:
🔹 Revenue: $5.09B-$5.10B (Est. $5.09B) 🟡
🔹 Adj. EPS: $6.08-$6.12 (Est. $6.04) 🟢
🔹 Oper Income: $2.07B-$2.08B (Est. $2.08B) 🟡
���� Free Cash Flow: $1.78B-$1.82B (Est. $1.79B) 🟡

Q3 Guide:
🔹 Adj. EPS: $1.46-$1.48 (Est. $1.50) 🔴
🔹 Revenue: $1.275B-$1.280B (Est. $1.28B) 🟡
🔹 Operating Income: $510M-$515M (Est. $524M) 🔴

Other Q2 Metrics:
🔹 NRR (Enterprise, TTM): 99%; up from 98% YoY
🔹 Customers >$100K: 4,625; +8.2% YoY

Capital Return:
🔹 Buybacks: ~3.7M shares in Q2; 44.2M total under plan
🔹 Remaining Authorization: ~$1.3B

Comments:
🔸 “Total revenue grew 4.9% year over year, anchored by 7.8% growth in Enterprise revenue, its strongest growth rate in three years.”

🔸 “Our AI-first Customer Experience portfolio continues to scale, delivering high-double-digit ARR expansion, driven in part by strong adoption of Zoom Virtual Agent, whose customer count increased 256% year over year.”

[3li.Mo7d @_3limo7d]

@Aly_jezzini الخلاص الوحيد من الحرب لايرانيين هوي أسعار الـ yields ع سندات الخزانة …. هي الوحيدة القادرة تخضع ترامب مثل مافعلت 2025 خلال أزمة التعريفات الجمركية وي الصينيين

[The Earnings Correspondent @earnings_guy]

$COHR (Coherent) #earnings are out:

[Neel Somani @neelsomani]

I'm a former Citadel quant who covered power & gas.

There's constant talk about chips & memory, but power is the central bottleneck for AI.

Very few people understand it, so I'm posting a canonical primer on power pricing & data centers: https://t.co/LO5ovj2imA

[3li.Mo7d @_3limo7d]

@wallstengine $ARRY plz

[Kris Patel 🇺🇸 @KrisPatel99]

Just went Long on $MU think this a good trade as it

Let the bounce begin...

[Investment Wisdom @InvestingCanons]

Stanley Druckenmiller:

“That would be my #1 advice to young people: Do not invest in the present. The present is not what moves stock prices. Change moves them.”

[3li.Mo7d @_3limo7d]

@StockMKTNewz @stocktalkweekly Damn, you look ready to open the market on the NYSE floor with that fit. Hope you’re taking home all the liquidity today sir @stocktalkweekly

[Garcia @GarciaCap]

$PLTR

[Wall St Engine @wallstengine]

HOW SITUATIONAL AWARENESS’S $45B FUND UNRAVELED:

Leopold Aschenbrenner, a 24-year-old former OpenAI superalignment researcher, launched Situational Awareness in 2024 after publishing a viral 165-page essay arguing that most of the world was underestimating how quickly AI would advance.

The fund was built to profit from that thesis through concentrated investments in companies expected to benefit from AI across semiconductors, energy, infrastructure and software.

Its investors included Stripe co-founders Patrick and John Collison, senior AI executives at Meta and Jane Street. Some investors committed at least $25M and agreed to lock up their capital for years.

As the returns accumulated, money and attention crowded around Aschenbrenner. His regulatory filings were treated by many investors as a guide to the AI trade, and stocks publicly associated with Situational became increasingly visible.

Major banks also lined up to finance the fund.

By May, Situational was up roughly 270% net for the year and more than 1,000% net since inception, with well over $20B under management. The WSJ now describes it as a $45B fund.

The problem was how those returns were being generated. For every $1 of capital, Situational borrowed another $3 to $4, and sometimes more, while also using options to increase its exposure.

That worked while its concentrated AI positions were rising, but left the fund with little room to absorb a broad decline.

Goldman Sachs, JPMorgan, Citigroup and Bank of America were among the banks financing its trades. Earlier this year, the fund reportedly approached additional banks seeking even more borrowing capacity.

As recently as July 10, confidence remained high enough for Situational to serve as an anchor investor in SK Hynix’s $27B U.S. listing. But by the middle of July, sentiment around the AI trade began to change.

The emergence of cheaper Chinese open-source models raised questions about how much value would ultimately remain with the semiconductor, infrastructure and cloud companies supporting AI.

Investors began reducing exposure to the trade, hitting stocks associated with Situational, including SK Hynix, Sandisk, Bloom Energy, Nebius, CoreWeave and Core Scientific.

The selloff was broader than one fund. Over the three trading days through Tuesday, hedge funds reduced positions at a pace not seen in three years, according to Goldman Sachs.

For Situational, however, the leverage made the decline much more damaging. Banks began monitoring the fund’s performance daily and issued margin calls requiring additional collateral.

Other hedge funds reportedly shared information about Situational’s positions and shorted some of its largest holdings, expecting the fund would eventually be forced to sell.

That created a feedback loop. Falling prices produced margin calls, margin calls forced Situational to sell, and those sales placed further pressure on the same stocks. Aschenbrenner later compared the process to a bank run.

Because many LPs were locked into the fund for years, this was not primarily a rush of investors demanding their money back. The immediate pressure came from prime brokers demanding collateral and other traders positioning for forced liquidations.

All of this unfolded during the week of Aschenbrenner’s wedding. As guests began arriving in Carmel for the multiday celebration, he and his team were negotiating late into the night to keep the fund operating.

Late Wednesday, Aschenbrenner reached an agreement to sell Situational’s $3.5B Anthropic stake to a group led by Greenoaks and Sequoia. By Thursday morning, he had reversed course, deciding to preserve the private-company portfolio and sell public stocks instead.

Citadel and Millennium then negotiated with Situational past midnight. Citadel ultimately reached a deal just before Thursday’s open to buy the vast majority of the fund’s public-equity portfolio at more than a 10% discount to prevailing market value.

That discount came on top of the losses those stocks had already suffered. It was not a discount to Situational’s original purchase prices. Citadel then benefited from an immediate rebound as many of the acquired AI and semiconductor stocks rallied Thursday.

Situational had met its margin calls through Wednesday and had not formally defaulted. The Citadel transaction was an emergency deleveraging that allowed the fund to repay its lenders and retain private-company stakes valued at more than $10B.

The fund finished July down 67%, though its investor letter said it remained up roughly 80% YTD. It plans to continue operating and investing in public equities, but will stop using bank leverage and make changes to its portfolio-management and risk teams.

“I take full responsibility for these events,” Aschenbrenner told investors.

In a 2024 interview, he had described the central rule of his investment strategy more simply: “Not blowing up is task No. 1 and 2.”

[Antonio Linares @alc2022]

The stock market will run a test once in a while to see how much of a pussy you are

[Michael Sikand @michaelsikand]

*TRUMP THREATENS TO WITHDRAW MILITARY SUPPORT FOR SOUTH KOREA IF THEIR STOCK MARKET CAN'T STOP RUGGING

[Jensen Huang @JensenHuang]

For my first post, I’m sharing a letter @NVIDIA signed on why open models matter.

AI will transform every industry, power every company, and be built by every country.

Open models strengthen safety and cybersecurity, accelerate innovation and diffusion, and enable sovereignty.

The world needs both frontier closed models and frontier open models.

https://t.co/AUKzoQ5Ikb

[Stock Talk @stocktalkweekly]

In light of this past week's releases of China's 'Moonshot Kimi K3', and now 'Alibaba Qwen 3.8 Max', many of the discussions are centered around reiterating the need for additional compute for inference. I think this is a fairly rudimentary, obvious observation. These discussions are overlooking the more significant realization, which is the "decentralization of compute".

The U.S. AI strategy has thus far been built on the notion that a handful of companies should spend trillions of dollars on massive datacenter buildouts in order to train & deploy models. That strategy is being challenged for the first time. Whether or not distillation is part of the Chinese strategy, the results are what matters. The results are competitive models being produced by more parties, with less spend. We haven't found a way to outright prevent distillation, anyhow. If it was easy, Anthropic & OpenAI would have spent whatever was necessary already in order to put those protocols in place.

So, yes, overall more compute will be needed for inference as AI proliferates, but the burden of the cost & control of that compute will likely become more fragmented. Smaller AI labs all over the world, and enterprises deploying custom models, will likely begin acquiring (or renting) stacks of compute of their own at much greater frequency than we’ve seen thus far. The hyperscaler giants need not bear the entirety of the capex burden, and in the coming months/years, the industry will likely begin to come to terms with that. Call it decentralization, fragmentation, redistribution — whatever you’d like, but the sources generating the capex will become more scattered.

I think Meta's recent foray into cloud is an early indication of this. While many initially interpreted it as a representation of excess compute capacity, it was much moreso an indication & realization of the impending need to distribute that capacity more broadly. In my view, this initially reads as structurally bullish for the cloud compute industry as a whole.

[Dean W. Ball @deanwball]

Some observations on Kimi:

1. It's a very good model! I don't think its performance can be explained away by distillation or anything like that. In agentic coding sessions, it seems pretty much on par with the best public models of Q1 2026. In my fairly limited use, it also seemed very token hungry. It's not obvious to me that this model is actually that cheap to run.

2. I am personally surprised the Chinese state continues to allow the open sourcing of models this good, given potential risks. To be clear, I *myself* might be fine with models presenting this level of marginal risk being open weight, but I am surprised that China is fine with it. I suspect the reason they are is 75% explained by strategic blindness/lack of AGI-pilledness (the CCP is very Yann Lecun-y in its views of AI). The other 25% or so is their lack of compute for customer inference (making China's open-weight strategy an unintended byproduct of US export controls) and the normal Chinese strategy of aggressive exports. For the companies, as opposed to the government, the decision to open source is partially ideological and partially because they are behind, and they know that very few people would pay for sub-frontier models from China.

3. Open-weight models are inherently decelerationist, and I'm continually surprised to see the so-called "accelerationists" so excited about open-weight models. I suspect the reason they are is that they know open-weight models are effectively ungovernable, and they simply like the overall cloak of ungovernability open-weight models create over the whole of AI. It's not a bad strategy; it reminds me of James Scott's recounting of the hill people in "the art of not being governed." Still, in the end, open-weight models deter further AI capex.

4. One probable outcome of an open-weight-model-dominant world is full AI communism, which is precisely what China proposes: rather than a market product, AI is a "public good" which will ultimately be provided by the state as a kind of "digital public infrastructure." This future strikes me as a dystopian hellscape, but I've never met an open-weight models advocate who doesn't ultimately concede this is where things end. You'd be surprised how many 'accelerationists' lobbied me, while I was in government, to support an eleven or twelve-figure federally funded data center so that startups could train models at a subsidy and then give them away for free. There was no other way for AI to progress, they said. Perhaps this is the logical end state of things. Nonetheless, I find myself surprised to see supposed accelerationists excited about such an outcome. I think many of them just don't know what they're doing. Many accelerationists do not view the creation and serving of frontier models as a legitimate business.

5. I would guess that the Trump Administration will at some point realize that their best strategy here would be to create large amounts of regulatory risk around the use of open-weight Chinese models. You don't need to "ban open source" (one of the dumber motifs of AI policy discussion). You just need to direct every agency to issue soft law that creates FUD. "A Federal Reserve Advisory Bulletin found that there may be backdoors in Chinese AI models." It needn't be that well justified. You just create enough regulatory risk that every regulated enterprise backs off. You probably don't want to create so much regulatory risk that you scare off the hyperscalers from serving Chinese models; this will just drive startups to sketchier providers. There's a happy middle ground here. I'd assume they will do some version of this.

6. It's probably true that open-weight models of this capability make the world a bit more dangerous, but not so much more that you'll really notice. At some point the models will be capable enough that you will notice. "A nonliving, invisible, dangerous, and infinitely self-replicating agent escaped from a Chinese lab," you say? Color me shocked.

[benny ッ @imbennythebull]

Me watching all my stocks slice straight through every single support level

[Wall St Engine @wallstengine]

$AAPL has sent legal letters to about 40 former Apple employees now at OpenAI, per FT.

The letters direct them to preserve documents and communications and demand meetings with Apple’s lawyers.

That is roughly 10% of the 400 former Apple employees now working at OpenAI.

Apple says its current evidence is only the “tip of the iceberg,” while OpenAI says it is not aware of evidence that the complaint has merit.

[Oguz Erkan @oguzerkan]

The argument that Kimi 3 changed the AI capex game is nonsense.

We saw the same panic with DeepSeek R1 release last year. Nasdaq dropped 4% while $NVDA dropped by almost 10%.

What happened then? Capex still kept growing, tech recovered, and delivered a 45% rally since then.

The critical point you need to understand is that if American labs didn’t make all those investments, Kimi wouldn’t be possible because distilling American models is how they achieve most of their training efficiencies.

It’s not that Chinese geniuses found a way to built a frontier model with less capex that American labs can’t solve.

This is why we saw capex estimates kept growing despite DeepSeek.

Nothing different this time.

This is not to say some companies in semiconductors and AI spaces aren’t extremely overvalued. They are, but what I mean is that it doesn’t make sense to sell $NVDA at 20x earnings or quality neoclouds like $CRWV and $NBIS as they are already not that expensive.

Long $NVDA $CRWV $NBIS

[Anika @AnikaSomaia]

a week ago @SemiAnalysis_ wrote that Chinese labs are "simply too compute poor to truly reach the frontier." today one of those "too compute poor" labs, a 300-person startup actually, shipped a model that compares to opus 4.8

the entire western consensus – export controls, the $650B hyperscaler capex race, the "compute moat" investment thesis – is built on one assumption: flops gate capability. if that were true, chip controls would keep chinese labs permanently behind the frontier.

but after reading through moonshot's stack i no longer think it is. training is efficiency-compressible: MoE routing, INT4-native quantization, better data curation, infra built around scarcity (their Mooncake stack exists because they don't have gpus!). a small lab with taste can compress the compute needed to make a frontier model, even if it can't afford to serve one

the frontier is no longer something money can buy