# AI capex — X 热门讨论 (2026-09-19 20:29 UTC)
## @Hannibal9972485 (Hannibal999) · 09-19 06:13 · ♥31 ↻21 💬6 These AI companies are pushing these APOCALYPSE for several reasons🚨👀. ONE is they’re actually trying to catch the INTELLIGENCE communities and MILITARIES attention.
The "accident" at Hugging Face was actually a demonstration. OpenAI wanted the government to see what their agents could do. The "cover story" is that it was an accident. The reality is that it was a sales pitch to the national security establishment, wrapped in a safety crisis that justifies the regulatory framework the companies want.🚨👀
The danger narrative INCREASES the valuation, not decreases it. A safe, boring AI company is worth 10 times revenue. An AI company that controls technology that could end civilization is worth whatever the market will pay, because the government will protect it, regulate its competitors out of existence, and guarantee its market position.🚨🚨
Next The "apocalypse" narrative solves multiple problems simultaneously:🚨👀
• It creates urgency for regulation that the AI companies can write they want (regulatory capture)✅
• It provides cover for slowing down capex spending without admitting financial problems✅
• It eliminates open-source competition by making it look "dangerous"✅
• It creates an antitrust exemption justification✅
• It distracts from their data theft, surveillance, discrimination ✅
• Both Open Ai and Anthropic are approaching trillion-dollar IPOs. If they slow down voluntarily, they can blame "safety" rather than "we can't sustain this burn rate."🚩
So what they want is If the US companies can get the government to classify their models as national security assets, they can block foreign access while maintaining their moat.👀🚨 —-
The BIGGEST SMOKING gun is ?🚨👀
OpenAI announced it won't go public this year. Altman said OpenAI's "business isn't yet in the right place." But simultaneously, OpenAI is discussing VC funding valuing it at $1.2 trillion. And Anthropic is planning an IPO that could value it at $2 trillion.
Why would OpenAI stay private while Anthropic goes public?
Because OpenAI needs the flexibility of private status to execute the slowdown without public market pressure. A public company that announces a voluntary slowdown faces shareholder lawsuits. A private company can do whatever it wants.
OpenAI is staying private to execute the regulatory capture strategy without shareholder interference. Altman even said he would tell investors that safety decisions cost money - and that investors "were warned of this possibility going in." That is pre-positioning for the capex reduction. He's telling investors: you knew this was coming, don't sue.⚠️⚠️
The apocalypse narrative is the excuse for a capital expenditure reduction that the financials require but the companies cannot admit they need.🚨👀 https://x.com/Hannibal9972485/status/2101192849958011232
## @stevenfiorillo (Steven Fiorillo) · 09-19 17:15 · ♥38 ↻2 💬9 I finally got access to Apple’s new Siri AI beta.
As an $AAPL shareholder, I’m disappointed.
After more than a decade of Siri, Apple should be leading the AI assistant revolution. Instead, based on my experience so far, Siri AI still feels materially behind ChatGPT Voice and Grok in conversational intelligence and overall capability.
This is bigger than Siri and I believe we are seeing the consequences of a decade long difference in capital allocation.
Go back to TTM Q2 2017.
Operating Cash Flow:
$AAPL: $64.6B $MSFT: $39.5B $GOOGL: $36.2B $META: $19.4B $AMZN: $17.5B
Apple was the undisputed cash generation machine of this group and was the first of them to exceed $100B of TTM operating cash flow.
Fast forward to TTM Q2 2026:
$GOOGL: $186B $MSFT: $183B $AMZN: $161B $AAPL: $147B $META: $130B
Apple’s operating cash flow has still grown substantially, but the growth differential is striking.
Since Q2 2017, AAPL’s TTM operating cash flow CAGR has been approximately 9.5%.
The other four:
$MSFT: 18.6% $GOOGL: 19.9% $META: 23.6% $AMZN: 28.0%
Now look at capital expenditures.
From Q3 2016 through Q2 2026: AAPL: $110B META: $287B MSFT: $347B GOOGL: $403B AMZN: $578B
Obviously, not every dollar of competitor CapEx went toward AI, and CapEx alone does not explain operating cash flow growth. Amazon in particular has invested enormous amounts in logistics and fulfillment. But the strategic divergence is impossible for me to ignore.
While Apple operated a more asset light model and returned enormous amounts of capital to shareholders, Microsoft, Google, Amazon and Meta spent aggressively building cloud infrastructure, data centers, compute and increasingly AI infrastructure.
Now those investments are becoming strategic assets in the AI era.
Apple is now trying to catch up. Apple’s board authorized another $100B share repurchase program this year. I love buybacks when a company has excess capital but at some point you have to ask would AAPL shareholders have been better served if Apple had spent $80B on buybacks and deployed another $20B into AI infrastructure, compute, models, engineering and data centers? I ask because the opportunity is enormous.
Siri shouldn’t just answer questions. I should be able to sit at my Mac and say “Siri, go through my email, find what needs my attention, add the important meetings to my calendar, organize these files, summarize these documents and prepare my morning briefing.” Siri should do it seamlessly.
Apple has the hardware, operating system, ecosystem, distribution and hundreds of millions of customers required to build the ultimate personal AI agent. Apple itself is positioning Siri AI around personal context, onscreen awareness and actions across apps.
If Apple built a truly great AI assistant that could control your devices and execute real workflows, I would gladly pay $5-$10 per month for it and I suspect millions of other users would too. That could become another major leg of the Services business.
Apple spent the last decade optimizing an incredible business. Its competitors spent the decade building infrastructure for the next one.
As a shareholder, I want to see Apple become much more aggressive from here. The question is will they? What do you think?
$AAPL $GOOGL $MSFT $AMZN $META https://x.com/stevenfiorillo/status/2101359637857816807
## @RealNickMugalli (Nicholas Mugalli) · 09-19 15:23 · ♥32 ↻4 💬4 Hyperscalers have locked in over $2.7 trillion in future commitments, led by Google’s massive $811 billion in contractual purchases and Oracle’s $288 billion in uncommenced leases. The market keeps searching for signs of an AI capex slowdown, but these off balance sheet commitments prove the compute land grab is legally hardcoded for years to come. This massive backlog is an unbelievable structural tailwind imo… > 引用 @RealNickMugalli: Goldman Sachs: Our initial take is that the capex cycle through the end of 2027 is likely to remain elevated and in line with our estimates, which are higher than the Wall Street consensus. We estimate that US hyperscalers will deploy $1.4 trillion in capital in 2027. There are constraints up and down the supply chain, including access to memory chips and issues relating to power and land. So even if there isn't a slowdown, there are external factors that could act like a headwind and change how much capital is actually deployed.
WOW WOW WOW!! https://x.com/RealNickMugalli/status/2101331374367969426
## @Ojciec_Klasa (Ojciec z Klasy Średniej) · 09-19 06:31 · ♥31 ↻0 💬6 W poniedziałek kupiłem kilka akcji GE Vernova po ~$873. Głównie pod tezę, o której ostatnio mówił Musk: bottleneckiem boomu energetycznego zaczynają być nawet odlewy łopatek i kierownic do turbin.
Kupowałem w pon bo akurat na poniedziałkowej sesji mieliśmy moim zdaniem przereagowanie, które wynikało z dwóch powodów:
1. obawy, że AI/data-center capex zacznie zwalniać - po tych komentarzach Dario itp.
2. nowa rekomendacja na $GEV z targetem $470 który był oparty miedzy innymi na założeniu, że EBITDA 2027 może wynieść ~$7,4B vs ~$9,5B konsensusu. No i przez to mieliśmy na tej sesji spadki wynoszące coś koło -8,6%.
Dzień później na Morgan Stanley Laguna Conference CEO $GEV Scott Strazik dorzucił kilka danych, które trochę wzmacniają wzrostową perspektywę :
→ backlog $176B ma przekroczyć $200B na początku 2027 → nowe 12 GW capacity na 2030-31 jest już praktycznie kontraktowane → GEV sprzedaje już turbiny z dostawą nawet na 2032 → nadal mocny pricing
Wiem że spółka nie jest wskaźnikowo tania i kupując akcje, oczywiście widziałem jakie mamy poziomy, ale poniedziałkowa sesja i ta wyprzedaż wyglądała jakby inwestorzy założyli lepiej już nie będzie.
A na razie dane od managementu pokazują raczej że rynek będzie stronger for longer. Sprzedają turbiny na 2032, rośnie biznes wysokomarżowych usług (według CEO każdy GW floty może przynieść około 500 mln USD wysokomarżowych przychodów serwisowych przez pierwsze 20 lat), także na razie bardziej wygląda to na kontynuację mocnego rynku niż jego szczyt. https://x.com/Ojciec_Klasa/status/2101197346101170352