Explosive Q3, Diverging CSPs—What Exactly Happened?

It's earnings season now, and the info overload is insane—these past few days of digesting reports have left my eyes blurry.

 

Breaking it down into a few points.

Capex: Keep the Music Playing, Keep Dancing 

As everyone knows, U.S. data center stocks, including NV, rely on the CSPs for their meals.

 

 

So, the marginal changes in CSP capex this earnings season remain a key focus for global investors.

 

As of today, Google, Microsoft, Meta, and Amazon have all reported. The fact is, no big player dares to fall behind in this AI race—they're still FOMO-ing, and next year's capex expectations keep getting ratcheted up.

 

Meanwhile, another CSP giant, Amazon, posted results with cloud growth at 20% (beating the consensus 18%), shaking off last quarter's gloom. It also announced $125 billion in full-year 2025 cash capex, sending the stock up 13% after hours.

 

 

Amazon's CEO commented: "AWS growth has reached levels not seen since 2022, with year-over-year growth re-accelerating to 20.2%."

 

With the latest North American CSP earnings season landing capex steady, the U.S. data center industry chain—and the broader sector—has swallowed a reassurance pill.

 

The AI sector continues charging ahead at the industry level!

Divergence: Not Afraid of Big Spending, Afraid of Lousy ROI

So why such huge stock splits after these tech giants reported?

 

Google surges, Amazon explodes, Microsoft dips a bit, Meta craters—what's behind it?

 

I think the core key is: ROI.

 

All are FOMO-ing, all are aggressively throwing money at GPUs and data centers, but Meta is different from the other big techs—how exactly?

 

AMZN, MSFT, GOOG are all public cloud providers. Meta has no public cloud. Every GPU it burns, every data center it builds, won't turn into customer bills—it just becomes internal cost.

 

So,  While others are spending hundreds of billions on AI as "water sellers," Meta is the one "drinking the water" at the end.

 

The market's doubt isn't about Meta's taxes hitting profits—it's whether Meta's sustained high investment can feed back into the income statement. Bluntly, there's skepticism about Meta's ROI, especially with the metaverse still bleeding money.

 

If I sum up this earnings season's market sentiment toward big tech in one sentence:  

 

The market has shifted from "invest in AI at all costs" to "show me the profits from AI investment." It's not just about input anymore—ROI matters more.

 

 

Clearly, this round: Google delivered the perfect answer, Microsoft and Amazon were solid, Meta fell short.

 

First, Google: Stock rose because it hit $100 billion quarterly revenue for the first time, with profit and revenue beating expectations. Q3 revenue $102.3 billion, EPS $2.87. Google Cloud up 34% to $15.16 billion, search and ads stayed resilient. Company raised full-year capex to $91-93 billion for AI infrastructure. Market loves this spending because Google's revenue and cloud numbers show it's already paying off in profits. Especially, the search core held up surprisingly well—no one’s talking anymore about it being disrupted by AI search companies.

 

Next, Microsoft: Actually a strong report—revenue and EPS beat, Azure cloud up 40%—but the stock had priced in perfection, pressured by two things: single-quarter capex spiking to $34.9 billion (above expectations), plus a $3.1 billion non-cash OpenAI restructuring loss. AI drove revenue growth, but high costs worried investors about margins, leading to a small pullback.

 

Then META: Core ad business actually performed well, but the stock tanked. Market fears Zuckerberg's massive and uncontrolled spending: one-time $15.9 billion tax hit profit down 83%, 2025 capex raised to $70-72 billion, CFO warns 2026 will be "significantly higher." Reality Labs lost another $4.4 billion—AI and metaverse keep "burning cash" with returns hard to see. One-sentence summary: Google's spending looks justified by cloud profits; Meta's looks like speculative "money-burning" into projects with no payback.

 

Finally, Amazon: Last Q2 cloud dragged and got punished; this round cloud bounced back strong (though growth still can't match Google/Microsoft). AWS revenue up 20% YoY to $33 billion, growth accelerating (from 17.5% last quarter)—fastest since 2022, debunking "lagging in AI cloud race" talk. AWS CEO confirmed investments paying off, reviving the core; retail strength was just a bonus.

 

Lastly, I have to say: Meta just underperformed this earnings round, but that doesn't mean next one won't be good. Like last Q when the market was bearish on AWS, Amazon struck back with data this Q—hope Meta makes a king’s return next Q.

 

Of course, as I said, in the Mag7, personally I only buy the 3 I like best.

 

Storage: The Hottest Thing Going

 

Whether it's SK Hynix earnings, Samsung news, or U.S. stocks like Seagate yesterday and Western Digital today—the info all points to the same answer.

 

Listen to Hynix's own words: Even under conservative assumptions, the HBM market is expected to grow at an average annual rate of more than 30% over the next five years.

 

So, strategy's simple: If you're already in, keep holding. If not, buy on dips. Still bullish.

 

That's all for now.