What to Buy as AI Stocks Diverge

Dear subscriber,

I don't have to tell you that everything AI has soared this year...

Or that it's led to inevitable predictions of an AI bubble.

That sort of talk has spread from financial circles to the general media and likely to conversations you have in your daily life.

Everyday publications like the New York Times, Reuters, and the Associated Press don't typically make stock predictions... But they've all declared that AI stock prices have gotten out of hand.

They're missing what an AI bubble really means...

A bubble is an investment frenzy that leads to a misallocation of capital.

There are hundreds of billions of dollars being poured into AI today. And some of it will go up in smoke.

That's how bubbles work. Someone has to fund the losers to find the winners.

But, as I'll discuss today, you don't want to be the one to fund the losers. You want to own the Amazon.com... not Pets.com.

Billions Do Matter

We've seen a handful of earnings reports and announcements from Big Tech over the past couple weeks. And we've learned a lot about the AI winners and losers.

For now, AI is still more about spending than it is about profits.

That's why earnings from Meta Platforms (META), Alphabet (GOOGL), and others were so closely watched.

Everyone wants to know if these companies will keep spending billions on AI infrastructure. After all, that's the fuel for the bubble.

And the answer is a resounding yes.

If you add up the capital expenditures from the big "hyperscalers" – Meta, Alphabet, Amazon (AMZN), and Microsoft (MSFT) – their combined spending for the trailing 12 months ending this September now tops $330 billion...

All this spending is fundamentally changing the earning power of these mega-cap tech companies.

To understand why, you have to look at the difference between earnings and cash flows...

On an earnings statement, a data center doesn't cost you that much.

If a data center will last 10 years, you book an earnings hit this year of one-tenth of the total cost of that data center. However, on the cash-flow statement, the total cost of the data center is accounted for. So you pay for the entire data center today.

One argument for why mega-cap tech stocks can continue to rise is that they're so profitable. Yes, they are huge – but they produce huge earnings to justify it.

Meta trades at a price-to-earnings ratio of just 22 times. Alphabet and Microsoft trade at more like 30 times... expensive, but not insane.

However, their cash flows tell a different story. Even with their historically huge profits, you can't spend hundreds of billions without it costing real money.

If a business's free cash flows match its net income, that's considered "high-quality earnings." But for Big Tech, total income now tops $360 billion, while free cash flows are about $207 billion...

These tech companies have gone from capital-light cash printers... to looking more like an industrial firm that has to build large factories or make other massive investments.

Spotting the Winners and the Losers

That brings me to a division we've seen over the past couple weeks – one that's pointing us to the winners and losers.

Last month, Meta announced that it would round out 2025 with about $72 billion in capital expenditures. And it said that 2026 would be "notably larger" due to investments on data centers and other AI infrastructure.

The market hated it. Shares dropped more than 15%...

But the opposite happened to Alphabet.

It announced spending of more than $90 billion for 2025 – up from previous projections – and said that 2026 would see a "significant increase." And Alphabet shares have spiked...

The difference here is twofold...

While Alphabet is spending on AI, it's also earning from AI. Google Cloud revenue grew 34% to $15 billion in the most recent quarter. And that growth was driven by companies using Cloud to host AI apps.

Overall, Alphabet notched its first-ever $100 billion quarter – making for a $400 billion annual run rate.

It was a similar story with Amazon's earnings. The company's Amazon Web Services cloud division grew 20% year over year – the fastest pace since 2022. And shares surged after lagging the rest of the "Magnificent Seven" for much of 2025...

Meta doesn't earn any revenue from compute like Alphabet, Amazon, and Microsoft do. And the market seems to realize that.

So the lesson here is that it's still better to be a recipient of AI spending than to be the AI spender.

Now, the entire promise of AI is that this equation will flip someday. AI companies like ChatGPT creator OpenAI will eventually see a return on all this spending and become profitable themselves.

But right now, the market's not rewarding them.

There's another feature separating the winners from the losers...

Meta is spending big on an AI future... while Alphabet is using AI today.

Yes, AI may help Meta's social networks and advertising business. It has an open-source model called Llama – but it has dropped into the also-ran category.

Meta isn't in the AI race to make incrementally useful models. Don't expect to switch from ChatGPT or Gemini to use something Meta has produced. Rather, Meta CEO Mark Zuckerberg's plan is to pursue AI "superintelligence."

Zuck's not pursuing a profitable consumer product (for now). He wants to be sure Meta doesn't get left behind when the robots take over.

Meanwhile, Alphabet is providing AI to millions of users right now. As of the most recent quarter, its flagship Gemini app had more than 650 million monthly active users. Query numbers tripled from the previous quarter. And, according to the company, AI is driving Google Search volumes higher.

The market says delivering real AI today is more valuable than promises of a far-off techno future.

Investors remember Zuck's wasteful foray into the "metaverse." And his superintelligence obsession sounds just a little too familiar.

Overall, you can sense a little wind coming out of the sails on the biggest AI bets.

The market is acting more rationally and rewarding firms with more concrete results.

The most levered AI plays have cooled a bit, too...

Take "neocloud" companies like CoreWeave (CRWV). These firms operate data centers specifically designed to take on marginal AI computing demand.

They're the most levered to AI's upside... and they'll be the first to suffer the downside. CoreWeave has already cooled 16% since last week...

Or take Palantir Technologies (PLTR), a pure-play AI company. It's instituting machine-learning tools into businesses and governments.

The issue is that Palantir's stock has run up to such a lofty valuation – at 460 times earnings – that it'll likely be the first struck by a decline in AI sentiment.

It's down about 7% after reporting earnings this week. But to be fair, that's just a blip on a long run...

The Truth About AI Today

The takeaway from the recent AI news is clear...

As we've covered, spending continues with no signs of slowing (yet).

For now, it's better to be on the receiving end of AI spending than to be sending all that cash out the door. And if you are spending, you need to create real AI products for consumers that they can use today.

This will make winners out of several mega-cap tech stocks...

I've long pointed to Alphabet as an AI beneficiary when others called it a victim.

At the release of ChatGPT, many considered it a "Search killer," and Alphabet shares plunged.

But that was shortsighted. Alphabet invented the transformer model that underpins large language models. Its AI tools like Gemini and NotebookLM often top the AI-model leader boards, and it already has billions of users it can put its AI tools in front of.

I called Alphabet the "hidden leader" in AI as early as February, and it has been a rocket of a stock.

Alphabet benefits both from AI spend (via its cloud division) and from its real-world applications, so it's still my favorite of the Magnificent Seven.

Of course, Nvidia (NVDA) is still the most direct beneficiary of AI. But at current valuations, investing in the stock is just plain crazy.

As early as July 2024, we proposed that Cisco Systems (CSCO) made for a better AI play. Its networking equipment is vital in the data-center build-out, and it's winning market share from Nvidia's networking products.

Cisco is up about 53% since then compared with Nvidia's 56% return. It has outperformed it for much of this year. And it's dramatically less risky to buy Cisco at 18 times earnings than it is to buy Nvidia at 45 times earnings...

The playbook going forward is clear...

1. Monitor AI spend and cash flow.

2. Focus on the clear winners (the ones providing real AI tools today).

3. And watch CoreWeave and Palantir as your bellwethers for AI sentiment. They'll be the first to show that investors are starting to worry about so much leverage.

To hear more about the developing winners and losers in AI, check out today's This Week on Wall Street video.

You can watch the entire episode on our YouTube page by clicking the image below. Be sure to like and subscribe to get more of our videos.

What Our Experts Are Reading and Sharing... 

• The headlines declare that Elon Musk was awarded a $1 trillion pay package. In reality, he'll get $1 trillion if he drives Tesla's (TSLA) stock value to $8.5 trillion, sells 20 million vehicles and 1 million humanoid robots, has 1 million robotaxis in operation, and the company reaches $400 billion in core profits. Lofty goals. But Musk's previous package only rewarded him if he drove Tesla's valuation from $60 billion to $650 billion. And he did it (though the $100 billion payday was invalidated by a judge).

• Democrats won elections this week campaigning on "affordability." Life's getting expensive, and people are angry. But economist Tyler Cowen warns that fixing affordability isn't as easy as campaigning on it.

• People are also angry that OpenAI CFO Sarah Friar suggested a government "backstop" for data center costs at a Wall Street Journal event this week. Friar and CEO Sam Altman went on a PR blitz to say the quote was taken out of context... But if you watch a video of the interview, it's clear that she meant it.

New Research in The Stansberry Investor Suite...

All investing boils down to supply and demand.

And when you can find a big supply-and-demand imbalance, it's the recipe for huge upside.

In this month's issue of Stansberry's Investment Advisory, Whitney Tilson and his team have found a mega-shortage in a key industrial ingredient.

It's not rare earths or critical minerals, like you've heard so much about in recent weeks in regard to semiconductor manufacturing.

But it is critical for energy generation. The average AI data center uses tens of thousands of tons of it.

That's why Goldman Sachs recently called this material the "new oil."

However, as the Investment Advisory team explains, production is down worldwide. New discoveries are rare. And that means prices for this material are about to spike.

Fortunately, they've found the perfect way to play this global shortage.

Stansberry Investor Suite subscribers can read the entire report here.

If you don't already subscribe to The Stansberry Investor Suite – and want to learn more about our special package of research – click here.

Until next week,

Matt Weinschenk
Publisher and Director of Research

What do you think about This Week on Wall Street? Send any and all feedback to thisweek@stansberryresearch.com. We read every e-mail you send in.