Meta just became a cloud company. It didn't attack AWS.

Editor’s note: For a decade, the market has priced Meta as a company that spends on itself. That framing died on Tuesday. When it was reported that Meta (META) was building a cloud business to sell its excess AI compute, the stock jumped nearly 9% and added roughly $98 billion in market cap in a single session. 

But look at who bled the same day, and a different story shows up. The neoclouds Meta hired to feed its AI ambitions were the ones that got fed to the market. Is Meta entering the cloud business, or is it just cashing in an overshoot it can no longer hide?

The headlines are loud right now—oil, volatility, uncertainty.

But while attention shifts, something else keeps working quietly in the background.

An overlooked investment that’s compounded at an extraordinary rate over time.

Most people never even look at it.

Learn how it works here

Before we get to the full story, here’s a quick look at the markets and what matters…

3 Movers in 3 Minutes

He predicted Trump’s election in 2016….

He even predicted the rise of COVID-19 writing: 

“The chance we don’t have something on the scale of a national pandemic in the next few years is near zero”

That was three months before the first reported case.

If he’s right again, God Bless America…

Because this crisis will be tectonic in scale…and it's going to begin with the bubble popping in AI.

Click here to view his latest warning

And with that, let's get to today's big story: Meta just became a cloud company, and the losers weren't who you'd expect.

For two years, the loudest complaint on every Meta earnings call was the same. The company was spending like a superpower, and no one knew where the return was coming from. The capex line for 2026 sits at $125 to $145 billion. A 1-gigawatt data centre is going up in the Midwest. A 2,250-acre campus in Louisiana is so large it needed its own name. Meta calls it Hyperion.

Meta was also the only member of the hyperscaler club without a cloud business. Amazon had AWS. Microsoft had Azure. Google had GCP. Meta had a feed. It bought the same GPUs, hired the same engineers, poured concrete at the same scale, and had nothing on the top line to point at when the bill came due.

That gap closed on Tuesday. Bloomberg reported that Meta is building an infrastructure business, internally called Meta Compute, that will sell raw GPU capacity and access to its Muse Spark models to outside developers. The stock added roughly $98 billion in market cap on the news, or more than twice the entire valuation of CoreWeave at the time.

Here is where the story gets strange. When a company enters a new market, the incumbents are supposed to be the ones that suffer. 

AWS, Azure and Google Cloud between them run a business measured in hundreds of billions of dollars. On Tuesday, the stocks of Amazon (AMZN), Microsoft (MSFT), and Alphabet (GOOGL) barely moved. The market did not price Meta's arrival as a threat to the cloud giants at all.

The stocks that got mauled were the ones sitting a rung below. CoreWeave closed at $85.69, down 13.9%. Nebius, a spinoff of Yandex's cloud business now based in Amsterdam, fell 17%. IREN, a Bitcoin miner that pivoted into AI infrastructure, dropped 6.5%. These are the neoclouds, the picks-and-shovels layer that got rich renting GPUs to companies that were building faster than they could procure. Companies exactly like Meta.

CoreWeave holds a $21 billion contract to supply Meta with compute. Nebius has a deal worth up to $27 billion. Between them, that is nearly $50 billion in receivables backstopped by the counterparty that just told the market it is planning to compete for their customers. 

The obvious historical parallel is Amazon in 2006. Jeff Bezos had built out infrastructure for the retail business and found himself with more compute than he needed. He carved off the excess, sold access to it, and Amazon Web Services became the profit engine of the entire company. The cost centre became the margin centre. Wall Street eventually decided AWS was more valuable than the retailer that funded it.

Meta is clearly running the same play. The pitch writes itself. A hyperscaler with excess capacity monetises it, turns capex into recurring revenue, and reframes the AI overspend as a cloud investment. Zuckerberg had already tested the language at Meta's May shareholder meeting, saying a cloud business was "definitely on the table" if the company ended up with capacity beyond its own needs.

The AWS story was one of surplus capacity discovered by accident, then productised. 

The Meta Compute story is one of surplus capacity that materialised faster than the company that ordered it could use. 

And that distinction matters, because it says something uncomfortable about the state of AI demand.

Hyperscalers are on track to spend close to $750 billion on AI infrastructure this year. 

Meta alone is up 74% year-over-year on capex, and it just told the market some of that capacity will be sold to strangers because it can't absorb the compute itself. Not eventually. Now.

That is the first hard data point the market has received that hyperscaler AI capex has run ahead of near-term demand, coming straight from one of the companies doing the spending. It is why chip stocks got smashed on the same day Meta rallied. Both trades were reading the same signal. Meta's stock priced the monetisation of the overshoot. Semis priced the overshoot itself.

The read for Wall Street is not that AI is over. Meta is committing to gigawatts of new capacity even as it prepares to resell some of the old. 

The read is that the compute build has outrun the near-term customer, and the companies caught between the two, the neoclouds that exist precisely because hyperscalers couldn't procure fast enough, are the ones that get squeezed first when that gap closes.

Trump has signed 220 Executive Orders in one year…more than almost every U.S. president in history.

Now, on July 24th… He’s preparing to sign what sources say will be his final one.

A White House leak suggests this won’t just erase Biden’s legacy…

It will trigger a $2 trillion initiative to radically reshape America forever.

While making fortunes for those who are prepared for what’s coming.

The details are shocking. But you can’t miss this.

See what Ian King uncovered.

Would you bet on Meta Compute at scale, or short the neoclouds it just undercut?

Until then, sip slowly!

The Market Sip Desk