FEATURED

Meta’s Next Business Might Surprise You

Meta did something this month that I did not have on my 2026 bingo card.

It is not just building more AI infrastructure. It is flirting with the idea of renting it out.

On July 1, Bloomberg reported Meta is developing plans for a cloud infrastructure business to sell access to AI computing power and models. Internally, it sits under the Meta Compute effort and it lines them up against AWS, Azure, and Google Cloud. Not tomorrow morning, but the intent is real. Zuckerberg even told Bloomberg the pricing would be “very aggressive and attractive.”

Then on July 9, Reuters reported a second piece: Meta plans to start manufacturing an in-house AI chip, code-named Iris, in September. The same memo talked about scaling compute capacity from about 7 gigawatts coming online in 2026 to 14 gigawatts by 2027, plus a faster-than-usual chip release cadence through 2027.

That pairing is the signal. Build a lot of capacity, get better at feeding it, then see if you can charge someone else for what you overbuilt.

Here’s the part people skip

The market is not confused about Meta’s revenue engine. It is still ads. Q1 2026 revenue was $56.31 billion, up 33% year over year. Ad impressions rose 19% and the average price per ad rose 12%. Operating margin was 41%.

The argument is about the bill. Meta raised its 2026 capex forecast to $125 billion to $145 billion (from $115 billion to $135 billion). And investors have been treating every incremental dollar like a tax.

Iris is the most straightforward “we heard you” response. Reuters says Meta is working with Broadcom on design and TSMC on manufacturing. The memo also said one version cleared bug testing in roughly six weeks without major issues. If that holds up at scale, it is not just a science project. It is cost leverage.

Slight tangent, but it matters: if you have ever watched a company try to turn a cost center into a product, it usually looks messy at first. Lots of internal politics. A few awkward customer experiences. Then one quarter where the line item starts moving and nobody can ignore it anymore.

Numbers, updated to today

As of today (Tuesday, July 14, 2026), META is around $657. That is down from the July 10 close of $669.21, but still well above late June levels. The market cap is about $1.69T.

• Q1 GAAP EPS was $10.44, but it included an $8.03B income tax benefit. Underlying EPS was about $7.31.
• Q2 revenue guidance is $58B to $61B.
• The market is still treating July 29 as the next big checkpoint. Multiple earnings calendars list it for Wednesday, July 29, 2026 (estimated), and Meta has not formally confirmed the date yet.

One more reality check: people keep throwing around “cheap” valuations for Meta. At today’s price, the trailing P/E is roughly 24x (based on current market data). So yes, it can be cheaper than its recent history, but it is not the bargain-basement multiple some posts imply.

What I’m watching into July 29

1) Does management give a real commercialization path for selling compute, or is it still just “we’re exploring options” language?

2) Do they hint at customer type? Startups, neocloud partners, enterprises, model access, or all of the above?

3) Any change to the capex range. Not the number itself, but the tone around it. Confident, defensive, evasive. You can usually tell.

If you want one clean question to keep in your head, it is this: is Meta building a bigger ad machine, or is it trying to become a platform company in a second category?

Worth a look this week: pull up a chart of META against the cloud names that sold off on July 1. That spread is telling you what the market thinks the endgame might be. It might be wrong, but it is telling you anyway.