Markets

Meta Platforms is spending like an AI hyperscaler—and getting grilled like a social platform

Date Published

Meta Platforms is spending like an AI hyperscaler—and getting grilled like a social platform

TL;DR

Quick Summary

  • Meta says 2026 capex will be $115–$135B, a huge jump from $72.22B in 2025—aimed at scaling AI infrastructure.
  • The core business is still ads: Q4 2025 impressions were up 18% and price per ad rose 6%.
  • Meta is also facing renewed scrutiny around privacy and child safety as encryption debates intensify.

#RealTalk

Meta is building an AI empire with ad money, but the company can’t outspend its way out of trust and safety controversies. Investors are being asked to underwrite both the infrastructure build and the social platform risks at the same time.

Bottom Line

Meta’s near-term narrative is a balancing act: enormous AI investment meant to reinforce an already-dominant ad business, set against persistent regulatory and reputational pressure. The key question isn’t whether Meta can spend big—it’s whether the spending strengthens the ad engine faster than the risks pile up.

What happened (and why it feels like two different companies)

Meta Platforms, Inc. (META) is having one of those weeks where the internet can’t decide whether it’s watching a sci‑fi origin story or a courtroom drama.

On one track: Meta is telling investors it plans to spend $115–$135 billion in capital expenditures during 2026 to build out AI infrastructure—data centers, chips, and the kind of compute that turns “we’re doing AI” into “we own the factory that makes AI go brr.” That range was discussed on Meta’s Q4 2025 earnings call held on January 29, 2026, and it’s a major step up from $72.22 billion of capex in 2025.

On the other track: Meta is back in the harsh spotlight on privacy and child safety—this time alongside Apple—after new reporting around litigation in New Mexico raised uncomfortable questions about what happens when messaging gets more private.

Both storylines are real. Together, they explain the core tension inside Meta right now: the company wants to be treated like an AI infrastructure leader, but it still earns its money like a mass‑market social platform that regulators and parents never stop watching.

The AI bill is enormous, but the ad machine is still the engine

Meta’s pitch is simple: the best AI in consumer apps requires the most compute, and the most compute requires the most spending. The spending looks dramatic because it is. Meta isn’t just buying GPUs; it’s building capacity so AI can power the things that actually make cash today—ad targeting, ranking, creative tools, and measurement.

And the cash engine has been humming. In Q4 2025 (reported in late January 2026), Meta posted about $60 billion in revenue, up 24% year over year, and about $201 billion for full‑year 2025, up 22%. Advertising was still nearly all of it. In that same quarter, ad impressions rose 18% year over year and the average price per ad increased 6%.

That’s the part investors care about when they hear “AI spend.” If AI helps Meta show better ads, in better places (Reels), to more people, more often—without torching user experience—then the capex starts to look less like a moonshot and more like a scale advantage.

But there’s a catch: Meta is funding the future by squeezing the present

Big ambition doesn’t show up only in data center invoices. It shows up in internal tradeoffs.

On February 19, 2026, reporting indicated Meta reduced annual equity awards by about 5% for most employees, following a prior cut the year before. Read that as: “We’re spending billions on AI infrastructure and talent, and we’re trying to keep everything else from ballooning.”

That kind of cost discipline can be good for shareholders in the abstract. In the real world, it can also affect morale, retention, and the vibes inside a company that already runs hot.

Meanwhile, the child safety and privacy conversation isn’t going away

The Meta story isn’t only about compute. It’s also about trust.

On February 20, 2026, reporting highlighted scrutiny on Meta and Apple related to child safety and privacy—specifically, concerns raised by unsealed documents tied to Meta litigation in New Mexico, including internal references to millions of child sexual abuse material reports that could become harder to surface after encryption changes.

Encryption is one of those issues where everyone wants a clean answer and nobody gets one. People want privacy. People also want platforms to stop the worst things imaginable. Meta sits in the middle of that contradiction, and every move it makes invites backlash from both directions.

What this means for Meta’s story from here

Meta is trying to do something rare: keep a giant, mature ad business growing while simultaneously building an AI infrastructure footprint that looks more like a hyperscaler than a social app company.

If it works, Meta’s advantage won’t be a single product feature—it’ll be that the company can afford to train, deploy, and iterate faster than competitors who have to rent their future. If it doesn’t, the question won’t be “did AI matter?” It’ll be “did Meta spend too much, too fast, while the platform kept accumulating social and regulatory risk?”