Meta (META) just did the rare thing for a Magnificent 7 stock: it crushed earnings, raised guidance, and got punished for it. Shares dropped 7–8% in after-hours and premarket trading, and the morning conversation isn’t “buy the dip” — it’s whether Mark Zuckerberg has finally crossed the line on AI spending.
The headline numbers were strong:
- Revenue: $56.31B vs. $55.45B expected (+33% YoY — fastest growth quarter since 2021)
- EPS: $7.31 adjusted vs. $6.79 expected
- Operating margin: still printing >40%
And then the call happened.
Meta raised 2026 capex guidance to a range of $125 billion to $145 billion, up from a prior $115–$135B. The high end of that band is now within hailing distance of Microsoft’s and Alphabet’s combined AI spend — and unlike those two, Meta doesn’t have a cloud business to sell its GPU capacity to. It’s spending all of it on internal model training and Reels recommendation infrastructure that indirectly drives ad ROI.
That’s where the bear case lands. Investors gave Meta a pass on capex through 2024 and 2025 because Reels monetization was ramping fast enough to absorb it. But user growth disappointed this quarter — partly because of “internet disruptions in Iran” (the Hormuz war is killing connectivity in much of the Middle East) — and ad pricing growth is decelerating. The bull thesis required revenue growth to keep accelerating into the spend. This print suggests it’s plateauing.
Zuckerberg’s defense on the call: data center component costs are rising, and the additional capex protects 2027–2028 capacity. Translation: spending now to make sure Meta has GPUs when others don’t. That’s a legitimate strategic argument. It’s also exactly what every CFO says to justify capex they didn’t pre-announce.
Watch out for: the next two earnings cycles. If revenue acceleration doesn’t return by Q3, Meta becomes the cautionary tale of the AI capex cycle — and the stock gets a multiple compression that resembles 2022 more than 2023.
The Play: If you’re long-term bullish on Zuckerberg as a capital allocator (his track record is, for what it’s worth, pretty good), this -7% will eventually be a buying opportunity. But not today. The technical breaks below the 50-day moving average for the first time since November, and analyst capex models need a full week to update. The cleaner trade: if you wanted exposure to the AI spend cycle, GOOGL is taking flows from META today as the “capex with monetization attached” name. Pair trade: long GOOGL, hedge with a small META put if you really want to play the divergence.