There’s a kind of earnings report I think of as technically excellent and emotionally catastrophic. Revenue beats. EPS beats by a wide margin. User growth holds. Operating income climbs. And then, somewhere in the prepared remarks, management drops a number that makes portfolio managers set down their coffee and stare at the ceiling.
For Meta’s Q1 2026, that number was $125 to $145 billion: the raised full-year capital spending guide, up from a prior range of $115 to $135 billion. The stock fell 10% the next day, on a quarter where almost every operating metric beat. That gap between the results and the reaction is the whole story, and it’s a useful case study in how the market is pricing AI spending right now. The read on which tech names to buy cautiously covers the same question across the rest of the sector.
What actually happened in Q1
Revenue was $56.31 billion, up 33% year over year. That’s Meta’s fastest growth since 2021, and unlike 2021 it isn’t a post-lockdown ad-price spike. It’s AI-driven ad efficiency showing up in the numbers. Operating income was $22.9 billion, up 30%. EPS came in at $7.31, helped by roughly $8 billion in one-time tax benefits, and still ahead of estimates. Q2 guidance of $58 to $61 billion points to another quarter near 30% growth.
None of those numbers explain a 10% drop. The capex line does. Full-year 2026 capital spending is now guided to $125–145 billion, against roughly $70 billion in 2025, close to a doubling of infrastructure spend in a single year. Meta attributed the increase to “higher component pricing and other data center costs to meet future year capacity.” In plainer terms: Nvidia chips didn’t get cheaper, and Meta decided to build more than it originally planned. I’ve written before about where this money goes across the sector.
| Meta Q1 2026 | Result | vs. year ago | vs. estimate |
|---|---|---|---|
| Total revenue | $56.31B | +33% | Beat |
| Operating income | $22.9B | +30% | Beat |
| EPS | $7.31 | Up sharply | Beat |
| Capex (Q1 alone) | $19.84B | — | Below $27.6B est. |
| Full-year capex guide | $125–145B | vs. ~$70B in 2025 | Raised |
| Q2 revenue guide | $58–61B | — | Above estimates |
The detail people missed: Q1 capex of $19.84 billion actually came in below the $27.6 billion analysts modeled. Meta didn’t spend more this quarter — it told the market it will spend more this year. The full-year guide moved the stock, not the quarterly print. Spending isn’t being cut. It’s being pulled forward.
The advertising business nobody wants to talk about
Everyone wants to discuss Llama, superintelligence, and Zuckerberg’s long game. We’ll get there. First, notice what the core ad business is doing: 33% revenue growth with no major new product.
That is not normal. Facebook is more than 20 years old. Instagram has existed since 2010. WhatsApp is around 17. These are mature platforms that, two years ago, most analysts expected to grow revenue in the mid-single digits. They’re growing at 33% because AI-driven ad targeting has measurably improved advertiser returns, and advertisers pay more per impression when the impressions convert. When you can target on behavioral signals across a family of apps rather than broad demographic buckets, you have the most efficient advertising machine ever built.
This matters for the investment case because it means the AI spending isn’t pure speculation. Some of it is already paying for itself in the ad column. The bet is that the next wave, from foundation models to an assistant across every app to enterprise tools that don’t exist yet, earns a similar return on a similar timeline.
The Llama strategy: clever or reckless
Llama has been downloaded more than 1.2 billion times. Anyone who knows open-source software should find that startling. Linux, the OS running most of the internet’s servers, most cloud infrastructure and every Android phone, took decades to reach that level of adoption. Llama got there in under two years.
Open-sourcing it wasn’t forced. It’s cheaper and faster for developers to adapt a good open model than to build one, and every developer who builds on Llama becomes part of Meta’s ecosystem. Each app built on it is a signal about what users want from AI, and every company shipping a Llama-based tool is a potential future enterprise customer. It’s the same playbook as Amazon pricing AWS to win the cloud, or Android given away to win distribution: ecosystem capture first, monetization later. Contrast that with Google’s closed Gemini or OpenAI’s API model.
The risk is timing. Ecosystem-capture plays run on a five-to-seven-year clock, not two to three. The $125–145 billion Meta spends in 2026 is funding a plan that pays off in 2028 at the earliest.
The Reality Labs problem
Reality Labs, Meta’s AR and VR division, lost about $4 billion in Q1 2026 on $402 million of revenue, roughly a $16 billion annual run rate. Cumulative losses since Meta began breaking the unit out now exceed $90 billion. The Ray-Ban Meta smart glasses are the division’s biggest commercial success, which tells you how far the rest of it is from paying off.
Zuckerberg isn’t backing away, and his dual-class share structure means shareholders can voice displeasure without changing the strategy. So the losses continue. Personally, I treat Reality Labs as a permanent drag on an otherwise exceptional business: a roughly $15–20 billion annual tax on owning Meta, payable in case the metaverse bet is right. Some investors won’t accept that, and valuation can’t settle the question for them. It’s a temperament call.
Regulation, which is always fine until it isn’t
EU probes. Disputes under the UK Online Safety Act. A failed China deal. The FTC antitrust case still grinding through the courts. Regulation has weighed on Meta for a decade, and over that same decade its revenue has grown roughly fivefold. Meta can absorb fines. The real threat is structural: a forced separation of Instagram from Facebook would matter, and antitrust lawyers have argued for it since 2020. It hasn’t happened.
I’m not dismissing the risk. The FTC case has run long enough that its outcome will mean something. But the market has priced in a regulatory catastrophe for Meta perhaps a dozen times in the past eight years, and each time the overhang eventually faded into “priced in.”
Valuation
Meta trades around $613, down roughly 23% from the $796 high it set earlier this year. Forward P/E is about 24 times consensus 2026 earnings.
That multiple isn’t demanding on its own. Twenty-four times forward earnings for a company growing revenue 33% would normally be a bargain. The catch is that the capex ramp will pressure the near-term earnings base. You’re not really paying 24x for 2026 earnings; you’re paying it for 2027 earnings that may run below the current rate while margins compress before they expand.
The more interesting question is what the business looks like in 2028 if the AI spend works. If Llama-based enterprise products bring in $20–30 billion a year by then, ad revenue keeps growing near current rates, and Reality Labs losses finally flatten (a real “if”), Meta’s earnings power looks very different from today’s.
Wall Street is largely on that side. The average rating sits near 96% Buy, with a price target around $840, roughly 37% above the current price. These analysts have modeled the capex. They’ve decided the ad engine and the option value of the AI ecosystem are worth owning through the spending cycle.
The bear case is simpler: what if the AI investment doesn’t produce a return on any reasonable timeline and Zuckerberg keeps spending $125 billion-plus a year anyway? That would hurt. It’s also the frustrating scenario where the core metrics stay strong and the stock still underperforms, purely on spending anxiety.
Where I land
Meta is more interesting at $613, after a 23% pullback, than it was at the highs. I won’t pretend the valuation is screaming cheap, because it isn’t. But after three years following this company, a few things pull me toward owning it.
The ad business is a higher-quality asset than it was in 2022. The AI targeting flywheel is turning: better targeting lifts advertiser ROI, which lifts ad spend, which funds more AI development, which improves targeting. That loop is already running and doesn’t depend on any future product to explain the current growth.
The Llama ecosystem play is a legitimate long-term bet with real logic behind it. Not guaranteed, but not crazy; it’s a platform strategy that has worked in adjacent contexts. And by most of the frameworks I use, a company growing revenue 33% with 40%-plus operating margins and no debt is not overvalued at 24x. The capex adds near-term uncertainty. It doesn’t change the quality of the underlying asset.
I’d own it here, and I’d add if the market hands us another capex-anxiety selloff. The number I’d watch every quarter is the ad revenue growth rate. If it decelerates toward 15% or below, the AI spending loses the cover it currently has from a core engine running this hot.
Financial disclaimer: The content on StockVane is for educational and informational purposes only and should not be construed as professional financial advice. Stock market investing involves risk of loss.