Meta Earnings Preview: Ad Growth vs AI Spending

Traders keep zooming in on the same two lines in Meta’s model: advertising revenue and capex. One is humming. The other is sprinting.

That’s the tension heading into this print. Has the ad machine recovered enough to comfortably pay for an AI buildout that keeps stretching timelines and budgets?

If you’re watching from the sidelines, the setup is clean: strong ad demand into a still‑aggressive spend cycle. The question is whose patience runs out first.

Meta is back to its roots: make ads work better, then pour the surplus into the next platform shift. Right now, that shift is AI, not just for research pride or consumer wow factor, but for the practical stuff that moves numbers: better targeting, higher relevance, and lower unit costs to run the whole system.

On the other side, investors are still nursing a hangover from the last spend cycle. Reality Labs burn didn’t vanish. It just got a roommate named AI infrastructure.

The market will forgive big cheques for GPUs and data centers if two things hold: ad growth doesn’t wobble, and AI features show up in revenue lines, not just keynote slides.

Where the ads are actually growing

Meta’s revenue is still overwhelmingly ads across Facebook and Instagram. The interesting bits aren’t new formats; they’re the pipes and pricing behind them. Three areas matter most this quarter.

Reels is maturing from time sink to revenue line

Reels used to dilute monetization per minute. Over the last year, Meta has repeatedly said the monetization gap versus Feed and Stories has narrowed as auction models and formats improved during training cycles on its ad systems. The tone on recent calls was that the gap keeps closing, but not fully closed. Expect watch‑time and pricing comments to be glued together.

Messaging is the sleeper growth lever

Click‑to‑message ads (WhatsApp, Messenger, Instagram DM) have been a multi‑billion‑dollar run‑rate business, highlighted repeatedly on earnings calls (The Motley Fool). The appeal is simple: measurable conversations that lead to conversions, especially in markets where storefronts are phones first. Watch for more automation here, including AI chat flows that handle the first few turns with a customer before a human steps in.

Signal loss is yesterday’s story, but only because AI learned new tricks

Meta’s Advantage+ tooling has helped rebuild performance after Apple’s ATT changes kneecapped some signal flows. Dynamic creative, broad targeting, and model‑driven budget allocation are doing more of the heavy lifting now (Meta for Business). Advertisers care less about explicit targeting knobs if outcomes stay consistent or improve.

Segment
Trend
What to listen for

Reels ads
Improving monetization efficiency
Pricing vs. Feed/Stories, watch‑time mix, creator payouts impact

Click‑to‑message
Steady share gains in commerce‑heavy regions
Automation/AI handoff, SMB adoption, measured ROAS anecdotes

Brand budgets
Mixed but stable
Macro sensitivity, retail and CPG commentary

App install/perf
Recovered from ATT trough
Advantage+ updates, attribution clarity, CPA stability

AI spend: from GPUs to Llama in your feed

Everyone quotes the capex number because it’s simple. The story underneath is not. Meta is building out training clusters, inference capacity close to the user, and the software stack that turns raw compute into better recommendations and ad outcomes.

The capex arc and what’s in the basket

In 2024, Meta raised its full‑year capex outlook to prioritize AI infrastructure, flagging heavier investment in data centers and silicon (Meta IR). External reporting at the time emphasized a multi‑year, GPU‑heavy build plan as the company scaled recommendation systems and model training (Reuters). That posture hasn’t softened. If anything, model sizes and inference demand are escalating the bill.

Llama isn’t a science fair project anymore

Meta’s Llama 3 release signaled a commitment to open(ish) models and developer reach, not just internal tooling (Meta AI). That’s not charity. A wider ecosystem can stress‑test and improve models while Meta reaps the benefits in its own ranking systems, safety layers, and ad products. The tight loop is: better models push more relevant content, better content boosts engagement, more engagement feeds better ads.

How the spending hits the financials

1. Upfront capex goes to buildings, networking, and accelerators. That shows up on the cash flow statement first.
2. Then depreciation flows through the income statement over years, widening the gap between GAAP earnings and free cash flow.
3. Operating expense creeps up with headcount and model operations work: evaluation, red‑teaming, safety, trust & support.
4. COGS pressure can rise with inference at scale. Personalization and generative features aren’t free to serve.
5. Payoff lands in two places: higher ad yield and new revenue lines (enterprise tools, API usage, subscription value‑adds).

Margins, buybacks, and the patience problem

It’s the classic trade: invest through the cycle or optimize margins now. Meta’s recent playbook has been invest first, then use buybacks to keep per‑share math friendly when cash gushes. The sticking point is timing. If capex keeps stepping up while ad growth settles into something closer to mid‑teens, the operating margin path can still look fine, but consensus patience shortens when guidance feels open‑ended.

What margin math the Street will test

Two sensitivities drive most back‑of‑the‑envelope models: ad pricing momentum versus user‑time growth, and the depreciation curve on the AI buildout. Strong Reels monetization and messaging adoption can offset a lot, but if Meta adds yet another capex tranche for fresh data center footprints, that depreciation tail gets longer. If guidance pairs that with clear AI revenue tie‑ins, it plays differently than a broad “we’re investing” line.

Reality Labs is still in frame

Even if the market is more forgiving on AI than on VR, consolidated results don’t care which unit overspends. Any widening RL losses will get questioned. Meta has framed VR/AR as a longer‑cycle bet, but the tolerance band narrows if AI capex also stretches. Expect at least one question pressing for clearer hurdle rates or milestones.

What to watch on earnings day

The setup can flip fast in the first five minutes of a call. Keep a tight checklist and map reactions to lines.

1. Ad growth ex‑currency. If it steps down meaningfully, every other line item gets interrogated twice.
2. Capex guide versus prior commentary. Any change in range or language will move the stock more than you expect.
3. Reels monetization vs time spent. The gap should keep narrowing; a stall invites margin questions.
4. Click‑to‑message updates. Concrete adoption anecdotes beat vague “strong interest.”
5. AI product tie‑ins. Examples where AI directly lifted an ad KPI or opened a revenue opportunity matter most.
6. Opex discipline. Hiring pace and areas of focus say more than headcount totals.
7. Buybacks and cash. If free cash flow is strong, capital returns can cushion heavier spend.

Why people in crypto and fintech should care

Two reasons. First, AI infrastructure spend pulls on the same supply chains that Web3 and quant shops increasingly tap for inference and model‑assisted trading. Scarcer high‑end accelerators and pricier cloud inference trickle into everyone’s budgets. Second, Meta pushing open models with Llama helps startups and protocols bolt AI into wallets, support flows, and creator tools faster, without vendor lock‑in. That speeds up the “AI inside Web3” story in real life, not just pitch decks.

There’s also the ad‑spend angle. When brands lean into performance channels that prove conversion, gaming, NFT projects, and fintech apps with clear CPA can benefit from incremental budget. If click‑to‑message commerce gets smarter, expect more experiments with payments flows that nudge closer to embedded finance. Not a stretch to imagine crypto on‑ramps side by side with chat‑based shopping in some regions, even if Meta itself stays cautious on custody.

Scenario map: how the quarter could break

You don’t need pinpoint forecasts to be prepared. You just need to anchor the likely paths and what would validate them.

Scenario
What we hear
Likely reaction

Goldilocks
Ad growth steady, capex flat to prior commentary, clear AI wins
Relief rally, estimates drift up

Spend first
Capex up again, benefits described but light on metrics
Knee‑jerk down, debate shifts to FY margin corridor

Ad wobble
Soft pricing or engagement mix issues
Multiple compression, heavier scrutiny on all spend

Mixed bag
Messaging strong, Reels fine, but RL losses widen
Choppy; stock trades headlines into guide

Risks & What Could Go Wrong

• Macro ad slowdown hits just as AI spend ramps, squeezing margins from both ends.
• Inference costs run hotter than expected if new AI features see heavy usage without offsetting monetization.
• Regulatory pressure on data usage or competition policy complicates model training and ad targeting.
• Supply chain or energy constraints delay data center buildouts, pushing spend into less efficient stopgaps.
• Reality Labs losses overshadow AI narrative if hardware cycles disappoint.
• Creator economy fatigue reduces Reels time spent, softening pricing power.

The biggest risk isn’t one ugly quarter; it’s a guidance reset that implies the capex curve keeps rising while revenue tailwinds fade.

If you want a clean daily pulse on how Big Tech’s AI bets intersect with Web3 and digital assets, I cover those crossovers regularly at Crypto Daily. The lens is practical: what shifts budgets, what changes unit economics, and where infra supply squeezes show up next.

Frequently Asked Questions

Why are investors so fixated on Meta’s capex guide?

Because it’s the clearest proxy for AI ambition and the longest shadow on future margins. Higher capex means more depreciation later, and the market wants to know if those dollars translate into better ad yield or new revenue lines rather than just bigger clusters.

What would qualify as proof that AI spending is working?

Concrete links: improvements in ad conversion and pricing attributed to model upgrades, lower content moderation costs from better classifiers, or early revenue from AI tools offered to businesses. Specific, repeated examples beat broad claims.

How important are Reels to this quarter?

Pretty important. Reels soaks up a lot of user time. If monetization per minute keeps catching up to Feed and Stories, the business can grow without needing a sudden spike in total time spent.

Is messaging commerce really a big deal or just hype?

It’s real in regions where shoppers prefer chat‑based buying. Meta has repeatedly highlighted click‑to‑message as a multi‑billion‑dollar run‑rate category on calls, and AI‑assisted chat flows can improve conversion and response times, which advertisers care about.

What could surprise to the upside?

Flat or lower capex guidance paired with stronger ad pricing would be the cleanest upside surprise. Clear metrics tying AI features to revenue, plus steady buybacks, would add fuel.

What about Reality Labs — does it matter for the stock reaction?

It can. Even if the Street is more charitable about AI, widening RL losses may compress the multiple if investors feel both big bets are crowding near‑term returns.

Where can I read official commentary on Llama and Meta’s AI plans?

Meta’s AI blog covers Llama releases and research direction (Meta AI), and Investor Relations posts guidance and call transcripts (Meta IR).

Disclaimer: This article is provided for informational purposes only. It is not offered or intended to be used as legal, tax, investment, financial, or other advice.