To The Moon

Meta’s latest earnings: the quarter was strong — but the capex number is the real story

Meta just reported a very unusual quarter.

On the surface, the numbers were excellent: Q1 2026 revenue reached $56.3 billion, up 33% year over year. Operating income rose 30% to $22.9 billion, while operating margin stayed at 41%. Family daily active people reached 3.56 billion, up 4% year over year. Ad impressions grew 19%, and average price per ad increased 12%. In other words, the core advertising machine is not broken. It is accelerating. 

But the market did not focus only on the beat. It focused on one line in the outlook:

Meta now expects 2026 capital expenditures of $125 billion to $145 billion, up from the previous range of $115 billion to $135 billion. Management said the increase reflects higher component pricing and additional data center costs to support future capacity. 

That is the story.

Meta is no longer just an advertising platform with high margins and enormous free cash flow. It is becoming one of the world’s largest AI infrastructure companies — but without selling cloud infrastructure the way Amazon, Microsoft or Google do.

That distinction matters.

The paradox: Meta is performing well, yet becoming more capital intensive

Meta’s ad business looks healthier than many expected. The combination of higher ad impressions and higher pricing suggests that AI is already improving ranking, targeting, creative tools and engagement. For advertisers, better AI means better conversion. For Meta, better conversion means more pricing power.

So the bull case is clear: Meta spends heavily on AI infrastructure, uses that compute to make its ad engine smarter, increases monetization across Facebook, Instagram, WhatsApp, Threads and business messaging, and eventually builds new AI-native products for billions of users.

The problem is that the investment required to get there is enormous.

Meta generated $32.2 billion in operating cash flow and $12.4 billion in free cash flow in Q1, but capex for the quarter alone was already $19.8 billion. That means the business is still hugely profitable, but a much larger share of its cash generation is now being redirected into servers, GPUs, data centers, networking equipment and power infrastructure. 

This is why investors are nervous. The question is not whether Meta can afford the spend. It can. Meta ended the quarter with $81.2 billion in cash, cash equivalents and marketable securities. The question is whether the returns on this spend will be as attractive as the returns Meta historically earned from its software-led advertising business. 

The bond sale says something important

Meta’s capex increase was followed by a $25 billion investment-grade bond sale, after a $30 billion bond sale last year. Reuters reported that Meta is increasingly using debt to fund its AI infrastructure ambitions, a shift from the years when internal cash flow funded most expansion. 

This does not mean Meta is financially stretched. Reuters also reported that S&P rated the new debt investment-grade and kept a stable outlook. But S&P also noted that Meta’s AI investment is starting to affect credit metrics. 

That is a subtle but important signal: AI infrastructure is changing the financial profile of even the strongest tech companies.

For years, the best internet platforms were “asset-light” compared with industrial companies. They scaled software, attention and data. Now, the AI race is pushing them toward an asset-heavy model: chips, energy, land, data centers, cooling systems, long-term supply agreements and depreciation.

Meta is still a high-margin platform business. But the infrastructure layer beneath that platform is getting much heavier.

The consequences of higher capex

The first consequence is pressure on free cash flow.

Capex does not immediately hit the income statement in the same way operating expenses do, but it does reduce free cash flow immediately. Over time, it also increases depreciation. So even if revenue keeps growing, investors will watch whether operating margins can hold once today’s infrastructure buildout flows through the P&L.

The second consequence is a higher bar for management credibility.

Meta has been here before. The company spent aggressively on the metaverse, and investors eventually forced more discipline. The “year of efficiency” restored confidence because Meta proved it could cut costs, improve margins and refocus. Now management is asking investors to trust another huge investment cycle — this time in AI.

The difference is that AI is already closer to the core business than the metaverse was. AI improves ads, feeds, recommendations, content creation, messaging, customer service and developer tools. But the size of the investment means “strategically important” is not enough. Meta will need to show measurable monetization.

The third consequence is opportunity cost.

Every dollar spent on AI infrastructure is a dollar not used for buybacks, dividends, acquisitions or other projects. Meta paid $1.35 billion in dividends and dividend equivalents in Q1, but the scale of capex means capital returns may become less central to the story. 

The fourth consequence is execution risk.

Building AI infrastructure at this scale is not just a finance decision. It involves supply chains, chip availability, data center construction, energy access, model efficiency, product integration and regulatory exposure. Meta also flagged ongoing legal and regulatory headwinds in the EU and U.S., including youth-related issues that could result in material losses. 

The opportunities are just as large

The most obvious opportunity is advertising.

If Meta can keep improving ad relevance, creative generation and conversion measurement, AI can increase both ad load efficiency and ad pricing. The Q1 numbers already point in that direction: impressions up 19%, average price per ad up 12%. That is a powerful combination. 

The second opportunity is business messaging.

WhatsApp, Messenger and Instagram DMs could become major AI agent platforms. Instead of businesses using Meta only to advertise, they could use Meta to acquire, serve and transact with customers inside conversations. If this works, Meta’s revenue model broadens from advertising into commerce enablement, customer support automation and enterprise tools.

The third opportunity is consumer AI distribution.

Meta has something most AI startups do not: distribution to billions of people. If Meta AI becomes deeply embedded across Instagram, WhatsApp, Facebook and smart glasses, Meta does not need to win the AI race only through standalone apps. It can win by making AI a default layer inside products people already use every day.

The fourth opportunity is hardware.

Reality Labs remains expensive, but smart glasses are strategically more interesting in an AI world than they were in a pure metaverse narrative. If AI assistants become ambient and visual, glasses could become a natural interface. That said, Reality Labs still needs discipline. The opportunity is real, but the losses cannot be ignored.

The key question: is Meta building a moat or buying optionality?

This is the central debate.

A positive interpretation is that Meta is building a deeper moat. AI infrastructure gives it better models, better ranking, better recommendations, better ad tools, better messaging automation and eventually new consumer hardware experiences. Under this view, today’s capex protects and expands one of the most profitable advertising ecosystems in history.

A more cautious interpretation is that Meta is buying expensive optionality in a race where the winners are not yet clear. Unlike Microsoft, Amazon or Google, Meta does not have a large public cloud business to directly monetize spare AI capacity. Its return on infrastructure must come mostly through its own apps, ads, messaging products, AI assistants and devices.

That makes Meta’s AI capex more internally leveraged. If the products work, the upside is enormous. If monetization lags, the market will question whether the company overbuilt.

My take

Meta’s latest earnings were not weak. They were strong.

But the company has entered a new phase where the quality of the business will be judged less by quarterly revenue growth and more by capital productivity.

The old Meta question was: “Can the company keep growing engagement and ads?”

The new Meta question is: “Can the company turn massive AI infrastructure spending into returns that justify the capital intensity?”

That is a much harder question.

The opportunity is huge: better ads, AI agents, business messaging, consumer assistants, smart glasses and stronger platform lock-in.

The risk is equally clear: lower free cash flow, higher depreciation, rising debt usage, regulatory pressure and the possibility that AI infrastructure becomes a cost of staying competitive rather than a source of differentiated returns.

For now, Meta has earned some trust because the core business is performing. But with capex now guided to $125 billion to $145 billion for 2026, the market is right to demand evidence.

AI is no longer a slide in Meta’s strategy deck.

It is becoming the company’s largest capital allocation decision.