Everything About AI
NextFin News - Meta Platforms reports third-quarter results after the U.S. market close on Wednesday, October 21, 2026, with the 4:30 p.m. ET earnings call set to deliver the clearest verdict yet on whether the company's $130 billion-to-$145 billion artificial-intelligence spending bet can pay off before investors lose patience. The quarter itself is almost a sideshow: Wall Street expects revenue of about $63.3 billion, up roughly 24% from a year earlier and near the top of management's own $61 billion-to-$64 billion guide, with earnings per share consensus estimates clustered between $6.54 and $6.77. What will move the stock is not whether Meta beats another ad-revenue bar, but whether CEO Mark Zuckerberg and CFO Susan Li can convince the market that free cash flow - which collapsed to just $784 million last quarter - is about to recover rather than deteriorate further.
The Quarter That Decides the AI Bet
Meta Platforms trades at $704.50, up 1.48% on the day, with a market capitalization of about $1.79 trillion. The stock has gained roughly 6% over the past month on a wave of AI product announcements, yet it has fallen more than 3% over the past year and has given back ground from its 52-week high of $779.82. That divergence captures the tension investors are bringing into the print: the company's advertising engine is firing, but its balance sheet is being asked to fund a second industrial revolution.
The setup is unusually well defined. Management already told investors to expect third-quarter revenue between $61 billion and $64 billion, and the consensus of $63.3 billion sits in the upper half of that range - a 23.5% year-over-year increase from the $51.24 billion reported in the third quarter of 2025. On the bottom line, analysts tracking the company expect earnings per share between $6.54 and $6.77. In other words, the headline numbers are largely pre-sold. The options market agrees: as of Tuesday, contracts were pricing an implied post-earnings move of about 6.9% in either direction.
The real question is what comes after the headline. Meta has guided full-year 2026 capital expenditures to a range of $130 billion to $145 billion, and the market wants to know whether that number holds, rises, or - crucially - shows any sign of peaking into 2027. Some analysts project 2027 spending could climb toward $193 billion, with Bank of America estimating as high as $205 billion. At those levels, Meta is spending more on AI infrastructure in two years than most companies spend in a decade.
The cost is already visible in the cash-flow statement. In the second quarter, capital expenditures reached $31.1 billion, leaving just $784 million in free cash flow - down from $8.55 billion a year earlier, a decline of roughly 91%. Operating margin compressed to 31% from 43% over the same stretch. For the third quarter alone, analysts expect capital expenditures of about $41 billion, up 117% year over year, alongside an operating loss of roughly $5.2 billion in the Reality Labs division that builds the company's virtual-reality hardware.
Against that backdrop, management's commentary will carry more weight than the reported figures. Investors are watching for three things: whether AI-driven ad targeting and recommendation systems continue to lift ad pricing, whether the Muse AI assistant is converting into a real revenue line, and whether 2027 spending guidance signals a peak or another leg higher. The answers will set the tone not just for Meta but for the entire AI infrastructure trade across the market.
The Spending Wall: $145 Billion and Counting
Meta's capital-spending plan is not a budget line; it is a strategic declaration. A $130 billion-to-$145 billion annual commitment puts the company in the same spending tier as the largest cloud providers, and it reflects a bet that owning the full AI stack - from chips and data centers to models and consumer applications - is the only way to avoid being disintermediated in the next era of computing. Zuckerberg has framed the choice in stark terms: spend now at a scale that strains near-term profitability, or risk becoming a content supplier to whoever controls the AI layer.
The mechanism is straightforward but expensive. Every dollar of AI infrastructure spending has to show up somewhere as revenue, or it shows up as margin compression. In the near term, the spending flows through depreciation and operating expenses, which is exactly why free cash flow fell to $784 million in the second quarter even as revenue grew. The bullish reading is that this is a timing problem: the data centers and GPU clusters being installed today will serve inference demand for years, and the marginal cost of serving an additional AI query declines as utilization rises. The bearish reading is simpler - the spending may never earn its cost of capital if monetization lags.
There is evidence on both sides. Meta's Family of Apps "other" revenue - which includes paid messaging on WhatsApp and subscriptions - reached $1 billion in the second quarter, up 73% year over year. That is still a small fraction of the advertising business, but it is the first concrete sign that users will pay Meta directly for AI-enhanced services. The Muse assistant, launched in September, reached 10 million downloads within 40 days of its U.S. debut, data from market-intelligence firm Sensor Tower show, and a Wells Fargo research note put the product at more than 2 million daily active users.
"Meta Muse Wonder reaches AI frontier, outperforming Fable and top models from OpenAI & Google, second only to Opus 5.5 on Artificial Analysis Intelligence Index," Wells Fargo analyst Ken Gawrelski wrote as the company rolled out its frontier-class Muse Wonder and Muse Wonder Full models.
The pricing architecture - a free tier alongside $20 and $100 monthly plans - gives Meta a direct read on willingness to pay, something the advertising model has never provided. If even a low-single-digit percentage of the company's billions of users convert to paid tiers, the revenue base would diversify meaningfully. But conversion is the hinge, and it is too early to know it.
The Cash-Flow Squeeze Is Real - But Is It the Point?
Here is the number that keeps portfolio managers awake: free cash flow down 91% year over year. It is easy to read that figure as a red flag, and for a value-oriented investor it would be. But Meta is not in a value phase; it is in a build phase, and the appropriate comparison is not to its own prior cash generation but to the capital intensity of the AI race itself.
The structural argument runs like this: the company that controls the cheapest, highest-quality inference at scale wins the platform layer. GPUs, data centers, and power contracts are lumpy, front-loaded investments - you build the capacity before the demand arrives, which means cash flow troughs before it recovers. Under this view, the $784 million second-quarter print is the bottom of the trough, and every subsequent quarter should show operating leverage as revenue grows into the installed base. Operating margin compression from 43% to 31% is the price of admission.
The cyclical counter-argument is that advertising demand is itself cyclical, and a downturn in ad spending would hit revenue just as the new capacity comes online - the worst possible combination. That is the scenario that turns a strategic build-out into a capital trap. It is also why the company's guidance for 2027 matters more than the third-quarter print: if management signals that spending will keep climbing toward $200 billion without a corresponding monetization roadmap, the market will re-rate the multiple downward even if revenue beats.
This is where the market's pricing discipline shows up. Meta's stock trades at a forward price-to-earnings ratio of more than 80 times on the next-period consensus, a premium that assumes both continued ad growth and a credible path to AI monetization. A premium multiple cannot survive indefinitely on promise alone. At some point, free cash flow has to re-accelerate, or the multiple compresses.
Muse and the Search for a Second Engine
Advertising remains the profit foundation, and it is the reason Meta can afford to spend $145 billion without threatening its solvency. The company ended the second quarter with substantial cash reserves and access to debt markets, and its advertising business continues to grow on the back of AI-driven targeting and recommendation improvements. But the strategic imperative is to build something that does not depend on advertisers' budgets.
Muse is that attempt. It is an agentic AI assistant - capable of sending emails, booking travel, and shopping - and Meta is embedding it across its apps and hardware, including the Ray-Ban Meta glasses unveiled at the company's Connect conference. The hardware lineup, including audio-only Ray-Ban Meta glasses and a forthcoming $1,299 Meta VR Glasses product, is designed to give the assistant a presence beyond the phone screen.
The commercialization test is whether Muse can move from a cost center to a revenue line within the next few quarters. The $1 billion in "other" revenue is a start, but it is not yet material relative to the $63 billion quarterly revenue base. Investors will listen closely for any disclosure of paid-subscription uptake, enterprise-agent adoption, or API monetization. RBC analyst Brad Erickson framed the question directly, noting that his team sees business agents as the company's top near-term monetization opportunity outside advertising and that he would be looking for color on how business-agent adoption is progressing after the company began monetizing earlier in the quarter.
There is also a defensive dimension. If AI assistants become the primary interface through which users discover content and services, the company that owns the assistant owns the gateway. Meta's argument is that its billions of users across Facebook, Instagram, WhatsApp, and Threads give it a distribution advantage that standalone AI labs cannot match. Whether that distribution converts into paid usage is the unanswered question.
The Counter-Thesis: What If the Payoff Never Arrives?
The strongest case against Meta is not that AI is a fad - it is that the returns to AI infrastructure may accrue to chip suppliers and cloud providers, not to the application layer. In this view, Meta is engaged in a capital-intensity arms race where every participant spends more but no one captures disproportionate profit. The beneficiaries would be the companies selling GPUs, networking equipment, and power infrastructure, while Meta's margins compress permanently.
There is institutional support for this skepticism. JPMorgan, which maintains an overweight rating and a $920 price target, recently cut its 2026 and 2027 earnings-per-share forecasts by 10% and 2% respectively, citing an expected $10 billion legal accrual tied to a settlement with 52 state attorneys general over the impact of the company's apps on children. The bank's analysts noted the fee may not be fully reflected in the company's guidance - meaning the reported quarter could look better than the underlying economics.
Deutsche Bank analyst Benjamin Black offered the bullish counter to that view: "We think we are likely at the start of a rich product cycle that will foreshadow the potential returns to the significant infrastructure and R&D investment Meta has made from AI agents, enterprise, and advertising." A Morgan Stanley research note similarly argued that recent AI launches have strengthened confidence in the company's AI execution and that Meta carries the least risk of AI disruption to its core business among the companies the bank covers.
Both sides can be right at different points in the cycle. The bull case requires patience through a cash-flow trough; the bear case requires only one thing - proof that spending keeps rising without monetization. That asymmetry is what makes the 2027 guidance the single most important number of the call.
What to Watch - and What Would Prove the Thesis Wrong
The central judgment: Meta's AI spending surge is a structural regime shift, not a cyclical overspend. The company has committed to a level of infrastructure investment that will not self-correct, because the strategic cost of falling behind in AI exceeds the cost of the capital itself. The free-cash-flow compression is the cyclical leg of the same story - a trough that should ease as revenue grows into installed capacity, provided advertising demand holds.
This distinction matters for how to read the stock. Short term, the shares are a volatility instrument: the options market prices a roughly 7% move, and the direction will hinge on 2027 spending signals and any commentary about the $10 billion legal accrual. Medium term, the stock is a cash-flow story - investors need to see free cash flow re-accelerate as the capital base gets utilized. Long term, it is a platform bet: if Muse and the agent ecosystem become a meaningful second revenue engine, today's spending looks cheap; if they do not, the multiple has further to fall.
Three scenarios frame the next twelve months. In the base case, revenue lands near $63 billion, management reaffirms the $130 billion-to-$145 billion capex range without raising 2027, and the stock trades in a range as investors wait for proof of monetization. In the upside case, Meta signals that 2027 spending will grow more slowly than the $193 billion some analysts project while Muse paid-tier adoption exceeds expectations - that combination could drive the shares toward the $900-to-$1,000 price-target cluster held by JPMorgan and Wells Fargo. In the downside case, 2027 capex guidance rises again and free cash flow stays negative, which would likely push the stock back toward its 50-day average near $659 and test the $600 support zone.
The falsifying signal is specific: if management guides 2027 capital expenditures materially above $200 billion while "other" revenue remains below a 5% share of total revenue, the structural-payoff thesis is wrong and the cash-flow trough is deeper and longer than the bull case assumes. Conversely, if paid Muse adoption inflects and capex growth decelerates, the bear case breaks.
For the broader market, Meta is the first hyperscaler to report this earnings season, so its tone on AI spending will set the template for how investors assess the entire infrastructure trade. A confident, peak-spending signal lifts the sector; a "spend more, ask later" posture spreads the skepticism.
Meta's bet is that the company which owns the AI stack owns the next decade. The market's reply is simpler: show us the cash flow. Until then, every dollar of guidance is a down payment on credibility, and the bill for $145 billion comes due one quarter at a time.
Explore more exclusive insights at nextfin.ai.
