Meta reports Q2 2026 earnings after the close today. And here is what is strange about it: the revenue beat is almost certain. The stock reaction is not.
That gap between expected fundamentals and uncertain price action is the actual trade.
Where the bar sits right now
Consensus expects Q2 revenue of about $60.2 billion, up roughly 27% year over year, and operating profit of $21.5 billion. Revenue is expected to grow roughly 26.6% year-over-year, driven by continued strength in digital advertising, though the pace may moderate from Q1’s 33% surge.
The ad machine is not the problem. It has not been for a while.
Meta’s advertising division generated $55 billion in revenue in Q1 2026, with ad impressions up 19% and the price per ad up 12% in the same quarter. That kind of combination, volume and pricing moving together, is rare. It means the underlying demand is real, not just inventory-driven.
But the market is not pricing Meta like an ad company anymore. The market is pricing Meta as an AI infrastructure growth stock. That is why capex guidance matters more than the EPS beat.
The $125 billion question
AI-related capital expenditures remain the dominant investor concern, with full-year 2026 capex guidance already raised to a range of $125 billion to $145 billion. BofA expects capex guidance to be updated to $135 billion to $150 billion, after reports of a potential compute deal with Anthropic that could add $5 billion in incremental AI revenues.
That Anthropic angle is the hidden catalyst tonight. Meta is currently the only major hyperscaler without a cloud business it can point to as proof its AI spending earns money outside advertising. If Meta announced any actual cloud customers tonight, that would be huge news, and given the market’s past reaction to third-party reports alone, it could boost the stock.
The problem is the history. Meta beat on both revenue and EPS last quarter and still fell 6-7% after hours purely on the capex number. A beat alone has not moved this stock higher in either of its last two reports.
So the setup is: great ad business, rising costs, a new AI model nobody fully understands yet, and a stock that has punished beats before.
The Muse Spark angle
This is where the longer-term story actually lives. On July 9, Meta released developer access to its Muse Spark 1.1 model, and for the first time Meta is charging for model usage, with pricing of $1.25 per million input tokens and $4.25 per million output tokens.
That is new. Meta built its entire AI strategy around open-source models and free distribution for years. Charging for API access is a different business entirely. With Muse Spark, Meta is exploring enterprise API licensing, potentially diversifying its revenue away from pure advertising for the first time in its public history.
Muse Spark powers AI shopping recommendations and ad targeting improvements, not user subscriptions, reflecting how Meta monetizes through advertising and commerce rather than charging consumers directly. But the API business changes the equation for enterprise clients, and that story is just getting started.
The valuation reality
Meta stock has fallen approximately 9.7% year-to-date and trades roughly 25% below its 52-week high of $796.25, reflecting anxiety over AI spending returns.
The bull case rests on real fundamentals: roughly 27% expected revenue growth, ad pricing power that has not cracked, and a stock trading at approximately 18 to 22 times forward earnings, below its own three-year average of about 23 times.
That is a compelling setup if you believe the capex eventually converts into durable revenue streams. The bear case is that the timeline is long and the spending keeps rising. Combined Big Three hyperscaler capex is on track to hit $801 billion in 2027, enough to push all three negative on free cash flow through 2028.
The ad business is fine. The AI bet is expensive and unproven. Tonight, we find out how much patience the market still has.
The Q3 guidance tone may matter more than anything else in the release.
