Meta’s Ads Grew 26%. Why Did Free Cash Flow Turn Negative?

Family daily active people (DAP) reached 3.64 billion, up 3% year over year. Ad impressions rose 10%, while the average price per ad increased 14%. [1] [2] [4]

Advertising revenue does not simply track the number of users. How often people see ads and the price advertisers pay per impression also matter. [1] [2] [4]

Impression growth slowed from 14% in Q2 to 10% in Q3, while average ad-price growth accelerated from 12% to 14%. Both are year-over-year growth rates; the comparison does not show a sequential decline or increase in the underlying price or impression levels. [1] [2] [4]

CFO Susan Li attributed ad-price growth to better ad performance and improved macroeconomic conditions. Strong impression growth in regions and surfaces with lower monetization partly offset the increase in average price. [1] [2] [4]

Growth therefore depends on more than adding users. Meta’s ability to deliver more effective ads to its existing audience and create greater value for advertisers has become increasingly important. [1] [2] [4]

Meta described examples of AI contributing to improvements in its existing advertising business. [1] [4]

Management said the expansion of GEM increased conversions by 1.7% for off-site conversion campaigns and by 9.6% for app ads. Improvements to Andromeda increased conversions on Facebook by 1.4% within the reported scope. [1] [4]

Nearly 10 million advertisers used at least one of Meta’s generative AI ad-creation tools during the quarter. [1] [4]

These examples show AI being used to improve the existing advertising business, alongside investment in new services. [1] [4]

The percentages should not be added together or used to calculate AI’s contribution to total advertising revenue. They relate to different campaigns and applications, and do not imply an equivalent improvement in company-wide average return on ad spend (ROAS). [1] [4]

The clearest economic evidence so far is AI’s contribution to recommendations, conversions and monetization in the existing advertising business. Revenue from new AI services needs separate verification. [1] [4]

Q3 GAAP operating income fell 36.8% to $12.988B. Operating margin declined from 40.1% to 20.1%. [1] [3] [4]

The quarter included a $9.89B litigation charge related to an agreement with U.S. state attorneys general. A simple exclusion of that charge gives the following comparison. [1] [3] [4]

This analytical adjustment separates the effect of the settlement-related charge. It does not replace reported GAAP operating income. [1] [3] [4]

It would therefore be inaccurate to attribute the entire decline in reported operating income to weaker advertising competitiveness. [1] [3] [4]

Excluding the litigation charge does not eliminate the wider cost burden. Meta reported increases in data-center operating costs, depreciation, third-party cloud spending and third-party AI token costs. [1] [3] [4]

Net income also requires care. Q3 2026 net income increased 264% to $9.860B, but the year-ago quarter included a large one-time tax charge related to U.S. tax legislation. [1] [3] [4]

The 264% increase in net income should not be treated as the underlying growth rate of the advertising business. [1] [3] [4]

Meta generated $32.786B of operating cash flow in Q3, an increase of $2.787B from $29.999B a year earlier. [1] [2] [3] [4]

Capital expenditure, including finance-lease principal payments, rose by $15.532B, from $19.374B to $34.906B. [1] [2] [3] [4]

Cash generated by the business increased, but cash spent on expanding infrastructure grew much faster. [1] [2] [3] [4]

Negative FCF is not the same as an operating loss. Meta generated substantial operating cash, but spending on property and equipment and finance-lease principal exceeded it. [1] [2] [3] [4]

Meta narrowed its full-year 2026 CapEx outlook from $130–145B to $135–145B, keeping the upper end unchanged and raising the lower end by $5B. [1] [2] [3] [4]

The amount invested is only part of the question. Investors also need to track how added servers and data centers translate into usage and revenue, and how much cash remains after the investment. [1] [2] [3] [4]

Meta said more than 10 million people in the U.S. and Canada had set up Muse personal agents in roughly six weeks after launch. [1] [4]

Management also said about 25% of those who had set up Muse used it on a given day. [1] [4]

Potential monetization routes include consumer and business subscriptions, transaction-related fees, APIs and enterprise services. [1] [4]

Meta did not disclose a separate Muse revenue figure in its Q3 results. [1] [4]

User adoption and a profitable business model are different milestones. [1] [4]

CEO Mark Zuckerberg said Muse was compute-intensive, making improved cost efficiency important for scaling the service. He expected cost efficiency to improve by more than tenfold over the following year; this was a forward-looking expectation, not an achieved result. [1] [4]

Assessing Muse’s economics requires evidence on paid conversion, recognized revenue and compute cost per user. [1] [4]

AI-led improvements in the existing advertising business should be distinguished from potential future revenue from new AI services. [1] [4]