Meta Platforms, Inc. META Moat

Key data

The moat

An advertiser in Jakarta or Sao Paulo pays Meta because the auction returns a measurable result per dollar that nobody else can match at that budget size, and because the social graph behind it cannot be rented elsewhere. A user stays because leaving means leaving the group chats, the photo archive, and the people. That is a network moat, primary classification, high confidence, with a secondary proprietary-tech layer in the ranking and targeting stack. Swap test: Alphabet owns intent, not the friend graph; Amazon owns purchase data, not the conversation; TikTok owns a recommendation engine but not the contact list.

What the moat produces is pricing power inside the auction. The FY2024 filing records average price per ad up 10% in 2024 after a 9% decline in 2023, with impressions up 11% versus 28% the year before. Meta sold fewer incremental impressions and still took price, which is what a working auction looks like when demand does the lifting. Revenue per employee moved 1,348 to 2,004 to 2,221 to 2,548 across 2022 to 2025, a four-period sequence on filed headcount and filed revenue, and the 2023 jump is partly the layoff year, headcount falling from 86,482 to 67,317. Call that inflection a cost action, not an organic efficiency gain.

Widening or narrowing

Gross margin runs 78.3%, 80.8%, 81.7%, 82.0% across FY2022 to FY2025, a four-period climb while revenue grew 72% over the same span. The nearest large peer on the vendor list, Alphabet, sits at 56.6%, 58.2%, 59.7% for 2023 to 2025. A roughly 22-point gross margin spread that has held or widened for three straight years is the hardest evidence here, and it is a measured spread between two filings.

Meta holds the fat slice of the social advertising layer but owns almost none of the layer below it. Capital expenditure including finance lease payments ran $31.1 billion in a single quarter per the July 2026 call, and R&D went 35.34 to 38.48 to 43.87 to 57.37 billion dollars across 2022 to 2025. The silicon and power vendors are capturing a rising share of every incremental dollar of ad performance.

The overrated case, stated plainly: on third-party measurement, Meta is projected at 20.9% of US digital ad spend in 2026 against Alphabet at 23.9% and Amazon at 17.3%, with Amazon the fastest riser. ROIC also broke its climb, 14.6% to 19.3% to 25.1% then down to 18.0% in 2025, as the capital base absorbed data centers faster than returns arrived. Reality Labs revenue has gone 2.27, 2.16, 1.90, 2.15, 2.21 billion dollars across five years, which is five periods of no growth on a filed line, so the distribution hedge is not working yet. Verdict: stable.

What breaks it, and who

Amazon is the live threat, not TikTok. At 17.3% of US digital ad spend on third-party projections, Amazon converts closed-loop purchase data into performance claims Meta can only infer, and it is taking the lower-funnel budgets that Meta's auction depends on.

The FY2025 filing names the DMA, DSA, the EU AI Act, the UK DMCC and an active FTC consent order, and flags the risk that invalidation of EU-US data transfer bases could stop Facebook and Instagram operating in Europe. Europe was $46.57 billion of 2025 revenue, 23% of the total.

The third pressure is open source, partly self-inflicted: Meta publishes its Llama models, which compresses the proprietary-tech layer it is spending $57.37 billion a year of R&D to build.

The read changes on three things in the next 12 to 18 months: whether average price per ad holds positive for a third consecutive year, whether Family of Apps other revenue past the $1 billion quarterly mark keeps compounding near 73%, and whether any EU data-transfer ruling lands.

Closing thoughts

The moat is real and currently stable, with the auction-performance layer widening and the distribution layer narrowing. The weakness is that the strongest forward claims, Meta One subscriptions, AI glasses, business agents on WhatsApp, rest on company statements from the July 2026 call rather than measured results, so they are asserted, not proven. The one checkable thing is average price per ad in the next annual filing, because that single line separates a network taking price from a network selling more inventory cheaper. The moat strengthens if average price per ad posts a third consecutive annual increase alongside rising impressions and weakens if price turns negative again while capital expenditure keeps climbing.