Meta Platforms, Inc.
Why it could work, how I could lose money, and what to watch — reopenable in six months.
- company
- Meta Platforms, Inc. (Nasdaq: META)
- sources used
- FY2025 10-K (filed 29 Jan 2026); Q2 2026 10-Q (filed 30 Jul 2026); Q2 2026 earnings call transcript (29 Jul 2026); Q2 2026 follow-up call transcript (29 Jul 2026). FY2023–FY2024 comparatives taken from the FY2025 10-K.
- latest period
- Q2 2026 — quarter ended 30 June 2026
- missing or stale data
- Ad load level and remaining headroom: not disclosed. Return metric on AI infrastructure: not disclosed; management declined a timeline. Maintenance-versus-growth split of capex: not disclosed. The MDL state-AG trial (began 12 Aug 2026) and the expanded New Mexico trial (begins 8 Sep 2026) post-date the latest filing in sources, so their outcomes are unknown here. No market or valuation data is used; this memo takes no view on price.
SECTION 01Business in one line
What it does, how it makes money, and which type it is.
Meta sells auction-priced advertising against the attention of 3.60 billion daily active people across Facebook, Instagram, WhatsApp and Messenger. Advertising was $196.18bn of FY2025's $200.97bn of revenue — 97.6% [FY2025 10-K, share computed]. Revenue is impressions multiplied by price per ad, sold in a real-time auction with no contracts and no backlog. Reality Labs, the second segment, produced $2.21bn of FY2025 revenue and lost $19.19bn [FY2025 10-K].
slow growerstalwartfast growercyclicalturnaroundasset play
Fast grower — revenue grew 30% in H1 2026 [Q2 2026 10-Q] — that has just bolted a heavy-industry capital cycle onto itself, with FY2026 capex guided to $130–145bn against FY2025 revenue of $200.97bn [Q2 2026 call; FY2025 10-K].
That matters because a fast grower is judged on the durability of its growth rate and a cyclical on where it sits in the capacity cycle. Meta is now both, and the clocks are not synchronised: the leases run 30 years, the auction reprices every quarter.
SECTIONS 02 & 03Bull case vs. Bear case
The Peter Lynch pitch against the Munger invert. Side by side puts each load-bearing assumption next to the attack on it; the toggle reads either case alone.
The simple reason this stock could work is that Meta is growing revenue roughly nine times faster than its user base, and the machine doing it is software Meta owns outright.
The most likely way I lose money is that Meta signs twenty- and thirty-year fixed obligations against a revenue stream that reprices every quarter, and the auction turns before the concrete does.
In Q2 2026 advertising revenue rose 27% while Family daily active people rose 3% [Q2 2026 10-Q]. The gap is ranking quality, ad load and new surfaces — internal levers, none dependent on taking users from a rival.
Why the market might be missing it: margin and free cash flow are the two numbers a screen reads, and both are collapsing mechanically: capex hits the cash flow statement in full the moment it is spent, while $80.35bn of it produces neither depreciation nor revenue yet [Q2 2026 10-Q]. If the ad gains are real — and price accelerating from 9% to 12% against adverse mix says they are — Q2's $784m of free cash flow is a construction schedule, not an earnings statement.
SECTION 04Signals to monitor
Every signal traces to a named assumption or risk above. Click a tag to jump to it.
SECTION 05Bottom line
Four sentences: why it works, what must go right, how it breaks, what changes my mind.
Meta could work because it converts a near-flat user base into revenue at roughly nine times the rate that base grows, using an ad system it owns end to end — price per ad rose 12% in Q2 2026 against 9% a year earlier, despite adverse mix [Q2 2026 10-Q].
What must go right is that price per ad stays in double digits while the $80.35bn now in construction in progress begins producing revenue before it begins producing depreciation [Q2 2026 10-Q].
The thesis most likely breaks not through the ad business failing but through its cost base becoming fixed and long-dated ahead of it — $278.99bn of leases not yet commenced with terms up to 30 years, plus $349.31bn of non-cancelable commitments, against auction revenue that reprices quarterly [Q2 2026 10-Q].
I change my mind if price per ad decelerates to mid-single digits while impressions keep growing, if Family of Apps operating margin falls below the mid-30s, or if a court orders engagement-limiting product changes rather than a penalty — and, on the other side, if buybacks resume at scale with revenue growth above 20%, which reduces the bear case to an argument about timing.
APPENDIXSources & disclosure gaps
Every figure above is stated in one of these documents, or derived from them and marked (computed).