Meta runs free apps that sell attention to advertisers
Facebook, Instagram
WhatsApp, Messenger and Threads
3.60 billion
people use them each day
99.3%
of revenue from adverts
Meta's apps are free to use. It earns by showing adverts, sold in a real-time auction to businesses that want to reach the people using them.
The company · 2 · What exactly does it sell?
Almost every dollar is advertising
Each $1 of sales, second quarter of 2026
98%2%1%
Adverts in its appsOther app revenueHeadsets and glasses
Reality Labs, its headsets and smart glasses, is under 1% of sales and loses billions a year.
The company · 3 · How does it make money?
More adverts, at higher prices, at once
+14%price per advert last quarter
38.1%operating profit per $1, highest of its rivals
$130B to $145Bdata-centre spending this year
Its AI predicts which advert each person will act on, so advertisers get more back per dollar and pay more. Running that AI needs vast data centres.
Is it a good business? · 4 · What gives it an edge?
Measured attention at a scale nobody can copy
3.60 billion
people a day
+12%
price per advert
Advertisers go wherever each dollar earns most, and Meta's predictions earn more per dollar than most rivals.
Is it a good business? · 5 · What is the catch?
TikTok and YouTube sell the same hours
$294.7BGoogle's advertising revenue
$201.0BMeta's
TikTokcompetes for the same hours, a judge ruled
Listed rivals: GOOGL · SNAP
Alphabet is the only listed company selling advertising at Meta's scale.
Is it a good business? · 6 · What does its record show?
Growth now costs more than it used to
Sales, $B
117.9
116.6
134.9
164.5
201.0
241.6
Dec 2021Dec 2022Dec 2023Dec 2024Dec 2025last 12 mo
Sales$201.0Bsales in 2025
Profit per $1Peaked in 2024profit per $1; research now rises faster
Commitments$454.5Bof purchases it cannot cancel
Selling and office costs keep shrinking per dollar, but research rises faster than sales, and data-centre spending has doubled as a share of sales.
Is it a good business? · 7 · How much room is there to grow?
Social media overtook search in US advertising
$294.6BUS internet advertising, 2025
+32.6%social media adverts, to $117.7B
Meta's own advertising is booked worldwide, so its share of this American market cannot be read off.
Is it a good business? · 8 · Who is in charge?
The founder controls most of the vote
2004Zuckerberg started Meta
13.5%of the company he owns
$1his salary
Through ten-vote shares Zuckerberg controls most of the vote, so outside shareholders cannot overrule him.
The story right now · 9 · Why is Meta spending so much?
Meta is building data centres for AI
$130B to $145Bplanned spending this year
-$2.1Bcash left after it, last quarter
Meta is spending on computers to run AI for its adverts and new products, such as assistants and smart glasses. Investors doubt it will all pay back, and Meta gives no figure for AI product revenue yet.
The story right now · 10 · Why is cash short while profit is fine?
Building now, cost counted later
Meta pays for a data centreCash leaves today.
Accounts spread the costOver its working life.
Profit looks fine todayCash after spending turns negative.
Later years carry the costProfit per $1 falls.
That yearly slice is depreciation. Our value lets it pull operating profit to about 34 cents per $1, and judges Meta on cash left after spending.
The story right now · 11 · What changed lately?
Sales grew 26%; a legal charge cut profit
+26%sales, to $64.6B
+14%ad prices
$9.89Bcharge settling 52 states' claims
About $1.5Bnet cash left
Mixed quarter
The advert business passed its test, but cash after spending turned negative and legal costs stepped up. Management forecasts $73B to $76B of sales next quarter.
The story right now · 12 · What is everyone arguing about?
Will the data-centre build pay back?
Many investors think
The spending is mostly money burned
We think
The advert engine is speeding up; the build is an asset
Against us: no revenue from the spending is disclosed, 2027 spending is not forecast, and debt is rising.
The price and our view · 13 · What is a share worth?
The price sits about a tenth below our value
$694what we think a share is worth · the last close is of that
$694our value
Attractive below $555
Neutral
Unattractive above $832
$358gloomy: the build earns little, growth falls to 8%
$694our estimate: growth slows as the base grows
$968sunny: AI assistants and glasses add revenue
Value per share in three futures. None is a prediction of the share price.
We estimate what a share is worth from the cash the business should earn over many years. Our rule: buy only when the price is at least a fifth below our estimate, as a cushion for being wrong.
The price and our view · 14 · How could we be wrong?
A build that may not pay back
Spending that never paysBiggest
No revenue from it has been disclosed.
Debt and new sharesShowing now
Net cash fell from $22.9B.
A hidden commitmentWatch
Up to $45.95B off the balance sheet.
What the spending earnsUnknown
Meta discloses no return.
Daily users grew 3% against 7% six months earlier, and more trials over young users are due.
The price and our view · 15 · How will we know?
Early 2027 shows whether the build earns
Sales of $76B or more, cash after spending positiveGood sign
Spending above $145B, or negative cash againBad sign
2027 spending above 2026's, no AI revenue shownWe turn negative
JAN 2027-
Fourth-quarter 2026 results, January. Companies report every three months; each report is a test of the story.
Our view also changes on price alone: we turn positive if the shares fall a fifth below our estimate, and negative if they rise a fifth above it.
The price and our view · 16 · So, what do we think?
A strengthening advert engine, fairly priced against its spending
$694our value
Attractive below $555
Neutral
Unattractive above $832
NeutralLow conviction
Meta shows more adverts and charges more for each at once. On profit it is the cheapest of its rivals, but on cash left after its data-centre spending it is about fair. Neutral, with low conviction.
↑ ↓ to move · C for contents
What it does
More about the company
Who it is
Meta runs Facebook, Instagram, WhatsApp, Messenger and Threads, used by 3.60 billion people a day. It charges users nothing. Advertisers pay to put a message in front of a chosen audience, and increasingly pay for a result, such as an install or a purchase, rather than for being seen.
Advertising is 99.3% of revenue. The rest is Reality Labs, which sells headsets and AI glasses at a large loss, paid for by the apps.
How it makes money
Every advertisement is sold by auction, in real time, so no advertiser is committed beyond the next one. Meta earns more by showing more advertisements and by charging more for each, and it did both in the June quarter.
Growth and profit
What growth costs
Growth used to make Meta more profitable, because selling and admin costs barely rise with sales. That has turned: research spending, which now carries AI staff and computing, is rising faster than sales, and the money spent on data centres has doubled as a share of sales and will reach profits slowly as wear-and-tear charges. Profit margins peaked in 2024.
Its record
The numbers
Is it growing?Fast
Sales grew 28% over the twelve months to Sep 2026, faster than the 22% in the year to Dec 2025.
Is it profitable?Strong
Operating margin is 33%, about where it was (35%) in the year to Dec 2023.
Does growth pay off?Costly
Running costs take 48¢ of each sales dollar, up from 41¢ in the year to Dec 2021.
Does it earn good returns?Strong
Each dollar put into the business earns 26% a year after tax, about where it was (31%) in the year to Dec 2021.
Is it making cash?Solid
It turned 12% of sales into free cash, $28.6B over the last twelve months, after spending 43% of sales on buildings and equipment.
Is the balance sheet safe?Light debt
It owes $1.22B more than the cash it holds, about 0.0 years of the cash its operations bring in.
Are shares being diluted?Shrinking
The share count fell 10% in 4.8 years.
Why it wins
The evidence
Scale nobody can assemble twice. More users produce more measured sales, more measured sales train a better prediction of who will buy, and better predictions return more for each advertising dollar. Advertisers have no loyalty; they follow the return.
Users can leave, and a federal court said so. What an advertiser cannot find elsewhere is this much measured attention in one place. The proof is price: Meta charged 12% more per advertisement while earning the highest operating margin of its rivals, 38.1%.
Who could beat it
The rivals in full
TikTok and YouTube. In 2025 a federal judge ruled that they compete for the same hours Facebook and Instagram sell. Alphabet, YouTube's owner, is the only listed company selling advertising at Meta's scale, with $294.7B of Google advertising revenue against Meta's $201.0B.
Each rival
TikTok, operated by ByteDanceprivate
The session itself. A United States District Court, after a six-week bench trial, found that TikTok and YouTube are competitors for the same user time that Facebook and Instagram sell, and that the distinction between social networking and social media has collapsed - 'that wall has since broken down'. The buyer who stops opening Instagram to watch short video is the substitution, and the advertiser's budget follows the attention rather than the brand.
YouTube, operated by AlphabetGOOGL
The same session, and then the same advertiser budget. Alphabet is the only listed company selling advertising against owned consumer attention at comparable scale, with Google advertising revenue of $294,691M against Meta's $200,966M in the latest full year, and the court named YouTube specifically as a substitute for the time Meta's apps sell.
OpenAI, through ChatGPTprivate
The discovery step, and potentially the surface. Pinterest's own annual report lists 'OpenAI (including ChatGPT)' among the competitors it faces for user time and commerce opportunity, alongside Meta and Google. A conversational assistant that answers the question a person would previously have browsed for removes the impression before it is served. Meta's own response is to build the same thing, which is an admission that the surface is contestable.
Snap, through SpectaclesSNAP
The glasses themselves. Snap's own annual report describes Spectacles as augmented-reality glasses on its own operating system, sold to developers today; a buyer who wants a camera and assistant on the face has a second listed seller.
Google, through GeminiGOOGL
The assistant a person talks to. Pinterest's own annual report lists Google's Gemini beside Meta AI among the competitors for the same attention; an assistant people already use on their phones leaves Meta's free one, and its agents, with nobody to charge.
How big it could get
The market and who has signed
The market
Americans spent a record $294.6B on internet advertising in 2025, up 13.9%. Social media, the kind Meta sells, grew fastest of the large formats: 32.6%, to $117.7B, overtaking search.
Meta's own advertising, $59.4B in the June 2026 quarter, is booked worldwide, so its share of this American market cannot be read off.
Who runs it
Leaders, pay and the cash
Who runs it
Its founder, Zuckerberg, started Meta in 2004 and still runs it. Susan Li has been finance chief since 2022, and a new president, Dina Powell McCormick, joined in January 2026. Zuckerberg owns about an eighth of the company but, through shares with ten votes each, controls most of the vote, so outside shareholders cannot overrule him. He takes a $1 salary. The other officers' bonuses are set by the board's judgement, not a formula, and their shares vest for staying, not for results.
What they did with the money
It is paying for the build partly with borrowed money. Debt rose about $24.9B in six months and net cash fell from $22.9B to $6.6B; on 3 November it sold $15B more in bonds. Buybacks, $26.26B in 2025, have stopped. It is also on the hook for up to $45.95B on a data-centre venture off its balance sheet.
The last report
The latest report
Meta's sales grew 26% to $64.6B but cash flow turned negative
What it did to the thesis Mixed quarter
The advertising engine passed its test, but cash flow turned negative and legal costs stepped up. The view holds; conviction is now low because the cushion is thinner.
The report
Sales beat the top of guidance, up 26% to $64.6B, with ad prices up 14% and ads shown up 10%. Profit fell 37% because of a $9.89B charge for the deal with 52 states; without it, profit would have risen.
The cash is the worry: spending of $34.9B left free cash flow at minus $2.1B, and net cash shrank to about $1.5B. A New Mexico court also ordered Meta to pay $25B, which it will appeal.
What's happening now
The background
The report is in. Sales of $64.6B beat the $61B to $64B guide, but free cash flow was negative and a $25B New Mexico order landed the same day, so the question is whether the spending pays back.
The next report, in early 2027, is the next test. Management guides $73B to $76B of sales and 2026 spending of $135B to $145B. A revenue line for the new AI products would answer the question the price is asking.
Where we disagree
Both sides in full
The market
The capital spending is mostly money burned, so the reported profit deserves less than rivals get. The market has paid up since August, but still thinks so.
We believe
The advertising engine is accelerating. The market is still discounting a build the accounts treat as an asset, though far less than a month ago.
The argument
The core business is doing what a strengthening business does: more advertisements shown and a higher price for each, at the same time. The market still pays less for profit it doubts will survive the spending, though the gap is smaller than in August.
The case against us is real. No revenue from the spending has been disclosed, management will not guide 2027 spending, and debt is rising. Earnings have missed forecasts in two of the last four quarters.
What it's worth
How we value it
An advertising engine growing near 28% while it spends on a build of unknown return is valued on the analysts' near-term path and a margin that absorbs the depreciation, slowing toward 4% by year ten, each year closing 30% of the gap.
What the price assumes
Sales growth in the first year, then slowing as ours does, at the same profit per sale.
The price assumed
19.9%
Our estimate
28%
Measured at the close of
6 Nov 2026
The three cases
Bear case$358
The build earns little (C10) and youth trials cost money (C18): growth falls to 8% and 5% after 2026, the margin drops to 28% as depreciation lands, and new capital earns only 9%.
Our value$694
2026 grows 28% to about $257B, the analysts' figure behind 14% more adverts at 12% higher prices (C06); then 18% and 14% as the base gets larger; the operating margin settles at 34%, about 2026's guided expenses of $165-169B (C09) once depreciation on the $130-145B build catches up.
Bull case$968
Business agents and glasses (C39) add revenue lines, growth holds at 22% and 19%, near the analysts' path, and the margin recovers to 38%, below 2025's 41.4%.
The bet
Seen as a bet
The bet is on Reality Labs, the glasses and headsets, together with the AI products Meta has begun to sell beyond advertising: subscriptions, business agents and its models. Neither makes money yet, so advertising is counted separately, at what the closest listed rivals are valued at.
Counted that way, advertising alone is worth more than the whole company today, so the price pays nothing for either. If both take off, the company is worth about half as much again, nearly all advertising; if they never earn, nothing is lost.
What could go wrong
Every risk
The risks
Spending that never pays back
$130B to $145B is guided for 2026 and no revenue from it has been disclosed.
Debt, then new shares, replacing cash
Net cash fell to about $1.5B from $22.9B, and on 26 October Meta launched about $35B of share sales with about $60B more planned.
A hidden commitment
Up to $45.95B of exposure to a data-centre venture sits off the balance sheet.
Commitments already signed
$454.5B of purchase commitments are non-cancelable, $121.4B of it due in 2027, so next year's spending is largely set.
Court cases and penalties
A New Mexico court ordered Meta to pay $25B and Meta will appeal; a $9.89B charge settled 52 states' claims; more youth trials are due.
Slower user growth
Daily users grew 3% against 7% six months earlier.
What we still do not know
What the capital spending earns. Meta says buyers have offered more for its computing power than it paid, and has put a number on none of it.
How much it will spend in 2027. The finance chief declined to say, though $121.4B of purchase commitments already fall due that year.
How fast its new data-centre spending wears out. Servers are written off over 5.5 years, lengthened in 2025.
How much stock it will sell. About $35B is launched and about $60B is planned with no agreement signed.
What changes our view
Every route to a new view
Every route to a new view
→ Attractive
The share price falls to about $555, 80% of what we think the business is worth.
→ Unattractive
Spending planned for 2027 is higher than 2026's, with no revenue from it disclosed. Operating profit falls from a year earlier in a quarter with no legal or severance charges. Advertising revenue growth drops below 10% from a year earlier, so the engine paying for the build is slowing.
→ Unattractive
The share price rises to about $832, 20% above what we think the business is worth.
The next test
It supports the case if
Fourth-quarter revenue lands at or above the top of the $73-76 billion guidance, impressions and price per ad both still rise, free cash flow is positive for the quarter, 2026 capital spending is at or below $145 billion, and the company quantifies revenue from the API, business agents or compute for the first time.
It pressures the case if
Revenue lands at or below the bottom of guidance, price per ad turns negative, 2026 capital spending exceeds $145 billion, free cash flow is negative again, or 2027 spending is guided above 2026's with no revenue line attached.
We would also be wrong if
Daily users stop growing or fall.
Net cash turns into net debt and the dividend is cut to pay for the build.
The verdict
Scores and how we weighed it
The scores
Growth
4 of 4
Business quality
3 of 4
Financial position
2 of 4
Execution record
3 of 4
Per share capture
2 of 4
Potential
3 of 4
Risk profile
3 of 4
Valuation
3 of 4
Evidence quality
4 of 4
How we weighed it
ValueMixed
The price is inside the margin of safety on cash after spending, with the bear case well below.
BusinessPositive
It shows more adverts and charges more for each across apps billions of people use.
GrowthPositive
Advertising grows fast and faster than it used to.
RisksNegative
The building bill may never pay back, debt is replacing cash, and large commitments sit off the balance sheet.
RivalsMixed
The cheapest of its rivals on profit and near the dearest on cash.
The groupMixed
The largest technology names move together on AI spending; its spending scale is its own.
Latest newsMixed
The advertising engine passed, but cash flow turned negative and legal costs rose.
In plain words
Depreciation
A data centre is paid for up front but used for years, so the accounts spread its cost over its working life, a slice each year. That yearly slice is depreciation. Spending today shows up as a cost in later years' profit.
In plain words
Free cash flow
The cash a company actually has left in a year after paying its bills and its investments. Profit can be shaped by accounting rules about when a cost is counted; cash cannot. When profit rises but free cash does not, investors ask why.
In plain words
Our estimate of value
We forecast how much cash the business can earn over the next ten years and beyond, then ask what that stream is worth today to an investor who wants an 8% yearly return. We do it three times: a gloomy future, an expected one and a sunny one. We call a share Attractive only when its price is at least 20% below the expected estimate, and Unattractive when it is 20% above.