META · Forward model · Superintelligence case
The Superintelligence case, 20 quarters out
Each segment is projected from its own operating driver, rolled up into consolidated cash flow, and discounted back to a fair value per share. The assumptions below are editable — change them and every number on this page moves with them.
Meta reports TWO segments, Family of Apps and Reality Labs, and splits Family of Apps into Advertising and Other revenue on the face of the release. This model carries those three reported lines exactly as disclosed for all six quarters shown - 2026 Q2 is $59,363M advertising, $1,007M other, $431M Reality Labs, $60,801M total, per the 8-K Exhibit 99.1 - and adds two more, Business AI and Meta AI, which have NO history and start from zero in the projection. Business AI is agents sold to businesses; Meta AI is the assistant sold to people, taking a share of what consumers pay OpenAI, Google, xAI and Anthropic today. Neither is a re-cut of anything Meta reported: both are held at zero across every reported quarter. Nothing here is apportioned: every historical number is a reported segment figure, which is why no quarter is flagged estimated. What IS assumed is the split of company-level costs and capital across the lines: Meta gives segment operating income but no segment depreciation or capex, so the margin and capex-intensity inputs on each vertical are ours. Advertising carries almost all of the AI infrastructure spend, because that is what the money is being spent on. Reality Labs is carried at a margin below -900% of its own revenue, which is what a $4.6B quarterly operating loss on $431M of revenue actually is.
Meta's own stated case, taken at face value: Meta Superintelligence Labs is worth the $130-145B a year it is costing, personal superintelligence becomes a product people pay for, business agents replace a chunk of what companies spend on software rather than a chunk of what they spend on ads, and Meta AI takes roughly a quarter of the consumer assistant market on distribution alone - about $26B a year, or four times what Meta earns from Reality Labs, Quest and glasses combined. Advertising still carries it - even here, the ad line is more than three quarters of terminal revenue, which is the finding rather than the flourish. What this case does NOT reach is a company where AI revenue rivals advertising: the two AI lines together would need to be roughly eight times the size this case gives them, and no disclosed driver produces that off two lines earning zero today. The honest version of the bet is that AI shows up as ad revenue - which is where this model already puts most of it.
Latest: $160.56B (2031Q2E)
| Period | Value |
|---|---|
| 2025Q1 | $42.31B |
| 2025Q2 | $47.52B |
| 2025Q3 | $51.24B |
| 2025Q4 | $59.89B |
| 2026Q1 | $56.31B |
| 2026Q2 | $60.80B |
| 2026Q3E | $63.67B |
| 2026Q4E | $66.62B |
| 2027Q1E | $70.44B |
| 2027Q2E | $74.01B |
| 2027Q3E | $77.75B |
| 2027Q4E | $81.67B |
| 2028Q1E | $85.76B |
| 2028Q2E | $90.04B |
| 2028Q3E | $94.53B |
| 2028Q4E | $99.23B |
| 2029Q1E | $104.14B |
| 2029Q2E | $109.29B |
| 2029Q3E | $114.69B |
| 2029Q4E | $120.34B |
| 2030Q1E | $126.27B |
| 2030Q2E | $132.48B |
| 2030Q3E | $139.00B |
| 2030Q4E | $145.84B |
| 2031Q1E | $153.02B |
| 2031Q2E | $160.56B |
What drives each segment
Advertising
Subscribers × ARPUFacebook, Instagram, WhatsApp, Messenger and Threads sold to advertisers. 97.6% of company revenue and effectively all of the profit. The people are already there - 3.60 billion daily - so this line is a monetisation story, not a user-growth story: impressions up 14% and price per ad up 12% is what a 27% quarter is made of.
Latest: $137.25B (2031Q2E)
| Period | Value |
|---|---|
| 2025Q1 | $41.39B |
| 2025Q2 | $46.56B |
| 2025Q3 | $50.08B |
| 2025Q4 | $58.14B |
| 2026Q1 | $55.02B |
| 2026Q2 | $59.36B |
| 2026Q3E | $62.04B |
| 2026Q4E | $64.78B |
| 2027Q1E | $67.62B |
| 2027Q2E | $70.58B |
| 2027Q3E | $73.65B |
| 2027Q4E | $76.85B |
| 2028Q1E | $80.17B |
| 2028Q2E | $83.62B |
| 2028Q3E | $87.20B |
| 2028Q4E | $90.93B |
| 2029Q1E | $94.80B |
| 2029Q2E | $98.82B |
| 2029Q3E | $103.01B |
| 2029Q4E | $107.35B |
| 2030Q1E | $111.87B |
| 2030Q2E | $116.56B |
| 2030Q3E | $121.44B |
| 2030Q4E | $126.51B |
| 2031Q1E | $131.78B |
| 2031Q2E | $137.25B |
Assumptions & reasoning
- Modelled as daily active people x revenue per person rather than as a growth rate, because the two halves move at completely different speeds: the audience grew 3% year over year and revenue grew 27%. A single growth number would hide that every dollar of upside here is price, not people.
- Revenue per person is derived from two disclosed numbers - $59,363M of advertising revenue over 3.60 billion daily active people - and is $5.50 a month. It is not a metric Meta publishes, and it moves with ad load, auction density and mix, not with a price list.
- The 2.5% quarterly drift on revenue per person is a deliberate deceleration: it compounds to about 10% a year against the 23% this line just ran. Holding today's monetisation growth for five years is the single easiest way to make this model say anything you like, so it is not held.
- The first projected quarter lands at about $61B, inside the company's own $61-64B guidance for Q3 2026, which is the only near-term check available on this line.
- Capex intensity starts at 70% of this segment's revenue and glides to 22%. That is not a guess: $50.9B was already spent in the first half against $130-145B of full-year guidance, which leaves about 68% of revenue a quarter for the back half. Almost all of it is AI infrastructure serving ranking, recommendation and generative ads, so it is charged here rather than spread evenly.
- The driver grows smoothly and Meta does not: advertising is seasonally strongest in Q4 and weakest in Q1, so individual projected quarters are wrong in a way that cancels over a year. Read the annual totals, not the Q4 line.
- The EBITDA margin of 50% is struck on a quarter carrying a $2.40B legal charge and $1.18B of severance; underlying is nearer 54%. It glides to 56% as those roll off and the infrastructure scales, which is the assumption most worth arguing with given depreciation is compounding behind it.
Family of Apps — Other
Growth pathEverything inside Family of Apps that is not an ad: WhatsApp paid messaging through the Business API, Meta Verified subscriptions, and the smaller service fees. It crossed $1B in a quarter for the first time in 2026 Q2, growing 73% year over year, which makes it the fastest line in the company and still 1.7% of it.
Latest: $7.11B (2031Q2E)
| Period | Value |
|---|---|
| 2025Q1 | $510M |
| 2025Q2 | $583M |
| 2025Q3 | $690M |
| 2025Q4 | $801M |
| 2026Q1 | $885M |
| 2026Q2 | $1.01B |
| 2026Q3E | $1.16B |
| 2026Q4E | $1.33B |
| 2027Q1E | $1.51B |
| 2027Q2E | $1.70B |
| 2027Q3E | $1.91B |
| 2027Q4E | $2.12B |
| 2028Q1E | $2.36B |
| 2028Q2E | $2.60B |
| 2028Q3E | $2.87B |
| 2028Q4E | $3.14B |
| 2029Q1E | $3.44B |
| 2029Q2E | $3.75B |
| 2029Q3E | $4.09B |
| 2029Q4E | $4.44B |
| 2030Q1E | $4.82B |
| 2030Q2E | $5.22B |
| 2030Q3E | $5.65B |
| 2030Q4E | $6.11B |
| 2031Q1E | $6.59B |
| 2031Q2E | $7.11B |
Assumptions & reasoning
- Carried as a growth rate because Meta discloses neither the paying business count nor the price per conversation, and inventing both to reach a number it already publishes would add precision without adding information.
- 12% a quarter decaying to 4% is a deceleration from the 13.8% it just posted. Paid messaging is priced per conversation in markets where WhatsApp is the default channel, so volume can keep compounding well after the growth rate falls.
- Margin is set below advertising at 35% gliding to 45%: messaging carries carrier and delivery costs an ad impression does not, and Meta Verified is priced against consumer willingness to pay rather than against an auction.
- This is the line that would absorb any consumer Meta AI subscription, so if that ships and is not broken out, it lands here rather than in Business AI.
Reality Labs
Growth pathQuest headsets, Ray-Ban and Oakley Meta AI glasses, and the software and content around them. $431M of revenue against a $4.62B operating loss in 2026 Q2 - it is 0.7% of the company's revenue and about a quarter of its operating income burned. Glasses are growing; headsets are shrinking.
Latest: $2.02B (2031Q2E)
| Period | Value |
|---|---|
| 2025Q1 | $412M |
| 2025Q2 | $370M |
| 2025Q3 | $470M |
| 2025Q4 | $955M |
| 2026Q1 | $402M |
| 2026Q2 | $431M |
| 2026Q3E | $471M |
| 2026Q4E | $513M |
| 2027Q1E | $558M |
| 2027Q2E | $606M |
| 2027Q3E | $658M |
| 2027Q4E | $713M |
| 2028Q1E | $771M |
| 2028Q2E | $834M |
| 2028Q3E | $901M |
| 2028Q4E | $972M |
| 2029Q1E | $1.05B |
| 2029Q2E | $1.13B |
| 2029Q3E | $1.22B |
| 2029Q4E | $1.31B |
| 2030Q1E | $1.41B |
| 2030Q2E | $1.52B |
| 2030Q3E | $1.63B |
| 2030Q4E | $1.75B |
| 2031Q1E | $1.88B |
| 2031Q2E | $2.02B |
Assumptions & reasoning
- Carried as a growth rate rather than as units x price because Meta has never disclosed headset or glasses unit sales. A units model here would be two invented numbers multiplied together to reproduce a disclosed one.
- The line is heavily seasonal - Q4 2025 was $955M against $470M the quarter before - and a smooth quarterly growth rate deliberately does not reproduce that. Read the annual totals, not the individual quarters.
- The margin of -970% is not a typo: it is a $4.18B EBITDA loss on $431M of revenue, which is what the reported $4.62B operating loss looks like once an assumed share of depreciation is added back. Expressing it any other way would quietly delete $15B a year of real spending from the free cash flow.
- It glides toward -250%, so the loss stays roughly flat in dollars as revenue grows rather than shrinking - which is what management guided to when it said Reality Labs operating losses would remain similar to 2025 levels.
- This is the cleanest single lever in the model: a reader who believes Meta caps the burn can drag the terminal margin and watch what closing the loss is worth, without touching a single advertising assumption.
Business AI
Subscribers × ARPUSelling AI to businesses rather than selling their ads: agents that answer, sell and support inside WhatsApp, Messenger and Instagram, priced per business rather than per impression. Zuckerberg called it 'entirely new enterprise opportunities' on the Q2 2026 call. It earns nothing today and Meta reports nothing for it, so it starts from zero here.
Latest: $7.67B (2031Q2E)
| Period | Value |
|---|---|
| 2025Q1 | $0.00 |
| 2025Q2 | $0.00 |
| 2025Q3 | $0.00 |
| 2025Q4 | $0.00 |
| 2026Q1 | $0.00 |
| 2026Q2 | $0.00 |
| 2026Q3E | $0.00 |
| 2026Q4E | $0.00 |
| 2027Q1E | $442M |
| 2027Q2E | $703M |
| 2027Q3E | $980M |
| 2027Q4E | $1.27B |
| 2028Q1E | $1.58B |
| 2028Q2E | $1.91B |
| 2028Q3E | $2.26B |
| 2028Q4E | $2.62B |
| 2029Q1E | $3.01B |
| 2029Q2E | $3.42B |
| 2029Q3E | $3.86B |
| 2029Q4E | $4.32B |
| 2030Q1E | $4.80B |
| 2030Q2E | $5.31B |
| 2030Q3E | $5.85B |
| 2030Q4E | $6.43B |
| 2031Q1E | $7.03B |
| 2031Q2E | $7.67B |
Assumptions & reasoning
- This vertical has NO reported history and no carved-out base. Meta has never disclosed a business AI revenue line, so rather than apportion one out of Other revenue, it is held at zero across every reported quarter and starts in the projection.
- Two quarters of zero before the first dollar, so nothing is booked before 2027 Q1. Anything Meta already charges businesses for messaging is in Family of Apps Other, not here, and double counting it would be the easy mistake.
- The base is the roughly 200 million businesses Meta has said use its apps. Attach starts at 0.5% and glides to 12%, which is a paying-business story, not a usage one - most businesses will use free agents and never pay.
- $60 a month is priced against small business software, not against enterprise AI seats. If this line is instead sold on usage at agent scale, the price is wrong by an order of magnitude in either direction, which is why it is a slider and not a forecast.
- Capex for this line is charged as a percentage of its own revenue, so during its two zero quarters it books no capital spending at all - real infrastructure the model charges nowhere. In practice that spend is already inside advertising's 60% intensity.
Meta AI
Capacity × utilisation × priceThe assistant sold as a product rather than as a feature: paid tiers, AI-native subscriptions and whatever Meta charges for intelligence directly, competing with ChatGPT, Gemini, Grok and Claude for the consumer assistant market. Meta's advantage is distribution — the assistant is already inside WhatsApp and Instagram, which is why usage in Europe has moved to it without anyone downloading anything — and its problem is that it has never charged for any of it. Modelled as market x share, because share is the only quantity this argument is ever made in.
Latest: $6.51B (2031Q2E)
| Period | Value |
|---|---|
| 2025Q1 | $0.00 |
| 2025Q2 | $0.00 |
| 2025Q3 | $0.00 |
| 2025Q4 | $0.00 |
| 2026Q1 | $0.00 |
| 2026Q2 | $0.00 |
| 2026Q3E | $0.00 |
| 2026Q4E | $0.00 |
| 2027Q1E | $310M |
| 2027Q2E | $424M |
| 2027Q3E | $556M |
| 2027Q4E | $708M |
| 2028Q1E | $881M |
| 2028Q2E | $1.08B |
| 2028Q3E | $1.30B |
| 2028Q4E | $1.56B |
| 2029Q1E | $1.84B |
| 2029Q2E | $2.16B |
| 2029Q3E | $2.52B |
| 2029Q4E | $2.92B |
| 2030Q1E | $3.37B |
| 2030Q2E | $3.87B |
| 2030Q3E | $4.43B |
| 2030Q4E | $5.05B |
| 2031Q1E | $5.74B |
| 2031Q2E | $6.51B |
Assumptions & reasoning
- No reported history and no carved-out base: Meta has never disclosed a dollar of Meta AI revenue, and every dollar the assistant earns today is an ad someone saw afterwards. The line is held at zero across all six reported quarters and opens in 2027 Q1.
- The unit is $1M of quarterly consumer-assistant market, so the whole line is pool x share. The pool is what people pay for assistants directly - subscriptions and consumer tiers - not the API and enterprise book that most of OpenAI's and Anthropic's revenue actually is.
- The finding is that the pool's growth rate matters more than Meta's share of it. At 12% share of a pool compounding 5% a quarter this is about $3B a quarter by 2031 - a real business, and still under 3% of Meta. Distribution wins the users; it does not automatically win revenue.
- Free distribution is exactly why the share can move and the revenue might not. Meta AI reaches people who never chose it, which is the opposite of the paying relationship ChatGPT has with a subscriber, and a bundled assistant inside WhatsApp is very hard to put a price on later.
- This line does NOT carry the ad revenue that better AI produces - that is already inside the advertising vertical's monetisation drift, and counting engagement twice is the easiest way to make this model lie.
- The compute behind it is charged in advertising's capex intensity, not here, so this vertical's own capex number covers only the incremental serving cost of the paid tier.
Where each case comes from
Superintelligence case — primary sources
The primary sources this case is built from — filings, calls and posts. Where they are claims by an interested party rather than disclosures, every number in the Superintelligence column is what happens if they are taken at face value.
What Meta says the AI is already doing
- Jul 29, 2026 AI is accelerating our core business today, powering our next generation of products, and opening the door to entirely new enterprise opportunities. The results are already showing, and I'm optimistic about the potential ahead.
- Jul 29, 2026 Ad impressions delivered across our Family of Apps increased by 14% year-over-year. Average price per ad increased by 12% year-over-year.
Selling intelligence, not compute
- Jul 29, 2026 We're getting a lot of offers for compute at a significant premium for what we paid for it.
- Jul 29, 2026 It would be foolish to basically just sell all of the compute and take a short-term profit, when the company expects a significantly higher margin on selling intelligence rather than selling compute directly.
What it is costing to find out
- Jul 29, 2026 We anticipate 2026 capital expenditures, including principal payments on finance leases, to be in the range of $130-145 billion, narrowed from our prior outlook of $125-145 billion.
- Jan 28, 2026 We anticipate 2026 capital expenditures to be in the range of $115-135 billion, with year-over-year growth driven by increased investment to support our Meta Superintelligence Labs efforts and core business.
From cash flow to fair value
| Present value of free cash flow, 20 quarters | $252.98B |
| Terminal-year revenue | $598.42B |
| Terminal-year EBITDA | $348.71B |
| Exit multiple, on revenue | 8.0x |
| Terminal value | $4.79T |
| Discounted at 8.5% a year, terminal value becomes | $3.18T |
| Enterprise value | $3.44T |
| Net cash | $6.60B |
| Equity value | $3.44T |
| Diluted shares | 2.57B |
| Fair value per share | $1,341.93 |
| Against the current price of $545.83 | +146% |
6x forward revenue against roughly 6.1x trailing at $543.67 today. The terminal year is growing near 15%, not 28%, so paying a premium to today's multiple for a slower business needs an argument this model does not make. It is still the largest single input: halve it and the fair value moves further than every operating assumption combined.
Read the other way round: at $545.83 the market is paying 2.9x terminal-year revenue, holding every other assumption on this page fixed. That is the number to argue about.
The projected path
| Quarter | Advertising | Family of Apps — Other | Reality Labs | Business AI | Meta AI | Revenue | YoY | EBITDA | Capex | FCF | R40 | PV of FCF |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 2026 Q3E | $62.04B | $1.16B | $471M | $0 | $0 | $63.67B | +24% | $33.17B | $40.81B | -$7.64B | +12 | -$7.48B |
| 2026 Q4E | $64.78B | $1.33B | $513M | $0 | $0 | $66.62B | +11% | $35.16B | $39.82B | -$4.67B | +4 | -$4.48B |
| 2027 Q1E | $67.62B | $1.51B | $558M | $442M | $310M | $70.44B | +25% | $37.22B | $39.22B | -$1.99B | +22 | -$1.88B |
| 2027 Q2E | $70.58B | $1.70B | $606M | $703M | $424M | $74.01B | +22% | $39.45B | $38.57B | $735M | +23 | $678M |
| 2027 Q3E | $73.65B | $1.91B | $658M | $980M | $556M | $77.75B | +22% | $41.79B | $38.03B | $3.16B | +26 | $2.85B |
| 2027 Q4E | $76.85B | $2.12B | $713M | $1.27B | $708M | $81.67B | +23% | $44.26B | $37.60B | $5.59B | +29 | $4.95B |
| 2028 Q1E | $80.17B | $2.36B | $771M | $1.58B | $881M | $85.76B | +22% | $46.86B | $37.28B | $8.05B | +31 | $6.98B |
| 2028 Q2E | $83.62B | $2.60B | $834M | $1.91B | $1.08B | $90.04B | +22% | $49.59B | $37.06B | $10.52B | +33 | $8.94B |
| 2028 Q3E | $87.20B | $2.87B | $901M | $2.26B | $1.30B | $94.53B | +22% | $52.44B | $36.94B | $13.02B | +35 | $10.84B |
| 2028 Q4E | $90.93B | $3.14B | $972M | $2.62B | $1.56B | $99.23B | +22% | $55.43B | $36.92B | $15.55B | +37 | $12.68B |
| 2029 Q1E | $94.80B | $3.44B | $1.05B | $3.01B | $1.84B | $104.14B | +21% | $58.56B | $36.99B | $18.11B | +39 | $14.47B |
| 2029 Q2E | $98.82B | $3.75B | $1.13B | $3.42B | $2.16B | $109.29B | +21% | $61.83B | $37.17B | $20.71B | +40 | $16.22B |
| 2029 Q3E | $103.01B | $4.09B | $1.22B | $3.86B | $2.52B | $114.69B | +21% | $65.25B | $37.45B | $23.35B | +42 | $17.91B |
| 2029 Q4E | $107.35B | $4.44B | $1.31B | $4.32B | $2.92B | $120.34B | +21% | $68.83B | $37.82B | $26.04B | +43 | $19.58B |
| 2030 Q1E | $111.87B | $4.82B | $1.41B | $4.80B | $3.37B | $126.27B | +21% | $72.57B | $38.29B | $28.79B | +44 | $21.20B |
| 2030 Q2E | $116.56B | $5.22B | $1.52B | $5.31B | $3.87B | $132.48B | +21% | $76.47B | $38.86B | $31.60B | +45 | $22.80B |
| 2030 Q3E | $121.44B | $5.65B | $1.63B | $5.85B | $4.43B | $139.00B | +21% | $80.56B | $39.53B | $34.47B | +46 | $24.37B |
| 2030 Q4E | $126.51B | $6.11B | $1.75B | $6.43B | $5.05B | $145.84B | +21% | $84.84B | $40.29B | $37.42B | +47 | $25.92B |
| 2031 Q1E | $131.78B | $6.59B | $1.88B | $7.03B | $5.74B | $153.02B | +21% | $89.31B | $41.16B | $40.45B | +48 | $27.46B |
| 2031 Q2E | $137.25B | $7.11B | $2.02B | $7.67B | $6.51B | $160.56B | +21% | $93.99B | $42.12B | $43.57B | +48 | $28.98B |
Every row is projected. A year-over-year change is shown only where the quarter it compares against exists — an em dash means there is no comparable quarter, not a flat year.
Model revisions
Assumptions are marked to reality as each quarter prints. Every change is appended here, with the fair value the model produced at the time, so the model's own history stays visible.
| Date | Changed | Fair value then | Note |
|---|---|---|---|
| 2026-08-19 | all | $757.11 | Initial model, built off the 2026 Q2 print: $60.801B revenue, $59.363B advertising, $1.007B Family of Apps other, $431M Reality Labs, 3.60B daily active people, $31.08B of quarterly capex against $130-145B of full-year guidance. Five verticals: the three reported lines plus Business AI and Meta AI, neither of which has any history, both opening in 2027 Q1. Meta AI is modelled as share of a consumer assistant market estimated at $6.5B a quarter. |