Sales
PricedIn$63.1B$60.9B – $65.1B
$60.9B$63.1B$65.1B
adjusted EPS
PricedIn$6.86$5.85 – $7.81
$5.85$6.86$7.81
Middlethe forecast80% range1 in 10 chance below this bar, 1 in 10 above it. These are the low and high printed by each number.
All six expect sales inside the outlook, from $62.5B to $64.0B. GLM and Muse Wonder are lowest at $62.5B; GLM sees growth 'decelerating as Meta laps strong impression gains'. Sol is highest at $64.0B because 'U.S. spending data have not turned down'. On earnings, Muse Wonder's $7.84 stands apart: it keeps 'costs inside the guided range even after the $10 billion legal accrual'.
Third-quarter sales of about $63.1B, in the upper half of the company's own outlook, as ad demand held up.
Sales$63.1B$60.9B – $65.1B · +23% on Q3 2025 ($51.2B)
Adjusted EPS$6.86$5.85 – $7.81 · +554% on Q3 2025 ($1.05)
Investment and cash generation
Base case
Capital spending$41.1B
Free cash flow−$9.05B
Free cash flow: operating cash flow less capital spending.
Meta reported second-quarter sales of $60.8B, up 28% from a year earlier, as ad impressions rose 14% and the average price per ad rose 12%. Costs jumped too: total expenses reached $42.0B, including a $2.4B legal charge and $1.2B of severance from the May job cuts, so the operating margin fell to 31% from 41% in the first quarter.
For the third quarter, which ended on 30 September, the company guided sales to between $61B and $64B, assuming currency trims growth by about 1%. In August it settled with state attorneys general over harm to young users and said it would book a legal expense of about $10B in the quarter, while leaving its sales outlook unchanged.
The six AI models put sales between $62.5B and $64.0B, averaging $63.1B, about 3.8% above the second quarter. Most sit near the top of the outlook, where the company has landed in recent quarters. GLM and Muse Wonder stay at the midpoint, stressing the reasons management gave for slower growth: harder comparisons, less personalized ads in Europe now fully rolled out, and integrity enforcement.
The settlement charge pulls the reported operating margin down to about 16% on average. Adjusted earnings on this page leave out only that charge. Oil climbed above $120 a barrel in September, and the models weigh whether that hurt late-quarter ad budgets, as the Iran conflict did early in the year.
Meta reports the quarter on 21 October 2026, after the market closes, and will give its outlook for the fourth quarter then.
Adjusted earnings per share average $6.86, up from $6.18 in the second quarter, with the $10B settlement charge added back and tax near the guided 15% to 17%. Muse Wonder is highest at $7.84, with a 22% reported margin against about 16% for the rest.
Outlook against result
Q1 2025+1.2% above$39.5B–$41.8B → $42.3Bguided “$39.5-41.8 billion”
Q2 2025+4.4% above$42.5B–$45.5B → $47.5Bguided “$42.5-45.5 billion”
Q3 2025+1.5% above$47.5B–$50.5B → $51.2Bguided “$47.5-50.5 billion”
Q4 2025+1.5% above$56.0B–$59.0B → $59.9Bguided “$56-59 billion”
Q1 2026inside$53.5B–$56.5B → $56.3Bguided “$53.5-56.5 billion”
Q2 2026inside$58.0B–$61.0B → $60.8Bguided “$58-61 billion”
Q3 2026inside$61.0B–$64.0B → $63.1Bguided “$61-64 billion”
Each outlook is the one the company gave before the quarter. The last row is the models' forecast.
The quarter forecast, line by line
The lines as the company's statement shows them. The forecast columns are the average of the models, line by line. Sales growth is the change on the same quarter a year earlier. Earnings per share is the headline's: the company's own for the reported quarter, and for a forecast quarter worked out from the averaged lines with the company's adjustments.
What drove second-quarter 2026 growth
Total revenue growth28%
Ad impressions growth14%
Average price per ad growth12%
Other app revenue growth73%
Reality Labs revenue growth16%
Where each number comes from
- Total revenue growth
Moving now to our consolidated results. Q2 total revenue was $60.8 billion, up 28% or 27% on a constant currency basis. Q2 total expenses were $42 billion, up 55% compared to last year, and included $2.4 billion in charges related to legal proceedings and $1.2 billion in severance expenses in connection with the May 2026 headcount reduction. Year-over-year growth was primarily driven by increases in employee compensation, infrastructure costs, legal-related costs, and third-party AI token costs.
- Ad impressions growth
In Q2, the total number of ad impressions served across our services increased 14%.
- Average price per ad growth
The global average price per ad increased 12% year-over-year, driven by ad performance gains, improvements in macro conditions relative to Q2 of last year, and currency tailwinds.
- Other app revenue growth
For the first time, quarterly family of apps other revenue reached $1 billion and grew 73% year-over-year
- Reality Labs revenue growth
Within our Reality Lab segment, Q2 revenue was $431 million, up 16% year-over-year due to strong growth in AI glasses revenue
The second quarter's growth by part, which the models carry, slowed, into the third quarter.
What it rests on
- Third-quarter sales outlook
- $61B to $64B, with currency about a 1% headwind
We expect third quarter 2026 total revenue to be in the range of $61-64 billion. Our guidance assumes foreign currency is an approximately 1% headwind to year-over-year total revenue growth, based on current exchange rates.
Second-quarter earnings release, July 2026 - Settlement charge
- A legal expense of about $10B booked in the third quarter
annualized evenly. We expect to accrue a legal expense of approximately $10
Attorneys general settlement call, August 2026 - Outlook kept after settlement
- Third-quarter sales outlook left unchanged in August
On this question, our Q3 revenue outlook remains unchanged. We have not
Attorneys general settlement call, August 2026 - Ad impressions
- Up 14% in the second quarter
In Q2, the total number of ad impressions served across our services increased 14%.
Second-quarter earnings call, July 2026 - Europe ad headwind
- Less personalized ads in Europe, now fully rolled out, weigh on growth
Second, we’re seeing the impact of less personalized ads offering in Europe
Second-quarter follow-up call, July 2026 - Teen revenue share
- Teens bring in less than 1% of revenue
for less than 1% of Meta's revenue.
Attorneys general settlement call, August 2026
Low and high
Basethe middle forecast
Key assumptions
The roughly $10B settlement charge is booked this quarter and left out of adjusted earnings. The company said in August it would accrue about $10B in the third quarter.6 of 6
- Opus 5.5I put it in third-quarter costs ($53.3 billion in total) and take it out of adjusted earnings per share as signed
- GPT-6.1 SolI remove that $10 billion only from the requested adjusted earnings in the quarter and year
- Grok 4.7I put that charge in third-quarter costs and removed it, after tax, from adjusted earnings
- GLM 5.3is excluded from adjusted earnings per share
- Gemini 4 Argonour corresponding 10 billion dollar adjustment to exclude this one-time charge
- Muse WonderI put the full accrual in third quarter costs and remove it, with its tax effect, for adjusted earnings
Sales land in the upper half of the $61B to $64B outlook. The company reaffirmed the range in August and has recently finished near the top of its outlooks.4 of 6
- Opus 5.5I put the third quarter at $63.2 billion: near the top
- Grok 4.7I expect the third quarter to land near the top of the company 61 to 64 billion dollar revenue range
- GPT-6.1 SolMy $64 billion middle is above the $62.5 billion guidance midpoint
- Gemini 4 Argonsupported our middle forecast of 63.2 billion dollars for third quarter 2026 revenue in the upper half of management guidance
Growth slows from the second quarter's 28% on a year earlier. Management named lapping strong impression growth, less personalized ads in Europe and integrity enforcement.4 of 6
- Opus 5.5I carry these forward and slow growth to about 24% in the third quarter
- GLM 5.3That is why my Q3 grows about 22 percent and Q4 about 21 percent against Q2's 28 percent
- GPT-6.1 SolI temper advertising growth to about 25% in the September quarter
- Muse Wonderthe third quarter guide embeds a modest constant currency deceleration
No sales cut for the teen time limits in the settlement. Teens bring in less than 1% of revenue, and the outlook was left unchanged.4 of 6
- Opus 5.5Teens are less than 1% of revenue, so I make no revenue cut for the two-hour limit
- Grok 4.7The third-quarter revenue range was left unchanged, so I did not cut sales for the teen rules
- GLM 5.3so I left revenue untouched
- Gemini 4 Argonour decision not to reduce third quarter or future revenue for the new two-hour teen daily time limit
Tax on normal profit near 16%, inside the guided 15% to 17%. The July release raised the tax outlook for the rest of 2026.3 of 6
- GLM 5.3I used the guided 15-17 percent rate for Q3 and Q4
- Muse Wondera 16% tax rate for the second half
- GPT-6.1 SolI use a 16% tax rate on normal second-half profit
If all of this happens
$63.1Bsales
$6.86adjusted EPS
$41.1BCapital spending
−$9.05BFree cash flow
Cash walk not supplied by every model.
Lowabout 1 chance in 10 of worse
What the low case assumes
Advertisers pull back after the oil spike, pushing sales to the bottom of the outlook. Advertisers cut spending when the Iran conflict began early in 2026.4 of 6
- Opus 5.5The autumn oil shock cuts advertiser budgets the way the Iran conflict did in early 2026
- Grok 4.7A lasting energy shock and weak household confidence cut advertiser budgets
- GLM 5.3ad demand softens with a weak consumer
- Muse WonderWeaker advertiser demand and pricing
Less personalized ads in Europe and integrity enforcement bite harder than planned. Management flagged both as headwinds now that the European offering is fully rolled out.3 of 6
- GLM 5.3European less-personalized-ads and integrity enforcement dragging more than expected
- Muse WonderEurope's less personalised ads and integrity work bite harder
- GPT-6.1 SolSlower engagement growth and tighter data-use restrictions
Costs run at the top of the range, or further legal charges arrive. Youth-related trials continue, and only the signed settlement is removed from adjusted earnings.3 of 6
- Muse Wondercosts stay at the top of the range
- GPT-6.1 Solother legal or tax losses can still depress results
- GLM 5.3Further material legal accruals from the youth-litigation docket
If all of this happens
$60.9Bsales · −4% vs base
$5.85EPS · −15% vs base
Sales would have to come in near the bottom of the outlook, about $60.9B on average, 3.6% under the base, with adjusted earnings per share near $5.85, 15% lower. That needs late-quarter ad budgets to buckle after oil climbed above $120 a barrel in September, as advertisers pulled back when the Iran conflict began, plus a heavier drag from Europe and integrity enforcement. Muse Wonder's sales low is furthest at $60.0B because it pairs weak demand with 'new legal charges follow the attorneys general accrual', so 'costs stay at the top of the range'. Sol's earnings low is furthest at $5.20: it removes only the signed settlement, so 'other legal or tax losses can still depress results'.
In early 2026 advertisers cut spending when the Iran conflict began, then recovered into April; the second quarter still grew 28%. The company reaffirmed its third-quarter outlook in late August, before oil jumped again.
The six lows run from $60.0B (Muse Wonder) to $61.5B (GPT-6.1 Sol).
First signThird-quarter sales near or below $61.5B in the 21 October 2026 release.
Highabout 1 chance in 10 of better
What the high case assumes
Ad performance gains keep prices and conversions climbing. The price per ad rose 12% in the second quarter, driven by ad performance gains.4 of 6
- GLM 5.3Ads performance gains compounding faster than expected
- GPT-6.1 SolStronger engagement and sustained advertising pricing
- Grok 4.7Ad demand and performance gains hold near the top of recent growth
- Muse WonderStronger engagement and advertiser returns from large language model based recommendations
Paid messaging and business agents add visible revenue. Other app revenue reached $1 billion in the second quarter, up 73%.4 of 6
- GPT-6.1 SolFaster monetization of messaging and business agents
- Grok 4.7Paid messaging, subscriptions, and business agents scale faster than expected
- Muse WonderNew revenue from business agents, subscriptions, paid messaging and AI glasses
- Gemini 4 ArgonRapid enterprise and small business adoption of paid Meta Business Agents
The September oil spike leaves ad budgets untouched. The second quarter still grew 28% despite high oil prices after the Iran conflict.2 of 6
- Grok 4.7with the late-summer rise in oil prices not hitting budgets
- Opus 5.5and the oil shock fades
If all of this happens
$65.1Bsales · +3% vs base
$7.81EPS · +14% vs base
Sales would have to beat the top of the outlook, about $65.1B on average, 3.1% above the base, with adjusted earnings per share near $7.81, 14% higher. Ad demand would need to keep its first-half strength, with the September oil spike not denting budgets and the new ranking systems lifting prices. Sol's high is furthest, at $66.5B in sales and $8.20 in earnings per share, because it alone has 'September sales exceed guidance' as 'subscriptions and business agents scale faster'; the others stop at $65.2B or below.
The second quarter grew 28%, with the price per ad up 12% on ad performance gains. Business agent charging began only in the second half, so new paid products are small this quarter.
The six highs run from $64.5B (GLM 5.3) to $66.5B (GPT-6.1 Sol).
First signThird-quarter sales above $64B in the 21 October 2026 release, with price per ad still growing strongly.
Watch
On 21 October 2026, after the close: sales against the $61B to $64B outlook, the fourth-quarter outlook, and any change to the $165B to $169B expense range. Sales near $61B or a new legal charge point lower; sales above $64B with strong ad prices point higher.
Sales
PricedIn$72.2B$67.9B – $76.2B
$67.9B$72.2B$76.2B
adjusted EPS
PricedIn$8.37$6.87 – $9.73
$6.87$8.37$9.73
Middlethe forecast80% range1 in 10 chance below this bar, 1 in 10 above it. These are the low and high printed by each number.
All six see a strong holiday quarter, from $70.5B to $74.5B. Muse Wonder is lowest at $70.5B, building from a third quarter 'near the middle of guidance'. Sol is highest at $74.5B, holding ad growth at 'about 25% in the September quarter and 24% in the holiday quarter'. Sol's range is much the widest: its $69.0B low starts from 'weaker holiday sales guidance'; its $80.0B high has 'holiday sales approach $80 billion'.
Holiday-quarter sales of about $72.2B, a smaller seasonal jump than usual as oil prices and a weaker euro weigh.
Sales$72.2B$67.9B – $76.2B · +21% on Q4 2025 ($59.9B)
Adjusted EPS$8.37$6.87 – $9.73 · −6% on Q4 2025 ($8.88)
Investment and cash generation
Capital spending: not supplied by every model
Free cash flow: not supplied by every model
Free cash flow: operating cash flow less capital spending.
The fourth quarter is the company's strongest season: last year it brought sales of $59.9B, up 17% on the third quarter. The company gives its outlook for this quarter with its third-quarter release on 21 October 2026, so the models forecast it before any guidance.
The six AI models put holiday sales between $70.5B and $74.5B, averaging $72.2B, about 14% above their third-quarter forecasts. Most hold the step-up below a normal holiday jump, because oil climbed above $120 a barrel in September and October and the euro weakened to about $1.12.
The company's reasons for slower growth carry into this quarter: lapping a year of strong impression gains, less personalized ads in Europe now fully rolled out, and integrity enforcement. Paid business agents began charging in August, but the models expect them to stay small this year.
With the settlement charge behind it, the reported operating margin recovers to about 35% on average. Adjusted earnings per share average $8.37, from $7.64 for Opus to $9.52 for Muse Wonder.
Sol's range is the widest, from $69.0B to $80.0B in sales, while GLM's runs only from $70.0B to $74.0B.
Fourth-quarter earnings per share rest on holiday ad sales, tax near 16% and costs that keep the year within the $165B to $169B expense outlook before the settlement. They average $8.37.
The quarter forecast, line by line
The lines as the company's statement shows them. The forecast columns are the average of the models, line by line. Sales growth is the change on the same quarter a year earlier. Earnings per share is the headline's: the company's own for the reported quarter, and for a forecast quarter worked out from the averaged lines with the company's adjustments.
What it rests on
- Full-year expense outlook
- $165B to $169B for 2026, set before the settlement
We now expect full year 2026 total expenses to be in the range of $165 to $169 billion.
Second-quarter earnings call, July 2026 - Operating income pledge
- 2026 operating income to top 2025's
We continue to expect to deliver operating income this year that is above 2025 operating income.
Second-quarter earnings call, July 2026 - Europe headwind
- Less personalized ads in Europe fully rolled out, a further headwind
now that it’s fully rolled out, and that may be an additional headwind. And
Second-quarter follow-up call, July 2026 - Business agent charging
- Charging businesses for Meta Business Agent from the second half
And in the second half, we plan to begin charging businesses for Meta Business
Second-quarter follow-up call, July 2026 - Iran shock precedent
- Advertisers cut spending when the Iran conflict began
advertiser spend coinciding with the beginning of the conflict in Iran at the end
First-quarter follow-up call, April 2026 - Brent oil price
- Brent oil above $120 a barrel in early October 2026
2026-10-02 135.51
Brent oil price series, October 2026
Low and high
Basethe middle forecast
Key assumptions
Holiday sales rise about 14% on the third quarter, less than a normal seasonal jump. Oil above $120 a barrel and a weaker euro weigh on fourth-quarter budgets and revenue.3 of 6
- Opus 5.5So I hold the fourth quarter at $71 billion (about 18.5% growth) instead of the 20%+ that extending the trend would give
- Grok 4.7The fourth quarter should rise from the third, but by less than the usual holiday jump
- GPT-6.1 SolThis modestly restrains holiday sales.
Growth on a year earlier keeps easing from the second quarter's 28%. Harder comparisons and Europe's less personalized ads carry into the holiday quarter.3 of 6
- GLM 5.3Q4 about 21 percent against Q2's 28 percent
- GPT-6.1 Sol24% in the holiday quarter
- Opus 5.5about 18.5% in the fourth
Costs return to a normal run rate, with no new legal charge. Further charges are possible, but their size and timing are unknown.3 of 6
- Grok 4.7I did not put another large charge in the base case beyond the 10 billion dollar accrual already announced
- GLM 5.3so I put no further accrual in the middle
- Opus 5.5I leave it out of the middle
New paid products stay small this year. Business agent charging began only in the second half of 2026.3 of 6
- GLM 5.3I kept them out of 2026 (too late in the year to be material)
- Grok 4.7Newer paid products stay small next to advertising through 2028
- GPT-6.1 SolGradual growth of paid messaging, subscriptions and business agents
Tax near 16%, inside the guided 15% to 17%. The July release set that range for the rest of 2026.3 of 6
- GLM 5.3I used the guided 15-17 percent rate for Q3 and Q4
- Muse Wondera 16% tax rate for the second half
- GPT-6.1 SolI use a 16% tax rate on normal second-half profit
If all of this happens
$72.2Bsales
$8.37adjusted EPS
Lowabout 1 chance in 10 of worse
What the low case assumes
The oil spike cuts holiday ad budgets with a lag. Advertisers pulled back when the Iran conflict began early in 2026.4 of 6
- Opus 5.5The autumn oil shock cuts advertiser budgets the way the Iran conflict did in early 2026
- Grok 4.7The jump in oil prices and very weak household confidence would cut holiday budgets
- GPT-6.1 SolConsumer demand and advertising budgets weaken after the energy shock
- Gemini 4 ArgonElevated energy prices above 100 dollars per barrel
Europe imposes further limits on personalized ads. The company is appealing the European decision, and further changes may be imposed.3 of 6
- Opus 5.5Europe imposes further changes to less personalized ads
- Grok 4.7Europe would impose a harsher limit on personalized ads
- Gemini 4 ArgonStricter European Union regulatory enforcement on less personalized ads
Another large legal charge lands, such as the $5.3B contingent piece of the settlement. Nearly 30% of the payments depend on YouTube and TikTok joining the deal.3 of 6
- Grok 4.7another large legal charge would land
- GLM 5.3including the roughly $5.3 billion contingent piece of the attorneys-general settlement if YouTube and TikTok join
- Muse Wondernew legal charges follow the attorneys general accrual
If all of this happens
$67.9Bsales · −6% vs base
$6.87EPS · −18% vs base
Holiday sales would have to fall to about $67.9B on average, 6.0% under the base, with earnings per share near $6.87, 18% lower. That takes the autumn oil spike cutting holiday ad budgets with a lag, harsher European rules and a weak consumer. Muse Wonder's sales low is furthest at $66.0B, starting from a weaker third quarter and adding 'new legal charges'; Opus's $66.5B low lets fourth-quarter growth slow 'to about 11%'. Sol's earnings low is furthest at $5.70, because it removes only the signed settlement, so 'other legal or tax losses can still depress results'.
The fourth quarter of 2025 rose 17% on the third. Early in 2026, advertisers cut spending when the Iran conflict began, then recovered into April.
The six lows run from $66.0B (Muse Wonder) to $70.0B (GLM 5.3).
First signA fourth-quarter outlook implying growth under 15%, given in the 21 October 2026 release.
Highabout 1 chance in 10 of better
What the high case assumes
Ad performance keeps compounding through the holidays. New ranking systems lifted ad clicks and conversions in the second quarter.3 of 6
- GLM 5.3Ads performance gains compounding faster than expected
- GPT-6.1 SolRecommendation improvements keep compounding
- Grok 4.7Ad demand and performance gains hold near the top of recent growth
Business agents and subscriptions scale faster. Business agent charging began in August, and other app revenue grew 73% in the second quarter.3 of 6
- GPT-6.1 Solsubscriptions and business agents scale faster
- Gemini 4 ArgonRapid enterprise and small business adoption of paid Meta Business Agents, MetaOne subscriptions
- Muse Wonderbusiness agents, paid messaging, subscriptions and the model API start to add visible revenue
The oil shock fades without hitting budgets. Performance ads kept growing through the earlier oil spike.2 of 6
- Opus 5.5and the oil shock fades
- Grok 4.7with the late-summer rise in oil prices not hitting budgets
If all of this happens
$76.2Bsales · +6% vs base
$9.73EPS · +16% vs base
Holiday sales would have to reach about $76.2B on average, 5.6% above the base, with earnings per share near $9.73, 16% higher. Ad gains from the new ranking systems would need to keep growth near the first half's pace, with the oil spike fading. Sol's high is furthest at $80.0B, because it alone lets holiday sales approach that level as 'subscriptions and business agents scale faster'; Grok stops at $78.2B and the rest at $75.8B or below.
The fourth quarter of 2025 rose 17% on the third. GLM's high has this fourth quarter growing about 24% on a year earlier.
The six highs run from $74.0B (GLM 5.3) to $80.0B (GPT-6.1 Sol).
First signA fourth-quarter outlook in the 21 October 2026 release implying growth above 22%.
Watch
The fourth-quarter outlook in the 21 October 2026 release: growth implied above 22% points higher, under 15% lower. Also watch for any new legal accrual and for early revenue from business agents.
Sales
PricedIn$252B$246B – $258B
$246B$252B$258B
adjusted EPS
PricedIn$31.84$28.78 – $34.43
$28.78$31.84$34.43
Middlethe forecast80% range1 in 10 chance below this bar, 1 in 10 above it. These are the low and high printed by each number.
The six agree closely, from $250B to $256B. Muse Wonder is lowest on sales at $250B, with a third quarter 'near the middle of guidance', yet highest on earnings at $33.94, keeping 'costs inside the guided range even after the $10 billion legal accrual'. Sol is highest on sales at $256B, with 'holiday revenue $74.5 billion'. Opus is lowest on earnings at $30.80, holding the holiday quarter back because 'Brent oil climbed again'.
2026 sales of about $252B, with profit held back by heavy data-center spending and a settlement charge.
Sales$252B$246B – $258B · +26% on FY2025
Adjusted EPS$31.84$28.78 – $34.43 · $16.62 reported so far
Base caseOperating margin30.7%41.4% in FY2025
Investment and cash generation
Base case
Capital spending$136B
Free cash flow−$1.65B
Free cash flow: operating cash flow less capital spending.
The first half brought $117B of sales and $16.62 of earnings per share. The first quarter's $10.44 included an $8.03B tax benefit; without it, the company said, earnings per share would have been $7.31. Adjusted figures on this page remove only the August settlement charge, so that benefit stays in.
The company expects 2026 expenses of $165B to $169B before the settlement, capital spending of $130B to $145B, and operating income above 2025's. Reality Labs, its glasses and headset unit, is expected to lose about as much as in 2025, when it cut operating profit by $19.19B.
The six AI models put 2026 sales between $250B and $256B, averaging $252B, about 26% above 2025's $201B. Their differences come almost entirely from the second half, where they weigh how much oil prices, Europe and harder comparisons slow growth.
Reported operating margin averages about 31%, down from 41% in 2025, because of the settlement charge, depreciation and cloud costs. Capital spending averages $136B, against $69.7B in 2025, so free cash flow is about −$1.7B on average, against $46.1B in 2025.
The capital spending outlook rose through the year: from $115B to $135B in January, to $125B to $145B in April on higher memory prices, then narrowed to $130B to $145B in July.
Adjusted earnings per share average $31.84, from $30.80 for Opus to $33.94 for Muse Wonder. The first half's $16.62, tax benefit included, is fixed; the second half rests on holiday sales and costs near the expense outlook.
Sales by quarter
Q1 2025$42.3B+16%
Q2 2025$47.5B+22%
Q3 2025$51.2B+26%
Q4 2025$59.9B+24%
Q1 2026$56.3B+33%
Q2 2026$60.8B+28%
Q3 2026$63.1B+23%
Q4 2026$72.2B+21%
Each percent is on the same quarter a year earlier. The forecast rows are the models' average.
What it rests on
- Expense outlook
- $165B to $169B for 2026, before the settlement charge
We now expect full year 2026 total expenses to be in the range of $165 to $169 billion.
Second-quarter earnings call, July 2026 - Capital spending outlook
- $130B to $145B, narrowed from $125B to $145B
We anticipate 2026 capital expenditures, including principal payments on finance leases, to be in the range of $130 to $145 billion, narrowed from our prior outlook of $125 to $145 billion.
Second-quarter earnings call, July 2026 - Tax rate outlook
- 15% to 17% for the rest of 2026
Absent any changes to our tax landscape, we expect our tax rate for the remaining quarters of 2026 to be between 15-17%, an increase from our prior outlook of 13-16%.
Second-quarter earnings release, July 2026 - First-quarter tax benefit
- An $8.03B benefit lifted first-quarter earnings
Our tax rate for the quarter was negative 23%, which was favorably impacted by a tax benefit of $8.03 billion.
First-quarter earnings call, April 2026 - Reality Labs losses
- Cut 2025 profit by $19.19B; 2026 expected similar
our investments in Reality Labs reduced our 2025 overall operating profit by approximately $19.19 billion, and we expect our 2026 Reality Labs operating losses to remain similar to 2025.
2025 annual report - Operating income pledge
- 2026 operating income to top 2025's
We continue to expect to deliver operating income this year that is above 2025 operating income.
Second-quarter earnings call, July 2026
Low and high
Basethe middle forecast
Key assumptions
Capital spending of about $136B, inside the $130B to $145B outlook. The company narrowed its outlook in July after raising it on memory prices.5 of 6
- GLM 5.3capital spending at $137.5 billion (middle of $130-145 billion)
- GPT-6.1 Solthe middle uses $134 billion in 2026
- Muse Wondercapital spending at $137.5 billion
- Grok 4.7Capital spending stays inside 130 to 145 billion dollars
- Gemini 4 ArgonHeavy capital spending of 134 billion dollars in 2026
Expenses land in the $165B to $169B range plus the roughly $10B charge. The settlement came after the July expense outlook was set.4 of 6
- Opus 5.5It comes on top of the July expense range, which was set before the agreement
- GLM 5.3I put full-year 2026 total expenses at roughly $176 billion
- GPT-6.1 Soltaking my full-year costs to about $177 billion
- Grok 4.7Full-year costs land near the middle of the July range plus that accrual
The first-quarter tax benefit stays in adjusted earnings. Only the signed settlement charge is removed from adjusted earnings.4 of 6
- GPT-6.1 Solmy 2026 earnings retain the first-quarter $8 billion tax benefit
- GLM 5.3the full-year 2026 adjusted figure of about $32 keeps that benefit in
- Gemini 4 Argonthe first quarter reported earnings per share of 10.44 dollars is carried directly into our full year 2026
- Opus 5.5which still includes the first-quarter tax benefit
Reality Labs loses about $19B, as in 2025. The company expects 2026 losses similar to 2025 and likely the peak.4 of 6
- Opus 5.5I hold Reality Labs losses near $19 billion in 2026
- GPT-6.1 SolI use about $19 billion of losses in 2026
- GLM 5.3I kept roughly $19 billion of Reality Labs losses in my 2026 cost lines
- Grok 4.7so I did not fade those losses in 2026
Third-quarter sales near the top of the outlook and a strong, if muted, holiday quarter. First-half sales of $117B are reported, so only the second half is open.3 of 6
- Opus 5.5Third-quarter revenue is about $63.2 billion, near the top of the $61-64 billion guide
- GPT-6.1 SolSeptember revenue reaches $64 billion and holiday revenue $74.5 billion
- Grok 4.7I expect the third quarter to land near the top of the company 61 to 64 billion dollar revenue range
If all of this happens
$252Bsales
$31.84adjusted EPS
$136BCapital spending
−$1.65BFree cash flow
Cash walk not supplied by every model.
Lowabout 1 chance in 10 of worse
What the low case assumes
Advertisers pull back after the oil spike, in both remaining quarters. Advertisers cut spending when the Iran conflict began early in 2026.3 of 6
- Opus 5.5The autumn oil shock cuts advertiser budgets the way the Iran conflict did in early 2026
- Grok 4.7A lasting energy shock and weak household confidence cut advertiser budgets
- Muse WonderWeaker advertiser demand and pricing
Further legal charges beyond the settlement. Youth trials continue, and nearly 30% of settlement payments are contingent.3 of 6
- GLM 5.3Further material legal accruals from the youth-litigation docket
- Grok 4.7another large legal charge would land
- GPT-6.1 Solother legal or tax losses can still depress results
Spending and costs run at the top of their ranges. Memory and component prices already pushed the spending outlook up twice this year.3 of 6
- GLM 5.3memory and component costs push capital spending to the top of the range
- Grok 4.7Spending on data centers would stay at the top of the company range
- Muse Wondercosts stay at the top of the range
If all of this happens
$246Bsales · −3% vs base
$28.78EPS · −10% vs base
2026 sales would have to slip to about $246B, 2.5% under the base, with earnings per share near $28.78, 9.6% lower. That takes a third quarter at the bottom of the outlook and a holiday quarter hit by the oil spike, with costs at the top of their range. Muse Wonder's sales low is furthest at $242B, adding 'new legal charges' to weak demand. Sol and Grok go furthest on earnings, at $27.00: Sol because 'other legal or tax losses can still depress results', Grok because 'another large legal charge would land'.
2025 sales grew 22% to $201B. Half of 2026 is already reported, $117B of sales, which keeps the full-year range narrow.
The six lows run from $242B (Muse Wonder) to $250B (GPT-6.1 Sol).
First signThird-quarter sales near or below $61.5B in the 21 October 2026 release.
Highabout 1 chance in 10 of better
What the high case assumes
Ad demand and performance hold near the first half's pace. The second quarter grew 28% with the price per ad up 12%.3 of 6
- Grok 4.7Ad demand and performance gains hold near the top of recent growth
- GLM 5.3Ads performance gains compounding faster than expected
- GPT-6.1 SolStronger engagement and sustained advertising pricing
Paid messaging and business agents scale faster. Other app revenue grew 73% in the second quarter.3 of 6
- GPT-6.1 SolFaster monetization of messaging and business agents
- Grok 4.7Paid messaging, subscriptions, and business agents scale faster than expected
- Muse WonderNew revenue from business agents, subscriptions, paid messaging and AI glasses
Spending and legal costs come in lighter than the base. The capital spending outlook still runs as low as $130B.2 of 6
- Grok 4.7Spending and legal costs come in lighter than in the base case
- GPT-6.1 SolBetter operating leverage and narrowing hardware losses
If all of this happens
$258Bsales · +2% vs base
$34.43EPS · +8% vs base
2026 sales would have to reach about $258B, 2.4% above the base, with earnings per share near $34.43, 8.2% higher. Both remaining quarters would need to beat, with third-quarter sales above $64B and spending toward the low end of its ranges. Sol's sales high is furthest at $263B, because it alone has 'September sales exceed guidance' and a holiday quarter nearing $80 billion. Muse Wonder's earnings high is furthest at $36.50, from its higher margin, 33% in its base against 31% on average.
The company has recently landed near the top of its sales outlooks: second-quarter sales of $60.8B sat just under the top of $58B to $61B.
The six highs run from $256B (GLM 5.3) to $263B (GPT-6.1 Sol).
First signThird-quarter sales above $64B in the 21 October 2026 release.
Watch
The 21 October 2026 release: third-quarter sales, the fourth-quarter outlook, and any change to the $165B to $169B expense or $130B to $145B spending ranges. A new legal accrual or a weak holiday outlook points lower; sales above $64B point higher.
Sales
PricedIn$296B$269B – $324B
$269B$296B$324B
adjusted EPS
PricedIn$31.43$23.33 – $39.25
$23.33$31.43$39.25
Middlethe forecast80% range1 in 10 chance below this bar, 1 in 10 above it. These are the low and high printed by each number.
All see 2027 sales growth slowing, to between 15% and 20%. Opus is lowest at $290B, holding that 'expenses grow about 23% as depreciation catches up with capital spending'. Sol is highest at $307B, with 'about 20% total sales growth in 2027'. Sol's range is much the widest: its $268B low has 'Committed computing costs and depreciation outrun revenue'; its $347B high has 'new services profitably use the computing capacity'.
Sales grow about 17% to $296B in 2027, but earnings per share stay roughly flat as data-center costs catch up.
Sales$296B$269B – $324B · +17% on FY2026
Adjusted EPS$31.43$23.33 – $39.25 · −1% on FY2026
Base caseOperating margin33.0%30.7% in FY2026
Investment and cash generation
Base case
Capital spending$163B
Free cash flow−$7.33B
Free cash flow: operating cash flow less capital spending.
The company has given no outlook for 2027 yet; GLM expects one with the January 2027 results. It has said its plans are geared to maximizing 2026 and 2027 capacity, and at the end of June it held $349.31B of non-cancelable commitments, $81.65B of them due in 2027.
The six AI models put 2027 sales between $290B and $307B, averaging $296B, about 17% above 2026. All expect ad growth to slow as this year's gains are lapped, with new paid products such as business agents and subscriptions adding a few billion dollars.
Costs are the crux. Capital spending averages $163B, against $136B in 2026, and servers depreciate over five to 5.5 years. So expenses grow faster than sales, and adjusted earnings per share average $31.43, about 1.3% below 2026, which still holds the one-time first-quarter tax benefit.
Operating margin recovers to about 33% from about 31%, since the settlement charge does not recur. Free cash flow averages −$7.3B, from −$20.0B for Muse Wonder to $6.0B for Grok, depending mostly on how far they let capital spending climb.
The ranges are wide: sales from $269B to $324B on average, earnings per share from $23.33 to $39.25. Sol's range is much the widest.
2027 earnings per share turn on whether sales outrun depreciation and cloud costs. They average $31.43; Opus expects a dip to $28.60 as expenses rise about 23%, Muse Wonder a rise to $35.40.
Sales by year
FY2023$135B
FY2024$165B+22%
FY2025$201B+22%
FY2026$252B+26%
FY2027$296B+17%
FY2028$340B+15%
The forecast years are the models' average.
Non-cancelable commitments at the end of June 2026
Total commitments$349B
Due in rest of 2026$54B
Due in 2027$82B
Contingent cloud purchases$15B
Where each number comes from
- Total commitments
As of June 30, 2026, we had $349.31 billion of non-cancelable contractual commitments, comprising both short-term and long-term arrangements. These commitments mostly relate to third-party cloud capacity arrangements and investments in servers and network infrastructure, data centers, and consumer hardware products in Reality Labs, with approximately $53.52 billion and $81.65 billion due in 2026 and 2027, respectively. In addition, as of June 30, 2026, we have contingent obligations to purchase up to $14.72 billion of cloud capacity over a five-year period, which may be reduced if the cloud service provider is able to sell such capacity to other customers. For agreements with variable terms, we do not estimate the total obligation beyond minimum quantities and/or pricing, as of the reporting date.
- Due in rest of 2026
As of June 30, 2026, we had $349.31 billion of non-cancelable contractual commitments, comprising both short-term and long-term arrangements. These commitments mostly relate to third-party cloud capacity arrangements and investments in servers and network infrastructure, data centers, and consumer hardware products in Reality Labs, with approximately $53.52 billion and $81.65 billion due in 2026 and 2027, respectively. In addition, as of June 30, 2026, we have contingent obligations to purchase up to $14.72 billion of cloud capacity over a five-year period, which may be reduced if the cloud service provider is able to sell such capacity to other customers. For agreements with variable terms, we do not estimate the total obligation beyond minimum quantities and/or pricing, as of the reporting date.
- Due in 2027
As of June 30, 2026, we had $349.31 billion of non-cancelable contractual commitments, comprising both short-term and long-term arrangements. These commitments mostly relate to third-party cloud capacity arrangements and investments in servers and network infrastructure, data centers, and consumer hardware products in Reality Labs, with approximately $53.52 billion and $81.65 billion due in 2026 and 2027, respectively. In addition, as of June 30, 2026, we have contingent obligations to purchase up to $14.72 billion of cloud capacity over a five-year period, which may be reduced if the cloud service provider is able to sell such capacity to other customers. For agreements with variable terms, we do not estimate the total obligation beyond minimum quantities and/or pricing, as of the reporting date.
- Contingent cloud purchases
As of June 30, 2026, we had $349.31 billion of non-cancelable contractual commitments, comprising both short-term and long-term arrangements. These commitments mostly relate to third-party cloud capacity arrangements and investments in servers and network infrastructure, data centers, and consumer hardware products in Reality Labs, with approximately $53.52 billion and $81.65 billion due in 2026 and 2027, respectively. In addition, as of June 30, 2026, we have contingent obligations to purchase up to $14.72 billion of cloud capacity over a five-year period, which may be reduced if the cloud service provider is able to sell such capacity to other customers. For agreements with variable terms, we do not estimate the total obligation beyond minimum quantities and/or pricing, as of the reporting date.
Mostly cloud capacity, servers and data centers; costs the company cannot quickly switch off if sales slow.
What it rests on
- Commitments due 2027
- $81.65B of non-cancelable commitments fall due in 2027
As of June 30, 2026, we had $349.31 billion of non-cancelable contractual commitments, comprising both short-term and long-term arrangements. These commitments mostly relate to third-party cloud capacity arrangements and investments in servers and network infrastructure, data centers, and consumer hardware products in Reality Labs, with approximately $53.52 billion and $81.65 billion due in 2026 and 2027, respectively. In addition, as of June 30, 2026, we have contingent obligations to purchase up to $14.72 billion of cloud capacity over a five-year period, which may be reduced if the cloud service provider is able to sell such capacity to other customers. For agreements with variable terms, we do not estimate the total obligation beyond minimum quantities and/or pricing, as of the reporting date.
Second-quarter 2026 quarterly report - Capacity plans
- Plans geared to maximizing 2026 and 2027 capacity
Consequently, our current plans are geared towards maximizing 2026 and 2027 capacity. When we have had incremental capacity in the past, it has proven extremely valuable in scaling experiences like reels, and we are confident that this will be true in this timeframe as well. Longer term, it's harder to predict the exact usage scaling curves, but we believe that our distribution advantages will give us the opportunity to serve AI products that are valuable for everyone, both our 3.6 billion users and millions of businesses.
Company earnings call - Server depreciation life
- Servers and network assets depreciate over five to 5.5 years
Servers and network assets Five to 5.5 years
2025 annual report - No margin target
- The company does not manage to a target margin
On your second question, we don’t manage to a target margin. We believe that
Fourth-quarter 2025 follow-up call, January 2026 - Automated ad tools
- Advantage Plus tools above $75B annual revenue run rate
Our AI-powered Advantage Plus end-to-end solutions continue to grow, reaching over $75 billion in annual revenue run rate this quarter.
Second-quarter earnings call, July 2026
Low and high
Basethe middle forecast
Key assumptions
Capital spending rises to about $163B, from about $136B. Plans are geared to maximizing 2026 and 2027 capacity.6 of 6
- Opus 5.5capital spending rising to about $165 billion in 2027
- GLM 5.3about $170 billion in 2027
- GPT-6.1 Sol$170 billion in 2027
- Muse WonderI therefore keep capital spending very high in 2027
- Gemini 4 Argon150 billion dollars in 2027
- Grok 4.7They also said they remain supply constrained through 2027, which keeps 2027 spending high
Ad growth slows toward the mid-to-high teens as this year's gains are lapped. Management named harder comparisons and Europe's rules as reasons growth is easing.4 of 6
- Grok 4.7In 2027 and 2028 advertising growth slows into the mid-teens and then the low teens
- GLM 5.3In 2027 revenue grows about 16 percent to roughly $292 billion
- Opus 5.5ad growth slowing from 28% to about 18-20% as comparisons get harder
- GPT-6.1 Solabout 20% total sales growth in 2027
Depreciation and cloud costs grow faster than sales, holding earnings roughly flat. Servers depreciate over about five years, so this year's spending hits profit from 2027.4 of 6
- Grok 4.7That is why 2027 earnings grow little despite mid-teens sales growth
- GLM 5.3why depreciation holds 2027 earnings per share roughly flat despite 16 percent revenue growth
- Opus 5.5That makes depreciation the main reason 2027 expenses grow about 23%
- GPT-6.1 SolRising depreciation and third-party computing costs offset much of the sales growth
New paid products add a few billion dollars. Business agent charging began in August 2026.3 of 6
- GLM 5.3gave them a few billion dollars by 2027-28 in the middle
- GPT-6.1 Solroughly $8 billion in 2027
- Grok 4.7newer paid products add only a little
Tax near 16% to 17%, with no repeat of the 2026 benefit. The 15% minimum tax now limits share-based tax benefits.2 of 6
- Opus 5.5So I hold the tax rate at 16-17% in 2027 and 2028
- GPT-6.1 SolI do not repeat the benefit in 2027
If all of this happens
$296Bsales
$31.43adjusted EPS
$163BCapital spending
−$7.33BFree cash flow
Cash walk not supplied by every model.
Lowabout 1 chance in 10 of worse
What the low case assumes
Ad prices stall as the energy shock and weak consumers cut budgets. Advertisers cut spending when the Iran conflict began early in 2026.4 of 6
- Opus 5.5advertiser pullback from the energy shock and weak consumer sentiment
- GPT-6.1 SolWeaker advertiser demand and smaller pricing gains
- Grok 4.7hold growth in 2027 and 2028 to the high single digits
- Gemini 4 ArgonMacroeconomic pressure from high oil prices and elevated interest rates
Legal charges beyond the settlement land in reported earnings. Youth cases and mass arbitrations remain open.3 of 6
- Opus 5.5new legal charges beyond the $10 billion accrual
- GLM 5.3Further material legal accruals from the youth-litigation docket
- Grok 4.7A worse European ads rule or another large legal charge
Committed computing costs outrun revenue. $81.65B of non-cancelable commitments fall due in 2027.3 of 6
- GPT-6.1 SolCommitted computing costs and depreciation outrun revenue
- Gemini 4 ArgonRapid growth in server depreciation and third-party cloud commitments outpacing revenue
- Opus 5.5depreciation from $130-185 billion a year of capital spending
Europe imposes further limits on personalized ads. The European decision is under appeal, and further changes may be imposed.2 of 6
- Opus 5.5further European restrictions on personalized ads
- Gemini 4 ArgonStricter European Union regulatory enforcement on less personalized ads
If all of this happens
$269Bsales · −9% vs base
$23.33EPS · −26% vs base
Sales would have to stall at about $269B, 9.2% under the base and about 6.5% growth, with earnings per share near $23.33, 26% lower. That takes ad prices stalling after the oil shock, further European limits and costs locked in by commitments. Opus's sales low is furthest at $260B: 'Ad prices stall while impressions keep growing, so 2027 sales grow only about 3-4%', and its margin 'sinks into the mid-20s'. Sol's earnings low is furthest at $18.00, because 'Committed computing costs and depreciation outrun revenue' while only the settlement is removed from earnings.
Opus notes the 2022 slowdown had special causes: Apple's tracking changes, a strong dollar and a Reels mix shift. Today $81.65B of commitments fall due in 2027, so costs cannot quickly shrink.
The six lows run from $260B (Opus 5.5) to $277B (GLM 5.3).
First signA fourth-quarter outlook implying growth under 15% in the 21 October 2026 release.
Highabout 1 chance in 10 of better
What the high case assumes
Ad performance gains from the new ranking systems keep growth near 25%. The company says it sees further room to improve recommendations into 2027.3 of 6
- Opus 5.5ad performance gains from the new ranking systems
- GLM 5.3Ads performance gains compounding faster than expected
- Gemini 4 ArgonUnified foundation models across organic content and advertising
New paid products become a visible revenue line. Business agents, subscriptions and the model API began charging in 2026.3 of 6
- Opus 5.5new paid products: business agents, subscriptions, the model API, and compute sales
- GLM 5.3New AI revenue lines scaling quickly
- GPT-6.1 SolFaster monetization of messaging and business agents
Expenses grow no faster than sales. The May job cuts and AI coding tools hold headcount costs down.3 of 6
- Opus 5.5operating leverage from a leaner workforce
- Gemini 4 ArgonStrong engineering productivity gains from agentic coding tools
- GPT-6.1 SolBetter operating leverage and narrowing hardware losses
If all of this happens
$324Bsales · +9% vs base
$39.25EPS · +25% vs base
Sales would have to reach about $324B, 9.3% above the base and about 28% growth, with earnings per share near $39.25, 25% higher. That takes the new ranking systems keeping growth near 25%, new paid products scaling, and expenses growing no faster than sales. Sol's high is furthest, at $347B and $46.00, because it alone has 'Profitable use of added capacity rather than idle infrastructure' and 'glasses losses narrow more quickly'; GLM caps at $306B, treating such a path as 'the upside tail'.
Second-quarter 2026 sales grew 28%, with the price per ad up 12%. Other app revenue reached $1 billion a quarter for the first time, up 73%.
The six highs run from $306B (GLM 5.3) to $347B (GPT-6.1 Sol).
First signThird-quarter sales above $64.5B on 21 October 2026, with a fourth-quarter outlook implying growth above 22%.
Watch
The 21 October 2026 release, then the first 2027 expense and capital spending outlook, which GLM expects in January 2027. Capital spending above $180B or expenses growing above 30% point lower; disclosed business agent revenue points higher.
Sales
PricedIn$340B$294B – $387B
$294B$340B$387B
adjusted EPS
PricedIn$37.03$24.00 – $48.42
$24.00$37.03$48.42
Middlethe forecast80% range1 in 10 chance below this bar, 1 in 10 above it. These are the low and high printed by each number.
All expect 2028 sales growth of 13% to 18%. Opus is lowest at $328B in sales and $31.90 in earnings, keeping capital spending high so 'Free cash flow stays near zero'. Sol is highest at $363B, with 'about 20% total sales growth in 2027 and 18% in 2028'. Sol's range is much the widest: at its $280B low 'low-case earnings can decline despite growing sales'; its $435B high needs 'new services profitably use the computing capacity'.
Earnings per share return to growth in 2028, about 18%, as sales outpace a slowing data-center build-out.
Sales$340B$294B – $387B · +15% on FY2027
Adjusted EPS$37.03$24.00 – $48.42 · +18% on FY2027
Base caseOperating margin34.2%33.0% in FY2027
Investment and cash generation
Base case
Capital spending$166B
Free cash flow$17.0B
Free cash flow: operating cash flow less capital spending.
2028 lies beyond any company outlook. Management has said its visibility declines for 2028 and beyond, and Grok reads its plan for that year as flexibility in land and power rather than locking in chips now.
The six AI models put 2028 sales between $328B and $363B, averaging $340B, about 15% above 2027. Ad growth slows into the low or mid teens, while new paid products grow from a small base.
Earnings per share average $37.03, up about 18% on 2027, because capital spending levels off: it averages $166B against $163B in 2027. Operating margin edges up to about 34%, and free cash flow turns positive at about $17.0B on average.
The spread is wide. Grok cuts capital spending to $135B and expects free cash flow of $38.0B; Sol raises it to $185B. Average earnings per share run from $24.00 in the low case to $48.42 in the high, and Sol's own range from $13.00 to $60.00.
Reality Labs losses are expected to ease gradually from a 2026 peak, and the 15% minimum tax keeps the tax rate near 16% to 17%.
2028 earnings per share rest on sales outpacing depreciation once the build-out levels off. They average $37.03, from $31.90 for Opus, which holds margin near 31%, to $42.50 for Muse Wonder.
Each model's growth in FY2028
Opus 5.5+13%
GPT-6.1 Sol+18%
Grok 4.7+13%
GLM 5.3+13%
Gemini 4 Argon+16%
Muse Wonder+15%
Growth of each model's middle on its own forecast for the year before.
What it rests on
- Visibility for 2028
- Management says visibility declines for 2028 and beyond
visibility declines as you get to ‘28 and beyond. And that’s why the ‘28 and
Second-quarter follow-up call, July 2026 - Frontloading capacity
- Build capacity early; spare compute goes to the core ads business
Some people think that we'll get there in a few years. Others think it will be 5, 7 years or longer. I think that it's the right strategy to aggressively frontload building capacity so that way we're prepared for the most optimistic cases. That way, if superintelligence arrives sooner, we will be ideally positioned for a generational paradigm shift in many large opportunities. If it takes longer, then we'll use the extra compute to accelerate our core business which continues to be able to profitably use much more compute than we've been able to throw at it.
Third-quarter 2025 earnings call - Reality Labs peak
- 2026 losses likely the peak, then gradually lower
I expect Reality Labs losses this year to be similar to last year, and this will likely be the peak as we start to gradually reduce our losses going forward while continuing to execute on our vision.
Fourth-quarter 2025 earnings call, January 2026 - Automated ad tools
- Advantage Plus tools above $75B annual revenue run rate
Our AI-powered Advantage Plus end-to-end solutions continue to grow, reaching over $75 billion in annual revenue run rate this quarter.
Second-quarter earnings call, July 2026
Low and high
Basethe middle forecast
Key assumptions
Sales grow about 15%, slowing as the business gets larger. Ranking gains are harder to repeat at scale, and visibility for 2028 is limited.4 of 6
- Grok 4.7In 2027 and 2028 advertising growth slows into the mid-teens and then the low teens
- GPT-6.1 Sol18% in 2028
- GLM 5.3by 2028 revenue reaches about $330 billion
- Gemini 4 Argon345.0 billion dollars in 2028
Capital spending levels off, averaging $166B. Management's 2028 plan leans on flexibility rather than locking in chips now.4 of 6
- Grok 4.7I therefore cut capital spending in 2028 to 135 billion dollars
- Muse Wonderonly step it down slightly in 2028
- GLM 5.3$180 billion in 2028
- Opus 5.5$175 billion in 2028
Operating margin edges up to about 34% as Reality Labs losses ease. Management called 2026 the likely peak for Reality Labs losses.3 of 6
- Gemini 4 Argonrecovering to 33.0 percent in 2027 and 35.0 percent in 2028 as revenue scales and Reality Labs losses moderate
- GPT-6.1 SolThe middle operating margin recovers only to about 32% and 34%
- GLM 5.3earnings per share about $35 as margins stabilise
New paid products grow into a meaningful revenue line. Business agent charging and subscriptions began in the second half of 2026.3 of 6
- GPT-6.1 Sol$14 billion in 2028
- GLM 5.3gave them a few billion dollars by 2027-28 in the middle
- Muse Wonderthe larger contribution only by 2028 in the middle case
Tax stays near 16% to 17%. The 15% minimum tax limits share-based tax benefits.2 of 6
- Opus 5.5So I hold the tax rate at 16-17% in 2027 and 2028
- GPT-6.1 Sol17% in 2027–2028
If all of this happens
$340Bsales
$37.03adjusted EPS
$166BCapital spending
$17.0BFree cash flow
Cash walk not supplied by every model.
Lowabout 1 chance in 10 of worse
What the low case assumes
Ad growth falls to the high single digits or lower. Higher oil prices and interest rates weigh on advertisers' budgets.3 of 6
- Grok 4.7hold growth in 2027 and 2028 to the high single digits
- Opus 5.5Ad prices stall while impressions keep growing
- Gemini 4 Argonslow global consumer spending and advertiser budgets in 2027 and 2028
Committed computing costs and depreciation outrun revenue. Commitments and data-center leases run for many years and cannot be switched off.3 of 6
- GPT-6.1 SolCommitted computing costs and depreciation outrun revenue
- Gemini 4 ArgonRapid growth in server depreciation and third-party cloud commitments outpacing revenue
- Opus 5.5depreciation from $130-185 billion a year of capital spending
Further legal and regulatory charges in the United States and Europe. Damages sought in youth cases range up to the high tens of billions of dollars.3 of 6
- Muse WonderFurther legal and regulatory charges in the United States and Europe
- Opus 5.5new legal charges beyond the $10 billion accrual
- GLM 5.3Further material legal accruals from the youth-litigation docket
If all of this happens
$294Bsales · −14% vs base
$24.00EPS · −35% vs base
Sales would have to stall near $294B, 14% under the base and about flat on 2027, with earnings per share near $24.00, 35% lower. Ad growth would need to fall to the high single digits while depreciation and committed cloud costs keep rising. Opus and Sol go furthest on sales, at $282B and $280B. Sol's earnings low is furthest at $13.00, because it builds the most capital spending, $185B in its base, and holds that 'low-case earnings can decline despite growing sales'.
Grok put a path of about 8% growth in 2028 near its own low case; GLM judged an outcome that weak closer to 1-in-20 than 1-in-10.
The six lows run from $280B (GPT-6.1 Sol) to $312B (GLM 5.3).
First signA first 2027 outlook, expected with the January 2027 results, implying expense growth above 30% or capital spending above $180B.
Highabout 1 chance in 10 of better
What the high case assumes
Ad growth stays above 20% on better recommendation models. Ranking gains lifted conversions, and the company sees further headroom into 2027.3 of 6
- Gemini 4 ArgonNext-generation unified recommendation models and generative ad creative tools unlock sustained twenty-plus percent advertising revenue growth through 2028
- Opus 5.5keep growth near 25% into 2027
- GPT-6.1 SolRecommendation improvements keep compounding
Business agents, the model API and compute sales add several billion dollars. The company says it gets offers for compute at a significant premium.3 of 6
- Gemini 4 Argonscales into a multi-billion dollar high-margin revenue stream
- Opus 5.5New revenue lines (business agents, subscriptions, the model API, and selling compute at a premium) add several billion dollars
- GLM 5.3add several billion dollars by 2027-28
Margins rise as Reality Labs losses fall and headcount stays flat. Management called 2026 the likely peak for Reality Labs losses.3 of 6
- Gemini 4 Argondriving operating margins above 38 percent
- GPT-6.1 SolBetter operating leverage and narrowing hardware losses
- Muse WonderReality Labs losses gradually falling after 2026
If all of this happens
$387Bsales · +14% vs base
$48.42EPS · +31% vs base
Sales would have to reach about $387B, 14% above the base and about 31% growth, with earnings per share near $48.42, 31% higher. Growth would need to stay above 20% on ad gains, with business agents, the model API and compute sales becoming large, and margins rising. Sol is furthest at $435B and $60.00, because it alone lets 'subscriptions and business agents scale faster' while new services use the capacity profitably; GLM caps at $358B, calling such growth 'demanding'. Gemini's high has margins 'above 38 percent'.
Opus's high reaches 2028 sales of $375B and earnings per share of about $41, from new revenue lines and a leaner workforce.
The six highs run from $358B (GLM 5.3) to $435B (GPT-6.1 Sol).
First signDisclosed revenue from business agents or compute sales, possibly as early as the 21 October 2026 release.
Watch
The first 2027 capital spending plan, the pace of business agent and subscription revenue, and Reality Labs losses. Capital spending leveling off near 2027's level points higher; another step-up toward $185B, or new legal charges, points lower.