# MSFT Full forecast · Q1 FY27

> A newer forecast, cutoff 30 October 2026, replaces this one. This page stays as published.

Canonical: https://pricedin.markets/msft/fq1-2027/forecast/

cutoff 27 October 2026, 06:00 ET

A newer forecast, cutoff 30 October 2026, replaces this one. This page stays as published.

[Full forecast, Q2 FY27 →](https://pricedin.markets/msft/fq2-2027/forecast/)

MSFT · as of 2026-10-27 06:00:22 ET · 6 AI forecasters · 80% ranges, 1 in 10 to 9 in 10

## Sales and earnings outlook

Next quarter · Q1 FY27

Sales $91.9B +1% above the $89.8B–$91.0B outlook

Adjusted EPS $4.84 +17% on Q1 FY26

Quarter after · Q2 FY27

Sales $95B +17% on Q2 FY26

Adjusted EPS $4.88 +18% on Q2 FY26

This year · FY2027

Sales $389B +17% on FY2026

Adjusted EPS $19.44

Next year · FY2028

Sales $452B +16% on FY2027

Adjusted EPS $22.58 +16% on FY2027

Year +2 · FY2029

Sales $514B +14% on FY2028

Adjusted EPS $25.93 +15% on FY2028

##### Sales

PricedIn **$91.9B** $88.9B – $94.2B

$88.9B $90.5B $91.9B $93.0B $94.2B

##### adjusted EPS

PricedIn **$4.84** $4.48 – $5.16

$4.48 $4.66 $4.84 $5.00 $5.16

**Middle** the forecast **50% range** 1 in 4 chance the result comes in below this bar, 1 in 4 above it. Half the time it lands inside. **80% range** 1 in 10 chance below this line, 1 in 10 above it. These are the low and high printed by each number.

Five of six sit above the guided midpoint, and all within a few billion dollars of each other. Muse Spark Open, lowest at $90.4B, takes the range at face value: it 'used mid-point 90.4B' to set its plan. Kimi, highest at $93.0B, expects Microsoft 'to keep beating conservatively framed quarterly guidance more often than not', and placed its middle above the range because the company beat its recent ranges 'across all three segments'.

The models expect first-quarter fiscal 2027 sales of about $91.9B, above the company's own range, as new cloud capacity is sold quickly.

Sales **$91.9B** +18% on Q1 FY26 ($77.7B)

*50% range* $90.5B – $93.0B *80% range* $88.9B – $94.2B

Adjusted EPS **$4.84** +17% on Q1 FY26 ($4.13)

*50% range* $4.66 – $5.00 *80% range* $4.48 – $5.16

#### Investment and cash generation

Base case

Capital spending **$45.5B**

Free cash flow **$6.75B**

Free cash flow: operating cash flow less capital spending.

Microsoft closed fiscal 2026 with fourth-quarter sales of $90.0B, up 18% on a year earlier, and earnings per share of $4.74. Sales for the year reached $332B. Capital spending rose to $116B for the year, and free cash flow slipped to $67.0B as that spending climbed.

For the July to September quarter the company guided revenue of $89.85B to $90.95B, with Azure, its cloud platform, growing about 45% in constant currency. It says customer demand still exceeds the capacity it can supply, and that personal computer sales will fall as component prices rise.

Five of the six models put the middle above the guided midpoint of $90.4B, and the average is $91.9B. Their reason is a record of beats: the June quarter landed at $90.0B against a range of $86.7B to $87.8B.

They differ on the size of the beat. Muse Spark Open stays at the midpoint; Sol sits about $2.1B above it, led by capacity becoming usable sooner. Opus trimmed its middle, judging that part of June's personal computing strength came from partners building stock.

Spending stays heavy: the models put capital spending near $45.5B for the quarter, leaving free cash flow of only about $6.8B on average.

Earnings per share, excluding the OpenAI investment impact, average $4.84, up 17% on a year earlier. Costs are expected near the company's guidance, with a tax rate of about 20%.

#### Outlook against result

Q2 FY26 +0.8% above $79.5B–$80.6B → $81.3B guided “USD 79.5 billion to USD 80.6 billion”

Q3 FY26 +1.4% above $80.7B–$81.8B → $82.9B guided “$80.65 to $81.75 billion”

Q4 FY26 +2.5% above $86.7B–$87.8B → $90.0B guided “between USD 86.7 billion and USD 87.8 billion”

Q1 FY27 +1.0% above $89.8B–$91.0B → $91.9B guided “between 89.85 and 90.95 billion U.S. dollars”

Each outlook is the one the company gave before the quarter. The last row is the models' forecast.

#### The quarter forecast, line by line

| Line | Q4 FY26 reported | Q1 FY26 reported | Q1 FY27 forecast | Q2 FY26 reported | Q2 FY27 forecast |
| --- | --- | --- | --- | --- | --- |
| Revenue | $90.0B | $77.7B | $91.9B | $81.3B | $95.3B |
| Sales growth, year on year | +18% | +18% | +18% | +17% | +17% |
| Intelligent Cloud | $39.3B | $30.9B | $41.9B | $32.9B | $43.9B |
| More Personal Computing | $12.9B | $13.8B | $12.6B | $14.2B | $12.8B |
| Productivity and Business Processes | $37.8B | $33.0B | $37.4B | $34.1B | $38.5B |
| Cost of revenue | $29.5B | $24.0B | $29.9B | $26.0B | $31.7B |
| Gross margin | $60.5B | $53.6B | $62.0B | $55.3B | $63.6B |
| Gross margin | 67.2% | 69.0% | 67.4% | 68.0% | 66.8% |
| Research and development | $10.00B | $8.15B | $8.92B | $8.50B | $9.40B |
| Sales and marketing | $7.59B | $5.72B | $6.07B | $6.58B | $6.70B |
| General and administrative | $2.29B | $1.81B | $1.93B | $1.93B | $2.02B |
| Operating income | $40.6B | $38.0B | $45.0B | $38.3B | $45.5B |
| Operating margin | 45.1% | 48.9% | 49.0% | 47.1% | 47.8% |
| Other income (expense), net | $3.44B | −$3.66B | −$0.18B | $9.97B | −$0.23B |
| Income before income taxes | $44.0B | $34.3B | $44.9B | $48.2B | $45.3B |
| Provision for income taxes | $8.28B | $6.55B | $8.97B | $9.79B | $9.05B |
| Tax rate | 18.8% | 19.1% | 20.0% | 20.3% | 20.0% |
| Net income | $35.8B | $27.7B | $35.9B | $38.5B | $36.2B |
| Diluted | 7.44B | 7.47B | 7.44B | 7.46B | 7.43B |
| Adjusted EPS | $4.74 | $4.13 | $4.84 | $4.14 | $4.88 |

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.

#### Quarterly revenue through fiscal 2026

Dec 2025 quarter $81.3B

Mar 2026 quarter $82.9B

Jun 2026 quarter $90.0B

**Where each number comes from**

**Dec 2025 quarter**

“This quarter, revenue was $81.3 billion, up 17% in constant currency.”

**Mar 2026 quarter**

“This quarter, revenue was $82.9 billion, up 18% and 15% in constant currency.”

**Jun 2026 quarter**

“This quarter, revenue was $90 billion, up 18 percent and 17 percent in constant currency.”

Each quarter came in above the range the company had given, the record behind the models' beat.

#### What it rests on

**Quarter revenue outlook**

$89.85B to $90.95B, growth of 16 to 17 percent “Therefore, at the total company level, revenue should be between 89.85 and 90.95 billion U.S. dollars, or growth of 16 to 17 percent, with accelerating commercial growth, partially offset by the impact from the PC market dynamics noted earlier.” July 2026 earnings call

**Azure outlook**

About 45% growth in constant currency, demand exceeding supply “In Azure, we expect revenue growth of approximately 45% in constant currency, and we remain focused on delivering efficiencies that help us bridge the gaps we see as customer demand continues to exceed supply.” July 2026 earnings call

**June quarter revenue**

$90.0B, up 18% “This quarter, revenue was $90 billion, up 18 percent and 17 percent in constant currency.” July 2026 earnings call

**June quarter outlook**

$86.7B to $87.8B, beaten by the result “Therefore, at the total company level, revenue should be between USD 86.7 billion and USD 87.8 billion or growth of 13% to 15%, with accelerating commercial growth, partially offset by our consumer business.” April 2026 earnings call

**Personal computing outlook**

$12.2B to $12.7B, hit by component prices and inventory “In more personal computing, we expect revenue to be $12.2 to $12.7 billion U.S. dollars as we continue to lap the strong prior year comparables noted earlier and navigate complex PC market dynamics impacted by component prices and inventory levels.” July 2026 earnings call

**Cost outlook**

Cost of revenue $29.6B to $29.8B; operating expense $16.8B to $16.9B “We expect COGS of 29.6 to 29.8 billion U.S. dollars, or growth of 23 to 24 percent, An operating expense of $16.8 to $16.9 billion U.S. dollars, a growth of 7 to 8 percent, driven by continued investment in R&D compute capacity and talent.” July 2026 earnings call

#### Low and high

Base the middle forecast

Key assumptions

- ****Sales land above the guided midpoint, repeating the company's habit of beating its own range.** June revenue of $90.0B beat a range of $86.7B to $87.8B. *5 of 6***
  - **Opus 5.5** I put the middle about 1.8 billion dollars above its midpoint
  - **DeepSeek V4.1 Flash** because the company finished above the top of its own range in each of the last three quarters
  - **GLM 5.3** shaded revenue slightly above the midpoint because of the company's persistent beats
  - **Kimi K3.1 Pro** placed the middle above it because Microsoft exceeded the preceding several quarterly revenue ranges across all three segments
  - **GPT-6.1 Sol** My September-quarter middle is $92.5 billion, about $2.1 billion above the guidance midpoint.
- ****Azure grows about 45% as new capacity is sold as soon as it arrives.** The company says demand still exceeds supply, and extra capacity in June was quickly sold. *5 of 6***
  - **Opus 5.5** Azure growth in the mid 40s as capacity comes online
  - **GLM 5.3** Azure grows about 45% in constant currency in the first quarter
  - **Kimi K3.1 Pro** Azure capacity additions and roughly mid-forties near-term growth
  - **DeepSeek V4.1 Flash** Azure and AI capacity arriving faster than planned, keeping Intelligent Cloud above the guided range
  - **GPT-6.1 Sol** Faster delivery and better utilization of cloud capacity
- ****Costs land near guidance, with a tax rate of about 20%.** Costs have tracked the company's guidance even in quarters when revenue beat. *5 of 6***
  - **Opus 5.5** Costs have landed near guidance even when revenue beat
  - **Muse Spark Open** COGS 29.6-29.8B and opex 16.8-16.9B directly set quarter plan lines and margins
  - **GLM 5.3** I built my statement lines on these
  - **DeepSeek V4.1 Flash** a 20 percent tax rate
  - **Kimi K3.1 Pro** Adjusted earnings use negative $100 million of quarterly other income and expense
- ****Personal computing shrinks but lands near the top of its $12.2B to $12.7B range.** Component prices and partner inventory weigh on personal computer demand, but the segment beat in recent quarters. *3 of 6***
  - **Opus 5.5** personal computing lands near the top of its guided range
  - **GPT-6.1 Sol** The September-quarter plan assumes a smaller shortfall than guidance in that segment
  - **DeepSeek V4.1 Flash** I take more personal computing to 12.7 billion dollars in the September quarter

If all of this happens

**$91.9B** sales

**$4.84** adjusted EPS

**$45.5B** Capital spending

**$6.75B** Free cash flow

Cash walk not supplied by every model.

Low about 1 chance in 10 of worse

What the low case assumes

- ****Personal computer and Xbox sales fall faster than the company guided.** Rising component prices raise device prices, and June's segment strength partly came from partners building stock. *4 of 6***
  - **Opus 5.5** Windows and Xbox revenue falling faster than guided
  - **Muse Spark Open** PC and Xbox weakness deepens
  - **Kimi K3.1 Pro** Personal-computing declines and higher component costs
  - **GPT-6.1 Sol** Weak personal-computer and gaming spending
- ****Azure growth falls short of the about 45% guide as capacity arrives late.** The company calls demand for cloud and AI difficult to forecast, and capacity timing moves Azure each quarter. *3 of 6***
  - **Opus 5.5** Azure growth limited by late capacity or softer AI demand
  - **GPT-6.1 Sol** Delayed power and equipment availability
  - **Kimi K3.1 Pro** Slower Azure consumption and delayed capacity monetization
- ****Infrastructure costs squeeze margins harder than planned.** Cost of revenue is guided to grow 23 to 24 percent, faster than sales. *3 of 6***
  - **Opus 5.5** Depreciation from record capital spending pressing gross margin
  - **Kimi K3.1 Pro** Margin pressure from infrastructure and operating leases
  - **Muse Spark Open** Gross margin compresses to 65% from under-utilized AI infrastructure and higher component costs

If all of this happens

**$88.9B** sales · −3% vs base

**$4.48** EPS · −7% vs base

For sales near the average low of $88.9B, 3.2% under the base, the company would have to miss its own range of $89.85B to $90.95B: cloud capacity arriving late, cloud usage softer, and personal computers falling harder than guided as component prices lift device prices. Earnings per share would fall to about $4.48, 7.4% under the base. Muse Spark Open goes furthest, to $87.5B and $4.20, because it alone starts from the guided midpoint rather than a beat, and it lets high rates and a sharply contracting PC market cut enterprise and consumer spending.

The June quarter's $90.0B came in above a range of $86.7B to $87.8B, and the models found beats in each recent quarter.

The six lows run from $87.5B (Muse Spark Open) to $89.8B (Opus 5.5).

First sign In the first-quarter report: Azure growth below about 45% in constant currency, or revenue near the bottom of the $89.85B to $90.95B range.

High about 1 chance in 10 of better

What the high case assumes

- ****Azure accelerates toward 50% as capacity arrives early and is sold at once.** June revenue beat on efficiency gains across the CPU and GPU fleet and earlier delivery of capacity. *6 of 6***
  - **Opus 5.5** Azure accelerating toward 50 percent
  - **GLM 5.3** letting Azure print above 47 percent growth each quarter
  - **DeepSeek V4.1 Flash** Azure above 47 percent constant currency through the year as new capacity is monetised on arrival
  - **Muse Spark Open** pushing Azure growth above 50 percent
  - **GPT-6.1 Sol** More cloud capacity monetized sooner
  - **Kimi K3.1 Pro** Faster data-center and processor deployment
- ****Usage-based Copilot billing lifts Microsoft 365 cloud ahead of plan.** Usage billing began in July alongside over 30 million paid Copilot seats. *4 of 6***
  - **Opus 5.5** Usage-based Copilot billing expanding Microsoft 365 revenue
  - **GPT-6.1 Sol** Large-scale Copilot and agent consumption
  - **Kimi K3.1 Pro** Rapid Copilot and premium-suite adoption
  - **DeepSeek V4.1 Flash** Copilot seat growth plus premium tier and usage billing accelerating Microsoft 365 commercial cloud
- ****Personal computing does better than its guided decline.** Partners restocking channel inventory lifted the segment in June. *3 of 6***
  - **Opus 5.5** Windows revenue better than the guided decline
  - **GLM 5.3** PC market holds up better than feared as inventories normalise
  - **DeepSeek V4.1 Flash** The personal computer decline bottoming sooner, limiting the drag from the consumer half

If all of this happens

**$94.2B** sales · +3% vs base

**$5.16** EPS · +7% vs base

For sales near the average high of $94.2B, 2.5% above the base, capacity would have to arrive even earlier and be sold at once, pushing Azure toward 50%; usage-based Copilot billing, begun in July, would have to add revenue quickly; and personal computing would have to beat its guided decline. Earnings per share would reach about $5.16, 6.6% above the base. Sol and DeepSeek go furthest on sales, to $95.0B. Sol reaches the highest earnings, $5.28, because its middle already sits about $2.1B above the guidance midpoint and it adds a smaller currency headwind and more cloud capacity monetized sooner.

The company credits June's beat to efficiency gains across its processor fleet and earlier delivery of new capacity.

The six highs run from $93.0B (GLM 5.3) to $95.0B (GPT-6.1 Sol).

First sign Azure growth above 46 percent in the first-quarter report, with a second-quarter outlook that rises again.

#### Watch

The first-quarter report: total revenue against $89.85B to $90.95B, Azure against about 45%, and cost of revenue against $29.6B to $29.8B. A beat smaller than recent quarters points lower; Azure above 46 percent and a stronger December outlook point higher.

##### Sales

PricedIn **$95.3B** $91.0B – $99.5B

$91.0B $93.2B $95.3B $97.2B $99.5B

##### adjusted EPS

PricedIn **$4.88** $4.48 – $5.31

$4.48 $4.68 $4.88 $5.10 $5.31

**Middle** the forecast **50% range** 1 in 4 chance the result comes in below this bar, 1 in 4 above it. Half the time it lands inside. **80% range** 1 in 10 chance below this line, 1 in 10 above it. These are the low and high printed by each number.

Five models cluster from $92.5B to $95.7B; Sol stands out at $98.0B. Muse Spark Open, lowest, starts from the guided midpoint and builds on 'Azure near 40 percent' where others see the mid 40s. Sol, highest, plans '$92.5 billion in September, $98 billion in December', assuming usage billing and premium subscriptions 'gradually accelerate Microsoft 365 revenue'. Sol also has the widest range, from $92.0B to $104B.

The models expect second-quarter fiscal 2027 sales of about $95.3B, up 17% on a year earlier, with cloud growth still leading.

Sales **$95B** +17% on Q2 FY26 ($81B)

*50% range* $93B – $97B *80% range* $91B – $99B

Adjusted EPS **$4.88** +18% on Q2 FY26 ($4.14)

*50% range* $4.68 – $5.10 *80% range* $4.48 – $5.31

#### 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 October to December quarter has no company outlook yet; Microsoft gives one with its first-quarter report. A year earlier the quarter brought sales of $81.3B and earnings per share of $4.14.

The company has said it expects growth in the first half of the fiscal year to accelerate, while warning that Azure growth can vary from quarter to quarter with capacity, timing and contract mix.

Five models put the middle between $92.5B and $95.7B, with Sol apart at $98.0B; the average is $95.3B. Most see sales rising 2.3% to 3.8% from the first quarter; Sol sees 5.9%.

Earnings per share barely move from the first quarter in most forecasts, as depreciation and operating costs absorb the added sales. Sol alone sees a clear step up, of 4.1%, to $5.09.

The spread is wider than for the first quarter, because the models must guess both the first-quarter result and the outlook that follows it.

Earnings per share average $4.88, up 18% on a year earlier but close to the first quarter's $4.84, as rising depreciation and operating costs absorb most of the added sales.

#### The quarter forecast, line by line

| Line | Q4 FY26 reported | Q1 FY26 reported | Q1 FY27 forecast | Q2 FY26 reported | Q2 FY27 forecast |
| --- | --- | --- | --- | --- | --- |
| Revenue | $90.0B | $77.7B | $91.9B | $81.3B | $95.3B |
| Sales growth, year on year | +18% | +18% | +18% | +17% | +17% |
| Intelligent Cloud | $39.3B | $30.9B | $41.9B | $32.9B | $43.9B |
| More Personal Computing | $12.9B | $13.8B | $12.6B | $14.2B | $12.8B |
| Productivity and Business Processes | $37.8B | $33.0B | $37.4B | $34.1B | $38.5B |
| Cost of revenue | $29.5B | $24.0B | $29.9B | $26.0B | $31.7B |
| Gross margin | $60.5B | $53.6B | $62.0B | $55.3B | $63.6B |
| Gross margin | 67.2% | 69.0% | 67.4% | 68.0% | 66.8% |
| Research and development | $10.00B | $8.15B | $8.92B | $8.50B | $9.40B |
| Sales and marketing | $7.59B | $5.72B | $6.07B | $6.58B | $6.70B |
| General and administrative | $2.29B | $1.81B | $1.93B | $1.93B | $2.02B |
| Operating income | $40.6B | $38.0B | $45.0B | $38.3B | $45.5B |
| Operating margin | 45.1% | 48.9% | 49.0% | 47.1% | 47.8% |
| Other income (expense), net | $3.44B | −$3.66B | −$0.18B | $9.97B | −$0.23B |
| Income before income taxes | $44.0B | $34.3B | $44.9B | $48.2B | $45.3B |
| Provision for income taxes | $8.28B | $6.55B | $8.97B | $9.79B | $9.05B |
| Tax rate | 18.8% | 19.1% | 20.0% | 20.3% | 20.0% |
| Net income | $35.8B | $27.7B | $35.9B | $38.5B | $36.2B |
| Diluted | 7.44B | 7.47B | 7.44B | 7.46B | 7.43B |
| Adjusted EPS | $4.74 | $4.13 | $4.84 | $4.14 | $4.88 |

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

**First-half outlook**

Growth expected to accelerate in the first half “Even with the strong close to Q4, we continue to expect H1 growth to accelerate.” July 2026 earnings call

**Year-earlier quarter revenue**

$81.3B “This quarter, revenue was $81.3 billion, up 17% in constant currency.” January 2026 earnings call

**Windows and devices outlook**

Revenue to decline in the high teens for the fiscal year “As a result, we expect revenue to decline in the high teens for the fiscal year.” July 2026 earnings call

**Microsoft 365 cloud outlook**

Growth to accelerate through the year with usage billing added in July “With the premium SKU momentum and the increased monetization opportunity from adding usage-based billing products alongside per-seat licensing in July, we expect to see acceleration in M365 commercial cloud revenue growth through this fiscal year.” July 2026 earnings call

**Contracted backlog**

Up 84% to $678B “Microsoft Cloud revenue was $59.3 billion and increased 27%, and commercial remaining performance obligation increased 84% to $678 billion.” July 2026 earnings release

#### Low and high

Base the middle forecast

Key assumptions

- ****Microsoft 365 cloud growth accelerates on Copilot seats, premium suites and usage billing.** Management expects acceleration in that business through the fiscal year after adding usage billing in July. *5 of 6***
  - **Opus 5.5** Copilot seats and usage-based billing lifting Microsoft 365 cloud
  - **GPT-6.1 Sol** Premium subscriptions and usage-based agent spending
  - **GLM 5.3** Microsoft 365 commercial cloud accelerates through the year on Copilot seat momentum
  - **DeepSeek V4.1 Flash** Copilot seats, the premium tier and usage billing keep Microsoft 365 commercial cloud growing around 16 to 17 percent
  - **Kimi K3.1 Pro** Copilot seats, premium suites and consumption revenue
- ****Personal computing keeps shrinking, with Windows and devices down in the high teens.** The company guided that decline for the fiscal year on component prices and inventory. *5 of 6***
  - **Opus 5.5** Windows OEM and devices revenue down in the high teens
  - **GLM 5.3** Windows OEM and devices declines in the high teens
  - **DeepSeek V4.1 Flash** Windows OEM and devices down high teens and gaming shrinking, dragging the consumer half
  - **GPT-6.1 Sol** Consumer weakness offsets part of commercial expansion
  - **Kimi K3.1 Pro** Moderating personal-computing and gaming revenue
- ****Azure keeps growing in the mid 40s as capacity keeps arriving.** The company expects first-half growth to accelerate and demand to keep exceeding supply. *5 of 6***
  - **Opus 5.5** supports a mid 40s Azure path in the first two quarters
  - **DeepSeek V4.1 Flash** Azure holds 44 to 45 percent constant-currency growth into the autumn and a little above that in December
  - **GLM 5.3** stays around 40-45% through the year as capacity is delivered
  - **Kimi K3.1 Pro** Azure capacity additions and roughly mid-forties near-term growth
  - **GPT-6.1 Sol** easing to about 46% for the full year
- ****Operating expenses grow slower than sales, offsetting part of rising infrastructure costs.** Headcount fell to about 223,000 from 228,000 a year earlier. *4 of 6***
  - **Opus 5.5** Operating expenses held to mid to high single digit growth
  - **GPT-6.1 Sol** Expense growth below revenue growth
  - **DeepSeek V4.1 Flash** Operating expenses up only about eight and a half percent and a twenty percent tax rate
  - **Muse Spark Open** Operating leverage from 7-8 percent opex growth offsets COGS up low-20s

If all of this happens

**$95B** sales

**$4.88** adjusted EPS

Low about 1 chance in 10 of worse

What the low case assumes

- ****Azure slows toward the low 30s as capacity arrives late or demand softens.** The company warns that demand for cloud and AI is evolving and difficult to forecast. *4 of 6***
  - **Opus 5.5** Azure growth limited by late capacity or softer AI demand
  - **Muse Spark Open** Azure growth decelerates to 20-25% as customers optimize spend
  - **GLM 5.3** Azure decelerates into the low-30s
  - **DeepSeek V4.1 Flash** Azure growth falling to the low thirties or below as enterprise AI budgets tighten
- ****Windows and devices fall more than 20 percent as device prices rise.** Component prices are raising personal computer prices, which cuts demand. *4 of 6***
  - **GLM 5.3** cutting Windows OEM and devices more than 20 percent
  - **DeepSeek V4.1 Flash** Windows OEM down 25 percent or more
  - **Kimi K3.1 Pro** Personal-computing declines and higher component costs
  - **Opus 5.5** Windows and Xbox revenue falling faster than guided
- ****Copilot seat additions flatten and usage billing adds little.** Usage billing began only in July, so its size is untested. *3 of 6***
  - **GLM 5.3** Copilot seat adds flatten and usage-based billing revenue disappoints
  - **Kimi K3.1 Pro** Weaker Copilot seat expansion and usage billing
  - **GPT-6.1 Sol** Slower paid-agent adoption or customer funding stress

If all of this happens

**$91B** sales · −5% vs base

**$4.48** EPS · −8% vs base

For sales near the average low of $91.0B, 4.5% under the base, the first quarter would have to disappoint and the December outlook come in soft: cloud capacity late, usage billing slow to add revenue, and personal computers weaker than guided. Earnings per share would fall to about $4.48, 8.4% under the base. Muse Spark Open goes furthest, to $88.5B and $4.30, because it alone lets Azure slow sharply, with growth that 'decelerates to 20-25%' as customers trim spending or move to competitors, where the others stop at the low 30s.

The year-earlier quarter's $81.3B already grew 17%, and the company has not yet given an outlook for this quarter.

The six lows run from $88B (Muse Spark Open) to $92B (GLM 5.3).

First sign A first-quarter report with Azure below about 45%, or a December outlook implying little growth from the first quarter.

High about 1 chance in 10 of better

What the high case assumes

- ****Azure accelerates toward 50% as capacity is monetized on arrival.** Efficiency gains and earlier capacity delivery drove June's beat. *5 of 6***
  - **Opus 5.5** Azure accelerating toward 50 percent
  - **Muse Spark Open** pushing Azure growth above 50 percent
  - **GLM 5.3** letting Azure print above 47 percent growth each quarter
  - **DeepSeek V4.1 Flash** Azure above 47 percent constant currency through the year as new capacity is monetised on arrival
  - **GPT-6.1 Sol** More cloud capacity monetized sooner
- ****Copilot seats and usage billing grow faster than planned.** Net Copilot seat additions more than doubled quarter over quarter in June. *4 of 6***
  - **GPT-6.1 Sol** Large-scale Copilot and agent consumption
  - **Kimi K3.1 Pro** Rapid Copilot and premium-suite adoption
  - **GLM 5.3** Copilot seat adds keep doubling sequentially toward 60 million-plus paid seats
  - **Muse Spark Open** Azure and Copilot usage-based billing exceed expectations
- ****Another investment gain lifts earnings per share.** June's adjusted earnings included a $3.2B gain from Anthropic. *3 of 6***
  - **GLM 5.3** Another large portfolio investment gain and a tax rate below 20 percent
  - **Opus 5.5** Gains like the Anthropic gain last quarter sit only in the high end.
  - **GPT-6.1 Sol** The upper earnings ranges allow investment gains that would remain in Microsoft's adjusted earnings

If all of this happens

**$99B** sales · +4% vs base

**$5.31** EPS · +9% vs base

For sales near the average high of $99.5B, 4.4% above the base, Azure would have to accelerate toward 50%, Copilot usage billing would have to scale quickly, and an investment gain like June's could lift earnings. Earnings per share would reach about $5.31, 8.7% above the base. Sol goes furthest, to $104B and $5.75, because it already expects the strongest December of any model, a step of 5.9% from the first quarter in its middle, and lets more cloud capacity be monetized sooner on top of that. Kimi follows at $101B.

June's adjusted earnings included a $3.2B gain from Anthropic, a kind of gain the company's adjusted figures do not exclude.

The six highs run from $97B (GLM 5.3) to $104B (GPT-6.1 Sol).

First sign A December-quarter outlook that rises again, with Azure above 46 percent in the first-quarter report.

#### Watch

The first-quarter report brings the company's December outlook. Watch its revenue range against $95.3B, the Azure growth it expects, and Copilot seat additions. A range well below that number points lower; a higher Azure outlook points higher.

##### Sales

PricedIn **$389B** $369B – $412B

$369B $380B $389B $400B $412B

##### adjusted EPS

PricedIn **$19.44** $17.77 – $21.37

$17.77 $18.62 $19.44 $20.32 $21.37

**Middle** the forecast **50% range** 1 in 4 chance the result comes in below this bar, 1 in 4 above it. Half the time it lands inside. **80% range** 1 in 10 chance below this line, 1 in 10 above it. These are the low and high printed by each number.

The models agree on the shape: cloud up strongly, personal computing down, margins a little lower. DeepSeek, lowest at $384B, has cost of revenue growing faster than sales and the consumer half pulling company growth to '13 to 15 percent in the second'. Sol, highest at $400B, expects 'a September sales beat and fiscal 2027 sales growth near 21%' because capacity becomes usable sooner and paid agent usage grows.

The models expect fiscal 2027 sales of about $389B, up 17%, as cloud and Copilot outgrow a shrinking personal computing business.

Sales **$389B** +17% on FY2026

*50% range* $380B – $400B *80% range* $369B – $412B

Adjusted EPS **$19.44**

*50% range* $18.62 – $20.32 *80% range* $17.77 – $21.37

Base case

Gross margin **66.4%** 67.9% in FY2026

Operating margin **46.5%** 46.8% in FY2026

#### Investment and cash generation

Base case

Capital spending **$179B**

Free cash flow **$38.3B**

Free cash flow: operating cash flow less capital spending.

Fiscal 2026 brought sales of $332B, up 18%, with Azure passing $100B for the first time. Capital spending rose to $116B from $64.6B, and free cash flow slipped to $67.0B.

For fiscal 2027 the company expects another year of double-digit revenue and operating income growth, operating margins down less than a point, a tax rate of about 20%, and free cash flow that stays positive.

The models' middles run from $384B to $400B, averaging $389B. Gross margin slides to about 66% as data center costs flow through, and operating margin to about 46%.

Building dominates the cash picture. The models put capital spending near $179B and free cash flow near $38.3B, though Kimi's $75.0B sits far above the rest because it assumes much lower spending, $155B.

Sol is highest at $400B, expecting capacity to be sold faster; DeepSeek is lowest at $384B, expecting costs to outrun sales and the consumer half to slow growth in the second half.

Earnings per share average $19.44, growing slower than sales: costs of the build rise faster than revenue, the tax rate is about 20%, and fiscal 2026's large investment gains are not expected to repeat.

#### Sales by quarter

Q3 FY25 $70.1B +13% · margin 68.7%

Q4 FY25 $76.4B +18% · margin 68.6%

Q1 FY26 $77.7B +18% · margin 69.0%

Q2 FY26 $81.3B +17% · margin 68.0%

Q3 FY26 $82.9B +18% · margin 67.6%

Q4 FY26 $90.0B +18% · margin 67.2%

Q1 FY27 $91.9B +18%

Each percent is on the same quarter a year earlier. The forecast rows are the models' average.

#### What it rests on

**Year outlook**

Another year of double-digit revenue and operating income growth “we continue to expect another fiscal year of double-digit revenue and operating income growth” July 2026 earnings call

**Operating margin outlook**

Down less than a point for the year “Even as we invest to meet growing demand, full fiscal year operating margins should be down less than a point.” July 2026 earnings call

**Tax rate outlook**

About 20% “And finally, we expect our FY27 effective tax rate to be approximately 20%.” July 2026 earnings call

**Calendar 2026 capital spending**

About $175B after moving leases to operating leases “However, the shift from finance to operating leases adjusts our expectation to approximately $175 billion.” July 2026 earnings call

**Copilot seats**

Over 30 million paid seats, net additions more than doubling “When it comes to knowledge work, we now have over 30 million paid Microsoft 365 co-pilot seats, with net seat ads more than doubling quarter over quarter.” July 2026 earnings call

**Contracted backlog**

Up 84% to $678B “Commercial remaining performance obligation grew 84% to $678 billion.” July 2026 earnings call

#### Low and high

Base the middle forecast

Key assumptions

- ****Azure grows in the low-to-mid 40s through the year as capacity comes online.** The company is roughly doubling capacity in two years and says demand exceeds supply. *6 of 6***
  - **GLM 5.3** Azure keeps growing in the low-to-mid 40s percent as Microsoft brings capacity online faster
  - **Muse Spark Open** Azure grows 40-45 percent in FY27 as capacity additions are quickly monetized and demand exceeds supply
  - **Opus 5.5** Azure growth in the mid 40s as capacity comes online
  - **Kimi K3.1 Pro** Azure capacity additions and roughly mid-forties near-term growth
  - **GPT-6.1 Sol** easing to about 46% for the full year
  - **DeepSeek V4.1 Flash** Azure and AI capacity arriving faster than planned
- ****Copilot seats, premium suites and usage billing accelerate Microsoft 365 cloud.** Over 30 million paid seats, and management expects acceleration through the year. *6 of 6***
  - **Opus 5.5** Copilot seats and usage-based billing lifting Microsoft 365 cloud
  - **GPT-6.1 Sol** Premium subscriptions and usage-based agent spending
  - **GLM 5.3** Microsoft 365 commercial cloud accelerates through the year on Copilot seat momentum
  - **Kimi K3.1 Pro** Copilot seats, premium suites and consumption revenue
  - **DeepSeek V4.1 Flash** Copilot seats, the premium tier and usage billing keep Microsoft 365 commercial cloud growing around 16 to 17 percent
  - **Muse Spark Open** Microsoft 365 Copilot and E5 E7 drive ARPU growth
- ****Operating expenses grow 7 to 8 percent, so operating margin slips less than a point.** The company guided margins down less than a point and headcount is falling. *5 of 6***
  - **GLM 5.3** offset by operating expense growth of 7-8 percent, so operating margin slips less than a point
  - **Muse Spark Open** Operating leverage from 7-8 percent opex growth offsets COGS up low-20s, keeping operating margin down less than a point
  - **Opus 5.5** Operating expenses held to mid to high single digit growth
  - **GPT-6.1 Sol** Expense growth below revenue growth
  - **DeepSeek V4.1 Flash** Operating expenses up only about eight and a half percent and a twenty percent tax rate
- ****Personal computing shrinks, with Windows and devices down in the high teens.** The company guided that decline on component prices, a Windows 10 comparison and inventory. *5 of 6***
  - **Opus 5.5** Windows OEM and devices revenue down in the high teens
  - **GLM 5.3** Windows OEM and devices declines in the high teens
  - **DeepSeek V4.1 Flash** Windows OEM and devices down high teens and gaming shrinking, dragging the consumer half
  - **Kimi K3.1 Pro** Moderating personal-computing and gaming revenue
  - **GPT-6.1 Sol** Consumer weakness offsets part of commercial expansion
- ****Capital spending near $179B leaves free cash flow thin but positive.** Management promises only that free cash flow stays positive in fiscal 2027. *4 of 6***
  - **Opus 5.5** free cash flow for the year is small, about 25 billion dollars
  - **DeepSeek V4.1 Flash** free cash flow falls to roughly 23 billion
  - **GLM 5.3** free cash flow around $30 billion, thin but positive as management promised
  - **GPT-6.1 Sol** New operating-lease payments restrain cash generation.

If all of this happens

**$389B** sales

**$19.44** adjusted EPS

**$179B** Capital spending

**$38.3B** Free cash flow

Cash walk not supplied by every model.

Low about 1 chance in 10 of worse

What the low case assumes

- ****Gross margin falls about three points more as component and depreciation costs rise.** Higher component pricing added about $25B to calendar 2026 capital spending. *6 of 6***
  - **GLM 5.3** Memory and component price inflation squeezes gross margin down 300-plus basis points
  - **DeepSeek V4.1 Flash** Cost of revenue climbing three points or more as a share of sales
  - **Muse Spark Open** Gross margin compresses to 65%
  - **Opus 5.5** Depreciation from record capital spending pressing gross margin
  - **Kimi K3.1 Pro** Margin pressure from infrastructure and operating leases
  - **GPT-6.1 Sol** Higher infrastructure costs and stronger dollar
- ****Azure slows to the low 30s or below as capacity lands into softer demand.** The company warns that overestimating demand may leave infrastructure underused. *5 of 6***
  - **GLM 5.3** Azure decelerates into the low-30s as the capacity surge lands into softer demand
  - **DeepSeek V4.1 Flash** Azure growth falling to the low thirties or below as enterprise AI budgets tighten
  - **Opus 5.5** Azure growth limited by late capacity or softer AI demand
  - **Muse Spark Open** Azure growth decelerates to 20-25%
  - **Kimi K3.1 Pro** Slower Azure consumption and delayed capacity monetization
- ****Windows and devices fall more than 20 percent, beyond the guided high-teens decline.** Rising component costs keep pushing personal computer prices up. *5 of 6***
  - **GLM 5.3** cutting Windows OEM and devices more than 20 percent
  - **DeepSeek V4.1 Flash** Windows OEM down 25 percent or more
  - **Opus 5.5** Windows and Xbox revenue falling faster than guided
  - **Kimi K3.1 Pro** Personal-computing declines and higher component costs
  - **GPT-6.1 Sol** Weak personal-computer and gaming spending
- ****Copilot seat additions flatten and usage billing disappoints.** Usage billing is new and its revenue is unproven. *3 of 6***
  - **GLM 5.3** Enterprise AI adoption stalls: Copilot seat adds flatten
  - **Kimi K3.1 Pro** Weaker Copilot seat expansion and usage billing
  - **GPT-6.1 Sol** Slower paid-agent adoption or customer funding stress

If all of this happens

**$369B** sales · −5% vs base

**$17.77** EPS · −9% vs base

For sales near the average low of $369B, 5.2% under the base and still 11% above fiscal 2026, Azure would have to slow toward the low 30s as new capacity meets softer demand, Copilot adoption would have to stall, and personal computers would have to fall more than 20 percent while component costs squeeze gross margin. Earnings per share would drop to about $17.77, 8.6% under the base. Muse Spark Open goes furthest on sales, to $362B, because it alone has Azure growth that 'decelerates to 20-25%' as customers optimize spend; Kimi has the lowest earnings, $17.40, as infrastructure and operating-lease costs compress margins.

Fiscal 2026 grew 18% to $332B, and roughly 30% of the $678B backlog is due to be recognized within twelve months.

The six lows run from $362B (Muse Spark Open) to $375B (GLM 5.3).

First sign An Azure growth figure below about 40 percent in a quarterly report, or Microsoft Cloud gross margin below 64 percent.

High about 1 chance in 10 of better

What the high case assumes

- ****Azure grows close to 50% for the year as capacity is sold on arrival.** The company reports demand still ahead of supply after a quarter of early capacity delivery. *6 of 6***
  - **GLM 5.3** Azure grows close to 50 percent for the year
  - **Opus 5.5** Azure accelerating toward 50 percent
  - **Muse Spark Open** pushing Azure growth above 50 percent
  - **DeepSeek V4.1 Flash** Azure above 47 percent constant currency through the year
  - **Kimi K3.1 Pro** Faster data-center and processor deployment
  - **GPT-6.1 Sol** More cloud capacity monetized sooner
- ****Copilot deployments and usage billing scale across large customers.** Net seat additions more than doubled in June, and usage billing began in July. *5 of 6***
  - **Opus 5.5** Usage-based Copilot billing expanding Microsoft 365 revenue
  - **GPT-6.1 Sol** Large-scale Copilot and agent consumption
  - **Kimi K3.1 Pro** Rapid Copilot and premium-suite adoption
  - **Muse Spark Open** Enterprise-wide Copilot deployments over 50k seats multiply
  - **DeepSeek V4.1 Flash** Copilot seat growth plus premium tier and usage billing accelerating Microsoft 365 commercial cloud
- ****Efficiency and pricing stop gross margin from falling.** The company credits efficiency gains in Azure and Microsoft 365 with offsetting AI costs. *4 of 6***
  - **DeepSeek V4.1 Flash** Gross margin stabilising as efficiency gains and price increases offset component costs
  - **Kimi K3.1 Pro** Greater infrastructure efficiency and operating leverage
  - **GPT-6.1 Sol** Lower cost per useful task and stronger pricing
  - **Muse Spark Open** Efficiency gains from Maya 200 and software optimization improve gross margin above 68 percent despite mix shift
- ****Personal computing stabilizes earlier than guided.** Channel inventory could normalize and component price pressure ease. *4 of 6***
  - **Opus 5.5** Windows revenue better than the guided decline
  - **GLM 5.3** PC market holds up better than feared as inventories normalise
  - **DeepSeek V4.1 Flash** The personal computer decline bottoming sooner
  - **GPT-6.1 Sol** Earlier advertising and consumer recovery

If all of this happens

**$412B** sales · +6% vs base

**$21.37** EPS · +10% vs base

For sales near the average high of $412B, 5.7% above the base, capacity would have to be monetized on arrival so Azure grows close to 50%, Copilot consumption would have to spread across the installed base, efficiency would have to protect margins, and personal computers would have to stabilize sooner. Earnings per share would reach about $21.37, 9.9% above the base. Sol goes furthest, to $432B and $22.90, because its middle already assumes 21% growth and its high adds lower cost per useful task and stronger pricing; Kimi follows at $420B.

In June the company said extra capacity was quickly monetized, and its Azure business passed $100B for the year.

The six highs run from $400B (GLM 5.3) to $432B (GPT-6.1 Sol).

First sign An Azure growth figure above about 47 percent, with guidance for further acceleration.

#### Watch

Each quarterly report: Azure growth against about 45%, Copilot seat additions, Microsoft Cloud gross margin, and capital spending. Azure below 40 percent or cloud margin below 64 percent points lower; Azure above 47 percent with margin holding points higher.

##### Sales

PricedIn **$452B** $409B – $498B

$409B $429B $452B $474B $498B

##### adjusted EPS

PricedIn **$22.58** $19.38 – $26.12

$19.38 $20.98 $22.58 $24.20 $26.12

**Middle** the forecast **50% range** 1 in 4 chance the result comes in below this bar, 1 in 4 above it. Half the time it lands inside. **80% range** 1 in 10 chance below this line, 1 in 10 above it. These are the low and high printed by each number.

All see growth of 13% to 20%. DeepSeek, lowest at $434B, ties its caution to 'One customer holding close to half the contracted backlog' and lets gross margin fall to 64.6 percent. Sol, highest at $480B, assumes 'roughly 20% growth in the newly defined Microsoft 365 cloud business in fiscal 2027 and 2028'. Sol's range is also widest, $417B to $554B: slower contract conversion and funding stress at one end, capacity monetized sooner at the other.

The models expect fiscal 2028 sales of about $452B, up 16%, as the contracted cloud backlog keeps turning into revenue.

Sales **$452B** +16% on FY2027

*50% range* $429B – $474B *80% range* $409B – $498B

Adjusted EPS **$22.58** +16% on FY2027

*50% range* $20.98 – $24.20 *80% range* $19.38 – $26.12

Base case

Gross margin **65.8%** 66.4% in FY2027

Operating margin **46.3%** 46.5% in FY2027

#### Investment and cash generation

Base case

Capital spending **$194B**

Free cash flow **$59.5B**

Free cash flow: operating cash flow less capital spending.

Beyond fiscal 2027 the company gives no outlook. The models lean on its contracted backlog of $678B, up 84%, of which roughly 45% came from OpenAI as of December 2025, and on its plan to roughly double data center capacity in two years.

Their middles run from $434B to $480B, averaging $452B. Capital spending keeps rising to about $194B, while free cash flow recovers to about $59.5B.

Most expect growth to slow as Azure's base surpasses $200B and supply catches up with demand.

Operating margin holds near 46% in most forecasts, but the models split on gross margin: Sol and DeepSeek let depreciation take it to 64% and 65%, while Kimi and Muse Spark Open keep it near 67%.

The ranges widen: the average low of $409B is 9.5% under the base and the average high of $498B is 10% above, with Sol's high reaching $554B.

Earnings per share average $22.58, up 16%, tracking sales: depreciation from heavier spending holds margins roughly flat, and buybacks trim the share count slightly.

#### Sales by year

FY2024 $245B

FY2025 $282B +15%

FY2026 $332B +18%

FY2027 $389B +17%

FY2028 $452B +16%

FY2029 $514B +14%

The forecast years are the models' average.

#### Commercial contracted backlog in fiscal 2026

Sep 2025 $400B

Dec 2025 $625B

Jun 2026 $678B

**Where each number comes from**

**Sep 2025**

“commercial RPO increased over 50% to nearly $400 billion with a weighted average duration of only 2 years”

**Dec 2025**

“Commercial remaining performance obligation, continues to be reported net of reserves, increased to $625 billion.”

**Jun 2026**

“Commercial remaining performance obligation grew 84% to $678 billion.”

Contracted cloud commitments not yet recognized as revenue; the models lean on them for later-year growth.

#### What it rests on

**Contracted backlog**

Up 84% to $678B “Microsoft Cloud revenue was $59.3 billion and increased 27%, and commercial remaining performance obligation increased 84% to $678 billion.” July 2026 earnings release

**OpenAI share of backlog**

About 45% “Approximately 45% of our commercial RPO balance is from OpenAI.” January 2026 earnings call

**Backlog timing**

Roughly 30% recognized within twelve months; duration 2.3 years “RPO, including OpenAI, has a weighted average duration of 2.3 years, and roughly 30% will be recognized in revenue in the next 12 months, up 37% year over year.” July 2026 earnings call

**OpenAI Azure contract**

Incremental $250B of Azure services “OpenAI has contracted an incremental $250 billion of Azure services, our rev share, exclusive IP rights and API exclusivity for Azure continue until AGI or through 2030.” October 2025 earnings call

**Demand risk**

Demand for cloud and AI is difficult to forecast “Demand for cloud-based and AI products and services is evolving and difficult to forecast.” 2026 annual report

#### Low and high

Base the middle forecast

Key assumptions

- ****Azure growth slows from the 40s toward the mid-30s as its base grows.** Backlog underwrites demand, but a much larger base and catching-up supply slow the rate. *4 of 6***
  - **GPT-6.1 Sol** It slows toward the mid-thirties and about 30% in the following years.
  - **GLM 5.3** For fiscal 2028 and 2029 I expect growth to moderate to the mid-teens
  - **Kimi K3.1 Pro** growth eventually normalizes as the revenue base enlarges and supply catches up
  - **DeepSeek V4.1 Flash** sixteen percent revenue growth this year and thirteen percent next year
- ****Copilot and usage billing keep Microsoft 365 cloud growing in the mid-to-high teens.** Over 30 million paid seats, with revenue per user rising on premium suites and usage. *4 of 6***
  - **GPT-6.1 Sol** roughly 20% growth in the newly defined Microsoft 365 cloud business in fiscal 2027 and 2028
  - **Kimi K3.1 Pro** Seat adoption, premium suites and usage billing support mid-teens productivity growth
  - **DeepSeek V4.1 Flash** Microsoft 365 commercial cloud growth hold around 16 to 17 percent
  - **GLM 5.3** management explicitly guiding Microsoft 365 commercial cloud growth to accelerate through fiscal 2027
- ****Free cash flow recovers to about $59.5B as operating cash catches up with spending.** Capital spending grows more slowly after the fiscal 2027 surge. *3 of 6***
  - **GLM 5.3** while operating cash flow catches up
  - **Kimi K3.1 Pro** free cash flow recovers as capital-spending growth slows
  - **Muse Spark Open** explains FCF pressure to ~45B in FY27 then recovery as spend moderates
- ****Depreciation from heavy spending pulls gross margin lower, toward 64% to 66%.** Capital spending rises to about $194B, and depreciation follows it. *3 of 6***
  - **Opus 5.5** I lower gross margin by about one point a year in later years
  - **DeepSeek V4.1 Flash** taking the gross margin from 67.9 percent last year to 66.2 percent and then 64.6 and 64.0
  - **GPT-6.1 Sol** annual gross margin falls to about 65% in fiscal 2027 and about 64% thereafter
- ****Personal computing stabilizes and stops dragging growth.** The fiscal 2027 decline rests partly on a one-time Windows 10 comparison. *2 of 6***
  - **GLM 5.3** with the segment stabilising in fiscal 2028
  - **GPT-6.1 Sol** modest consumer stabilization

If all of this happens

**$452B** sales

**$22.58** adjusted EPS

**$194B** Capital spending

**$59.5B** Free cash flow

Cash walk not supplied by every model.

Low about 1 chance in 10 of worse

What the low case assumes

- ****Gross margin falls three points or more as infrastructure costs outrun sales.** The company says infrastructure costs are reducing operating margins. *5 of 6***
  - **DeepSeek V4.1 Flash** Cost of revenue climbing three points or more as a share of sales
  - **GLM 5.3** squeezes gross margin down 300-plus basis points
  - **Muse Spark Open** Gross margin compresses to 65%
  - **Kimi K3.1 Pro** Margin pressure from infrastructure and operating leases
  - **Opus 5.5** Depreciation from record capital spending pressing gross margin
- ****OpenAI and other large AI customers draw on their commitments more slowly.** About 45% of the contracted backlog came from one customer. *4 of 6***
  - **DeepSeek V4.1 Flash** Money coming out of the contracted backlog more slowly, or being renegotiated, as the largest customer's own growth slows
  - **GLM 5.3** with OpenAI consuming its commitments more slowly than contracted
  - **GPT-6.1 Sol** The low allows slower contract conversion
  - **Opus 5.5** demand from a few large AI customers slows
- ****New capacity lands into softer demand and sits under-used.** The company warns overestimating demand may leave infrastructure underused and lead to impairment. *4 of 6***
  - **GLM 5.3** the capacity surge lands into softer demand, the oversupply scenario investors fear
  - **DeepSeek V4.1 Flash** The capacity still arrives, but it is under-used
  - **Muse Spark Open** AI demand disappoints and capacity overbuild
  - **Kimi K3.1 Pro** Slower Azure consumption and delayed capacity monetization
- ****High interest rates squeeze corporate technology budgets.** The 10-year Treasury yield rose to 5% by October 2026. *3 of 6***
  - **DeepSeek V4.1 Flash** Long-term interest rates at five percent and oil above a hundred dollars squeeze corporate technology budgets
  - **Muse Spark Open** enterprise IT budgets cut as rates stay high
  - **GPT-6.1 Sol** The Federal Reserve ten-year Treasury yield at about 5% supports caution about financing new data centers

If all of this happens

**$409B** sales · −9% vs base

**$19.38** EPS · −14% vs base

For sales near the average low of $409B, 9.5% under the base and only 5.0% above fiscal 2027, the AI build would have to prove narrower than the backlog implies: OpenAI drawing on its commitments more slowly, enterprise budgets tightening with long-term rates near 5%, and new capacity sitting under-used while costs rise. Earnings per share would fall to about $19.38, flat on the year. DeepSeek goes furthest on sales, to $392B, because it alone builds its low on the backlog's concentration: if that customer re-phases its commitments, the effect lands in fiscal 2028 and 2029. Sol has the lowest earnings, $18.60, letting margins fall with customer funding stress and expensive components.

The 10-year Treasury yield stood at 5% at the cutoff, and about 45% of the backlog came from OpenAI as of December 2025.

The six lows run from $392B (DeepSeek V4.1 Flash) to $424B (GLM 5.3).

First sign Commercial bookings excluding OpenAI turning negative, or Azure growth below 40 percent, in a fiscal 2027 quarterly report.

High about 1 chance in 10 of better

What the high case assumes

- ****Capacity is monetized on arrival and Azure stays above 40% growth for longer.** The company says new capacity was quickly monetized in June. *4 of 6***
  - **GPT-6.1 Sol** More cloud capacity monetized sooner
  - **Kimi K3.1 Pro** Azure remains above forty percent growth for longer
  - **DeepSeek V4.1 Flash** Azure above 47 percent constant currency through the year as new capacity is monetised on arrival
  - **GLM 5.3** Capacity delivered even faster than planned
- ****Copilot and usage-based products spread across the installed base.** Net Copilot seat additions more than doubled quarter over quarter. *4 of 6***
  - **GPT-6.1 Sol** Large-scale Copilot and agent consumption
  - **Kimi K3.1 Pro** Copilot plus usage-based products accelerate across the installed base
  - **Muse Spark Open** Enterprise-wide Copilot deployments over 50k seats multiply
  - **GLM 5.3** Copilot seat adds keep doubling sequentially toward 60 million-plus paid seats
- ****Demand broadens beyond OpenAI, so backlog growth comes from many customers.** All sequential backlog growth in June came from customers outside frontier model companies. *4 of 6***
  - **GPT-6.1 Sol** The high needs broad demand outside those customers as well.
  - **GLM 5.3** commercial remaining performance obligation growth excluding OpenAI re-accelerating above 30 percent
  - **DeepSeek V4.1 Flash** with the growth coming from outside the largest customer
  - **Kimi K3.1 Pro** Large backlog and strong growth excluding OpenAI support above-trend cloud revenue through fiscal 2028
- ****Efficiency gains and pricing hold gross margin up.** The company credits efficiency gains with offsetting AI infrastructure costs. *3 of 6***
  - **DeepSeek V4.1 Flash** Gross margin stabilising as efficiency gains and price increases offset component costs
  - **Kimi K3.1 Pro** Greater infrastructure efficiency and operating leverage
  - **GPT-6.1 Sol** Lower cost per useful task and stronger pricing

If all of this happens

**$498B** sales · +10% vs base

**$26.12** EPS · +16% vs base

For sales near the average high of $498B, 10% above the base, capacity would have to be monetized on arrival for a second year, Azure would have to stay above 40% growth for longer, Copilot consumption would have to spread across the installed base, and demand would have to broaden beyond OpenAI. Earnings per share would reach about $26.12, 16% above the base. Sol goes furthest, to $554B and $30.80, because its middle already grows 20% and its high adds large-scale Copilot and agent consumption with more capacity monetized sooner; Kimi follows at $525B.

All of the backlog's growth in the June quarter came from customers outside frontier model companies.

The six highs run from $472B (DeepSeek V4.1 Flash) to $554B (GPT-6.1 Sol).

First sign Backlog growth excluding OpenAI re-accelerating above 30 percent, with Azure above 47 percent.

#### Watch

Each quarter's backlog, split between OpenAI and other customers; Azure growth; Microsoft Cloud gross margin; and capital spending against free cash flow. Bookings outside OpenAI weakening or cloud margin below 63 percent points lower; broad backlog growth points higher.

##### Sales

PricedIn **$514B** $448B – $584B

$448B $481B $514B $548B $584B

##### adjusted EPS

PricedIn **$25.93** $21.04 – $31.23

$21.04 $23.53 $25.93 $28.42 $31.23

**Middle** the forecast **50% range** 1 in 4 chance the result comes in below this bar, 1 in 4 above it. Half the time it lands inside. **80% range** 1 in 10 chance below this line, 1 in 10 above it. These are the low and high printed by each number.

Five models sit between $484B and $530B; Sol is far above at $567B. DeepSeek, lowest, holds gross margin at 64.0 percent and says 'One customer holding close to half the contracted backlog is the main reason my FY2029 low is 430 billion dollars'. Sol, highest, has operating margins of '46.7% and 47.4%' from 'measured efficiency gains'. Sol's range is much wider: at $449B, delayed power and excess capacity; at $698B, faster paid agents and broad demand.

The models expect fiscal 2029 sales of about $514B, up 14%, though their spread is the widest of any year.

Sales **$514B** +14% on FY2028

*50% range* $481B – $548B *80% range* $448B – $584B

Adjusted EPS **$25.93** +15% on FY2028

*50% range* $23.53 – $28.42 *80% range* $21.04 – $31.23

Base case

Gross margin **65.6%** 65.8% in FY2028

Operating margin **46.5%** 46.3% in FY2028

#### Investment and cash generation

Base case

Capital spending **$200B**

Free cash flow **$89.7B**

Free cash flow: operating cash flow less capital spending.

Fiscal 2029 is three years out, and the company has given no outlook for it. The models extend their fiscal 2028 paths with growth slowing to between 12% and 18%.

Their middles run from $484B to $567B, averaging $514B. The average low of $448B sits 13% under the base and the average high of $584B 14% above.

Most expect capital spending to keep climbing, to about $200B on average, but Muse Spark Open has it falling to $135B, which lifts its free cash flow to $115B; Kimi expects $130B.

The backlog's concentration matters most here: DeepSeek says a slowdown at the largest customer would land in fiscal 2028 and 2029, not this year.

Sol stands apart, at $567B in the middle and up to $698B, expecting Microsoft 365 and Azure to keep compounding as operating margin improves to 47%.

Earnings per share average $25.93, up 15%. Operating margin holds near 47% on average, so earnings track sales; the spread from $23.80 to $29.10 mirrors the sales spread.

#### Each model's growth in FY2029

Opus 5.5 +13%

GPT-6.1 Sol +18%

GLM 5.3 +12%

Kimi K3.1 Pro +15%

Muse Spark Open +12%

DeepSeek V4.1 Flash +12%

Growth of each model's middle on its own forecast for the year before.

#### What it rests on

**OpenAI contract term**

Revenue share and Azure exclusivity run until AGI or through 2030 “OpenAI has contracted an incremental $250 billion of Azure services, our rev share, exclusive IP rights and API exclusivity for Azure continue until AGI or through 2030.” October 2025 earnings call

**Leases not yet started**

$329.1B, commencing fiscal 2027 to 2033 “As of June 30, 2026, we had additional leases, primarily for datacenters, that had not yet commenced of $329.1 billion, with some arrangements subject to certain contractual conditions being met. These leases will commence between fiscal year 2027 and fiscal year 2033 with lease terms of 1 year to 20 years.” 2026 annual report, leases note

**OpenAI share of backlog**

About 45% “Approximately 45% of our commercial RPO balance is from OpenAI.” January 2026 earnings call

**Margin risk**

Infrastructure costs are reducing operating margins “We are incurring significant costs to build and maintain infrastructure to support cloud-based and AI services, reducing operating margins.” 2025 annual report, risk factors

**Backlog timing**

Weighted average duration of 2.3 years “RPO, including OpenAI, has a weighted average duration of 2.3 years, and roughly 30% will be recognized in revenue in the next 12 months, up 37% year over year.” July 2026 earnings call

#### Low and high

Base the middle forecast

Key assumptions

- ****Contracted demand keeps converting to revenue without a wholesale loss of large contracts.** OpenAI contracted an incremental $250B of Azure services, with terms running through 2030. *4 of 6***
  - **GPT-6.1 Sol** The middle assumes no wholesale loss of those contracts.
  - **GLM 5.3** Multi-year contracted demand of this scale
  - **Kimi K3.1 Pro** Large backlog and strong growth excluding OpenAI support above-trend cloud revenue through fiscal 2028, with gradual normalization by fiscal 2029
  - **DeepSeek V4.1 Flash** Doubling the data-centre footprint in two years, with more than eighty percent more AI capacity in a single year
- ****Gross margin settles near 64% to 67% as depreciation and efficiency balance out.** Past build-outs saw depreciation peak and margins stop falling once spending slowed. *4 of 6***
  - **DeepSeek V4.1 Flash** why I let gross margin stop falling after FY2029
  - **GPT-6.1 Sol** about 64% thereafter
  - **Opus 5.5** I lower gross margin by about one point a year in later years
  - **GLM 5.3** toward about 66 percent in fiscal 2027 before stabilising
- ****Growth slows to the low-to-mid teens as AI revenue gets larger.** A much bigger base and supply catching up with demand slow the rate. *3 of 6***
  - **GLM 5.3** let revenue growth decelerate to around 12 percent by fiscal 2029
  - **Kimi K3.1 Pro** gradual deceleration in fiscal 2029
  - **GPT-6.1 Sol** assume slower Azure growth, continued premium and usage adoption, and modest consumer stabilization
- ****Capital spending keeps rising, to about $200B on average, holding free cash flow below earnings.** Data center leases of $329.1B not yet commenced run to fiscal 2033. *3 of 6***
  - **GPT-6.1 Sol** $235 billion in fiscal 2029
  - **GLM 5.3** capex keeps rising to roughly $200 and $215 billion
  - **DeepSeek V4.1 Flash** only recovering to about 72 billion by FY2029

If all of this happens

**$514B** sales

**$25.93** adjusted EPS

**$200B** Capital spending

**$89.7B** Free cash flow

Cash walk not supplied by every model.

Low about 1 chance in 10 of worse

What the low case assumes

- ****Capacity arrives in bulk and pricing and utilization fall.** The company warns that misaligned capacity investment may lead to impairment. *4 of 6***
  - **GLM 5.3** when capacity finally arrives in bulk, pricing and utilisation risk rise
  - **GPT-6.1 Sol** excess capacity later
  - **Muse Spark Open** AI demand disappoints and capacity overbuild
  - **DeepSeek V4.1 Flash** The capacity still arrives, but it is under-used
- ****Infrastructure costs keep squeezing margins.** The company says infrastructure costs are reducing operating margins. *4 of 6***
  - **DeepSeek V4.1 Flash** Cost of revenue climbing three points or more as a share of sales
  - **Muse Spark Open** Gross margin compresses to 65%
  - **Opus 5.5** Depreciation from record capital spending pressing gross margin
  - **Kimi K3.1 Pro** Margin pressure from infrastructure and operating leases
- ****The largest customer slows or re-phases its commitments.** About 45% of the contracted backlog came from OpenAI. *3 of 6***
  - **DeepSeek V4.1 Flash** If that customer slows or re-phases its commitments, the effect lands in FY2028 and FY2029, not this year.
  - **GLM 5.3** with OpenAI consuming its commitments more slowly than contracted
  - **GPT-6.1 Sol** The low allows delayed purchases or financial stress at model developers.
- ****Competitors limit prices and adoption.** The company reports intense competition across all its markets. *2 of 6***
  - **Muse Spark Open** Intense competition and risk that cloud/AI investments may not achieve expected returns
  - **GPT-6.1 Sol** the low allows competitors to limit prices and adoption

If all of this happens

**$448B** sales · −13% vs base

**$21.04** EPS · −19% vs base

For sales near the average low of $448B, 13% under the base and roughly flat on fiscal 2028, the AI build would have to turn into overcapacity: the largest customer re-phasing its commitments, pricing falling as supply catches up, competition limiting prices and adoption, and depreciation on a fleet that cannot be filled. Earnings per share would fall to about $21.04, 19% under the base. DeepSeek goes furthest on sales, to $430B, because the backlog's concentration in one customer is the main reason for its low; Sol has the lowest earnings, $19.20, combining excess capacity later with expensive components and slower adoption.

OpenAI's revenue share and exclusive rights run through 2030, and leases of $329.1B not yet commenced stretch to fiscal 2033.

The six lows run from $430B (DeepSeek V4.1 Flash) to $468B (GLM 5.3).

First sign Backlog growth outside OpenAI slowing, or Microsoft Cloud gross margin below 63 percent, in fiscal 2027 or 2028 reports.

High about 1 chance in 10 of better

What the high case assumes

- ****Azure stays above 40% growth for longer as capacity is sold on arrival.** Capacity is on track to roughly double, and demand still exceeds supply. *4 of 6***
  - **Kimi K3.1 Pro** Azure remains above forty percent growth for longer
  - **GPT-6.1 Sol** More cloud capacity monetized sooner
  - **DeepSeek V4.1 Flash** Azure accelerates past 47 percent constant-currency growth into the spring
  - **GLM 5.3** Capacity delivered even faster than planned
- ****Copilot and agent consumption compound across the installed base.** Over 30 million paid seats, with usage billing added on top of seats. *4 of 6***
  - **GPT-6.1 Sol** Large-scale Copilot and agent consumption
  - **Kimi K3.1 Pro** Copilot plus usage-based products accelerate across the installed base
  - **Muse Spark Open** Enterprise-wide Copilot deployments over 50k seats multiply
  - **GLM 5.3** Copilot seat adds keep doubling sequentially toward 60 million-plus paid seats
- ****Efficiency and pricing lift margins.** Efficiency gains in Azure and Microsoft 365 already offset part of AI costs. *3 of 6***
  - **GPT-6.1 Sol** Lower cost per useful task and stronger pricing
  - **Kimi K3.1 Pro** Greater infrastructure efficiency and operating leverage
  - **DeepSeek V4.1 Flash** Gross margin stabilising as efficiency gains and price increases offset component costs
- ****Demand broadens well beyond OpenAI.** Recent backlog growth came from customers outside frontier model companies. *3 of 6***
  - **GPT-6.1 Sol** The high needs broad demand outside those customers as well.
  - **GLM 5.3** commercial remaining performance obligation growth excluding OpenAI re-accelerating above 30 percent
  - **DeepSeek V4.1 Flash** with the growth coming from outside the largest customer

If all of this happens

**$584B** sales · +14% vs base

**$31.23** EPS · +20% vs base

For sales near the average high of $584B, 14% above the base, the company would need three more years of fast growth: Azure staying above 40% for longer on a much larger base, Copilot consumption across the installed base, demand broadening beyond OpenAI, and margins rising with efficiency. Earnings per share would reach about $31.23, 20% above the base. Sol goes furthest, to $698B and $41.20, because its middle already compounds at 20% and then 18% a year, and its high adds more capacity monetized sooner with lower cost per useful task and stronger pricing; Kimi follows at $630B.

Fiscal 2026 sales grew 18%, the fastest of the recent years, from a base far smaller than fiscal 2029's.

The six highs run from $535B (DeepSeek V4.1 Flash) to $698B (GPT-6.1 Sol).

First sign Azure growth holding above 40% through fiscal 2028 while Microsoft Cloud gross margin holds or improves.

#### Watch

Over fiscal 2027 and 2028: how fast Azure slows as its base grows, whether backlog outside OpenAI keeps growing, the cloud gross margin, and whether capital spending peaks. Slowing bookings and a falling margin point lower; Azure above 40% and steady margins point higher.

The years at a glance · sales

FY2024 **$245B**

FY2025 **$282B**

FY2026 **$332B**

FY2027 **$389B** $369B – $412B

FY2028 **$452B** $409B – $498B

FY2029 **$514B** $448B – $584B

$300B $400B $500B $600B

50% range 80% range

The years at a glance · adjusted EPS

FY2024 ○ **$11.80**

FY2025 **$14.13**

FY2026 **$17.28**

FY2027 **$19.44** $17.77 – $21.37

FY2028 **$22.58** $19.38 – $26.12

FY2029 **$25.93** $21.04 – $31.23

$15.00 $20.00 $25.00 $30.00

50% range 80% range

○ FY2024: as the company first printed it, on its earlier definition; it has not reprinted the year on the current one.