Leadership Under Uncertainty

The numbers are stated judgments. The record began on 13 October 2026 with starting confidences that the system set from the previous three weeks of reports, and a published rule has made every change since. They are not measured probabilities. The question and the rule are fixed before the record begins, and each change shows its evidence and its reasons. The rule and what is not claimed are on the method page.

The alternative

The cost of the AI capacity Microsoft builds rises faster than the revenue that capacity brings, so that by the end of 2027 Microsoft's operating profit stops growing or Microsoft cuts back its capital spending.

67.5% stated confidence, Likely (55–80%). Starting point on the record’s first day: Likely (55–80%), 67.5 per cent. Attention level: For information. Trend: Steady. Activation: Dormant.

Stated confidence, one point per daily state, from 13 October 2026. Hollow points were recorded late.
Key judgment

The judgment as stated before the record began

This is the leading alternative to the thesis. Microsoft is spending on new capacity at a record rate, and roughly two thirds of that spending goes on chips and servers that are written off over several years, so the cost arrives quickly while the revenue builds over time. If AI does not pay for itself at this scale, the first signs would appear in costs: a falling gross margin on the Microsoft Cloud, free cash flow shrinking as spending climbs, and operating profit growing more slowly than revenue. Commitments made years ahead, including data centre leases not yet started, would make those costs slow to cut. Microsoft would then face a choice between slower profit growth and a slower buildout.

Starting point

How the record started this assessment

On 13 October 2026, before the record began, the system read the 10 developments below and chose Likely (55–80%). The record started the assessment at the middle of that range, 67.5 per cent. The reading is kept in the first day's record.

Its reason, machine-written by anthropic/claude-sonnet-4.6: Multiple developments (1, 3, 6, 9, 10) report TSMC raising foundry prices 3 to 15 percent from 2027, directly raising the cost of AI capacity without a matching revenue guarantee, which supports factor C3. Development 2 shows capital expenditure hitting $116 billion against Azure at $100 billion, a ratio that pressures free cash flow (C2). Development 8 on the looser OpenAI alliance raises the risk that committed capacity earns less than planned (C4). These cost pressures are real and accumulating, making it more likely than not that margins or spending face stress by end of 2027, though the strong Azure revenue growth keeps the outcome from being very likely.

  • 1. First reported 23 September 2026, 3 outlets. Cited by the reading.
    “TSMC plans to lift wafer foundry prices by 3% to 6% from 2027” · eastasiabrief.com · 28 September 2026 · Open the report
    “TSMC reportedly to raise foundry prices 3%-6% from January 2027, with advanced nodes leading the increase” · finance.biggo.com · 23 September 2026 · Open the report
    “TSMC Foundry Prices to Rise 3%-6% in 2027 as 2nm, 3nm Demand Stays Booked to 2030 | AlphaPilot” · alphapilot.tech · 24 September 2026 · Open the report
  • 2. First reported 9 October 2026, 2 outlets. Cited by the reading.
    “Azure Tops $100B as Cloud Growth Accelerates | TMC Insight” · insight.tmcnet.com · 9 October 2026 · Open the report
    “Azure Surpasses $100 Billion as Microsoft Capital Expenditures Hit $116 Billion - Time News” · time.news · 10 October 2026 · Open the report
  • 3. First reported 23 September 2026, 2 outlets. Cited by the reading.
    “Report: TSMC to Raise Wafer Foundry Prices by 3% to 6% Starting January Next Year, Order Visibility Extended to 2030 | Bitget News” · bitget.com · 23 September 2026 · Open the report
    “TSMC to Raise Wafer Foundry Prices 3%–6% From January — Die Signal” · diesignal.com · 29 September 2026 · Open the report
  • 4. First reported 12 October 2026, 1 outlet.
    “This Is the Key for Microsoft Stock, Says Top Investor” · tipranks.com · 12 October 2026 · Open the report
  • 5. First reported 9 October 2026, 1 outlet.
    “Who Actually Pays for the AI Data Centers Built for OpenAI and Anthropic” · coinpaper.com · 9 October 2026 · Open the report
  • 6. First reported 6 October 2026, 1 outlet. Cited by the reading.
    “TSMC Foundry Prices Expected to Rise Up to 10% Next Year, Signaling Domino Effect Across Fabless Industry” · finance.biggo.com · 6 October 2026 · Open the report
  • 7. First reported 5 October 2026, 1 outlet.
    “[News] TSMC Earnings Preview: Terafab Talks, Price Hikes and A14-Intel Race Lead Five Key Themes” · trendforce.com · 5 October 2026 · Open the report
  • 8. First reported 1 October 2026, 1 outlet. Cited by the reading.
    “Microsoft and OpenAI Rewrite Their AI Alliance: Freedom, Caps and a Looser Bond” · webpronews.com · 1 October 2026 · Open the report
  • 9. First reported 29 September 2026, 1 outlet. Cited by the reading.
    “TSMC is raising 3nm chip prices up to 15% and nobody can say no - Startup Fortune” · startupfortune.com · 29 September 2026 · Open the report
  • 10. First reported 26 September 2026, 1 outlet. Cited by the reading.
    “Price increases spread from foundry wafers to power chips and Azure renewals — Zero One | Zero One Investment Research” · 01.co · 26 September 2026 · Open the report
Uncertainty

Causes of uncertainty

  • From July 2026 Microsoft treats its data centre buildings as lasting 25 years in place of 15, which slightly lowers depreciation and moves more of its new data centre leases out of the capital spending it reports, without changing what is built.
  • The gross margin on the Microsoft Cloud mixes AI with older cloud and software business, so a fall can reflect a change in the mix as much as the cost of AI.
  • A growing part of Microsoft's AI capacity is rented from specialized providers under contracts whose terms are only partly public.
What would change it

What would change the judgment

  • Microsoft's operating profit falling from a year earlier while its capital spending keeps rising.
  • Microsoft lowering its capital spending outlook and citing demand or returns.
  • An AI model developer reducing, delaying or renegotiating a large commitment for Azure capacity.
Causal factors

The causal factors, and which way each one cuts

FactorWhich way it cutsWhat evidence looks likeLast moved
C1. Cloud gross margin falling as AI costs rise
The gross margin on the Microsoft Cloud, which Microsoft reports each quarter, falling because the cost of AI capacity, mainly depreciation and running costs, grows faster than the revenue it brings. The Microsoft Cloud is Microsoft's measure that combines Azure, its business Microsoft 365 cloud services and its industry cloud services.
Supports. A stronger factor supports the assessment.
  • Microsoft reports a lower gross margin percentage for the Microsoft Cloud and links it to spending on AI capacity.
  • A quarterly filing shows depreciation growing faster than revenue.
  • Microsoft reports a higher cloud gross margin and cites efficiency, which bears on the factor in the other direction.
Not yet
C2. Free cash flow squeezed by spending
The cash Microsoft's operations generate, less what it spends on property and equipment, falling as spending rises faster than the cash its revenue brings in.
Supports. A stronger factor supports the assessment.
  • An earnings release shows free cash flow lower than a year earlier because of higher capital spending.
  • Microsoft relies more on leases or borrowing to pay for new capacity.
  • Microsoft reports a quarter in which its spending on property and equipment exceeded the cash its operations generated.
Not yet
C3. Higher prices for chips, memory and power
Rising prices for the components and power that AI capacity needs, which raise the cost of each unit of capacity without raising the revenue it brings.
Supports. A stronger factor supports the assessment.
  • Microsoft links part of a rise in its capital spending to higher prices for components.
  • Makers of memory chips or AI accelerators raise their prices.
  • Power contracts for new data centres are signed at higher prices.
Not yet
C4. AI model developers paying less, later or elsewhere
AI model developers that have committed to Azure capacity reducing, delaying or renegotiating those commitments, or moving work to other providers, which leaves capacity built for them earning less than planned.
Supports. A stronger factor supports the assessment.
  • Microsoft reports a reduction or renegotiation of a large commitment from an AI model developer.
  • An AI model developer reports difficulty raising the money to pay for computing it has contracted.
  • An AI model developer moves work from Azure to another cloud provider.
Not yet
C5. Operating profit growing alongside spending
Microsoft's operating profit, its operating income, growing year over year while its capital spending rises. This shows revenue covering the rising cost, which weakens this assessment.
Weakens. A stronger factor weakens the assessment.
  • An earnings release shows operating income up from a year earlier while capital spending rises.
  • Microsoft keeps its outlook for operating margin close to the previous year's.
Not yet
C6. Projects slowed, paused or dropped for demand or returns
Microsoft slowing, pausing or cancelling data centre projects, or walking away from leases and contracts for rented capacity, where the reason given is demand or returns. A change Microsoft attributes to accounting or to the timing of deliveries does not count.
Supports. A stronger factor supports the assessment.
  • Microsoft lowers its capital spending outlook and cites demand or returns.
  • Microsoft is reported to have walked away from data centre leases or contracts for rented capacity.
  • Microsoft pauses a data centre project and cites demand.
Not yet
Assumptions

What the assessment assumes, and what would break it

Alternatives

The rival explanations

Evidence moves the assessment only when it bears differently on the assessment than on at least one of these. Evidence that fits them equally carries no weight.

  • CX1. Margins fall because Microsoft builds ahead of business it has already signed, and they recover as that business arrives as revenue.
    No development was recorded on this assessment in the last 7 days.
    Favoured by:
    • The cloud gross margin falls while revenue under contract and not yet recognized keeps rising.
    • Microsoft says new capacity is committed to customers before it is built.
    Ruled out by:
    • The cloud gross margin keeps falling after the contracted business begins to arrive as revenue.
  • CX2. Spending slows because capacity becomes cheaper to build and run, while revenue from AI keeps growing.
    No development was recorded on this assessment in the last 7 days.
    Favoured by:
    • Microsoft reports a lower cost for each unit of AI work it delivers.
    • Growth in capital spending slows while Azure keeps growing quickly.
    Ruled out by:
    • Capital spending and Azure growth slow together.
  • CX3. Spending moves into leases and rented capacity, so Microsoft's own capital spending understates the cost of its AI capacity.
    No development was recorded on this assessment in the last 7 days.
    Favoured by:
    • Microsoft signs large contracts to rent computing capacity from specialized providers.
    • Leases not yet started grow while capital spending levels off.
    Ruled out by:
    • Leases and contracts for rented capacity shrink along with capital spending.

A development is recorded when the rule applies it to the assessment or sets it aside. Each count comes from the readings sealed with the changes, and a development set aside as fitting every alternative equally tells no alternative apart.

Evidence

Evidence behind changes in the last 30 days

Each item is a report or data series cited by a sealed change on this assessment. The update loop reads headlines and summaries, and no summary is shown as the evidence.

No evidence has been cited in the last 30 days.

Considered, set aside

Evidence considered and set aside

Developments the update loop read in the last 30 days and set aside because they fitted this assessment and each of its alternatives equally. Under the rule such evidence carries no weight, so none of it moved anything.

Nothing on this assessment was set aside in the last 30 days.

Changes

Changes to this assessment

No change to this assessment yet.

The other assessments

The rest of the judgment

The question’s text on this page is as sealed on the record’s first day. Fingerprint . Fingerprint pending. Records are fingerprinted each day at 23:59 UTC.