Businesses across many industries, paying for AI in Azure and in Microsoft 365, are the main source of Microsoft's AI growth through 2027, ahead of OpenAI and the other AI model developers that rent its computing capacity, and ahead of consumers.

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

The judgment as stated before the record began

The main assessment rests on this one, because revenue from many businesses is steadier than revenue from a few AI model developers, and more likely to keep covering the cost of new capacity. Microsoft earns from businesses' use of AI in two main ways: through computing and AI services in Azure, its cloud platform, and through Copilot and AI agents built into Microsoft 365 and its other business software. A large share of the new business it has signed, however, comes from AI model developers, the companies such as OpenAI that build AI models and rent computing capacity on a scale no ordinary business approaches. This assessment holds that businesses across many industries remain the larger source of growth. It would weaken if Microsoft's contracted business outside OpenAI stopped growing, if OpenAI came to account for most of the growth in Azure, or if paid use of Copilot stalled while AI model developers kept signing large commitments.

How the record started this assessment

On 15 October 2026, before the record began, a model read the 10 developments below five separate times, first reported in the three weeks before and linked to this assessment's causal factors or to its rivals', and chose a term each time. A second model, from a different family, checked each development a reading relied on, disputed 5 developments, and the five readings were made again without them. The record started the assessment at the average of the five final terms, taken in log-odds: 67.5 per cent, Likely (55–80%). Every reading and every check is kept in the first day's record.

The five final readings, in order: Likely, Likely, Likely, Likely, Likely. Their average, taken in log-odds, is 0.73, a stated confidence of 67.5 per cent. The readings were made by Claude Sonnet 4.6 (Anthropic) and checked by Gemini 4 Argon (Google).

Factors named above: R1 (Azure growth spread across many customers), R2 (Paid Microsoft 365 Copilot seats), R4 (Growth that rests on a few AI model developers), R5 (Consumer AI growth at Microsoft) and R6 (AI agents at work and charges based on use).

The developments shown

1. 1. First reported 14 October 2026, 1 independent publisher. Linked to this assessment's own causal factors.
   “Q3 2026: AI Labs Became Direct Buyers. Here Is Who Stands Behind Them.” · globaldatacenterhub.com · 14 October 2026 · Open the report · archived copy
   First round: relied on by 5 readings as counting against it. Removed from the second round, because the second model disputed it.
   • Check of its use as counting against the assessment: Disputed by the second model, so the second round was read without it, run on Gemini 4 Argon (Google).

     Its reason, as the model wrote it: Factor R4 measures the share of Microsoft's AI revenue and contracted business that comes from a few AI model developers. Report 1 describes an industry trend of AI labs contracting directly for data-center capacity and does not mention Microsoft or Azure at all. Therefore, it does not report any change in Microsoft's revenue or contracted business from AI model developers.

     Its ground: what the reports state is not a change in this causal factor.
     • Report 1, quoted: “AI Labs Became Direct Buyers.”
     • Report 1, quoted: “The report says AI labs are increasingly contracting directly for data-center capacity, citing large commitments by Anthropic and OpenAI that illustrate the race to secure infrastructure.”
2. 2. First reported 13 October 2026, 1 independent publisher. Linked to this assessment's own causal factors.
   “Broadcom CEO follows a proven blueprint for labs buying its chips” · finance.yahoo.com · 13 October 2026 · Open the report · archived copy
   First round: relied on by no reading. Second round: relied on by no reading.
   No reading relied on it, so it was not checked.
3. 3. First reported 13 October 2026, 1 independent publisher. Linked to this assessment's own causal factors.
   “Land Your Offer | Anatomy of a Revenue-Generating Partner Offer, Part 3: Copilot” · techcommunity.microsoft.com · 13 October 2026 · Open the report · archived copy
   First round: relied on by 2 readings as support for the assessment. Removed from the second round, because the second model disputed it.
   • Check of its use as support for the assessment: Disputed by the second model, so the second round was read without it, run on Gemini 4 Argon (Google).

     Its reason, as the model wrote it: Factor R2 measures the number of paid Microsoft 365 Copilot seats reported by Microsoft and the number of employers paying for Copilot for most of their staff. Report 1 describes partner guidance and an incentive to encourage active deployment rather than reporting an increase in paid seats or employer purchases. If anything, noting the gap between purchased seats and actual adoption does not indicate growth in paid seats across employers.

     Its ground: what the reports state is not a change in this causal factor.
     • Report 1, quoted: “Land Your Offer | Anatomy of a Revenue-Generating Partner Offer, Part 3: Copilot”
     • Report 1, quoted: “Microsoft’s new partner incentive rewards deployment and growth in active Microsoft 365 Copilot users, highlighting the gap between purchased seats and actual adoption among business customers.”
4. 4. First reported 12 October 2026, 1 independent publisher. Linked to this assessment's own causal factors.
   “October 2026 announcements - Partner Center” · learn.microsoft.com · 12 October 2026 · Open the report · archived copy
   First round: relied on by no reading. Second round: relied on by 4 readings as support for the assessment.
   No reading relied on it, so it was not checked.
5. 5. First reported 9 October 2026, 1 independent publisher. Linked to this assessment's own causal factors.
   “Who Actually Pays for the AI Data Centers Built for OpenAI and Anthropic” · coinpaper.com · 9 October 2026 · Open the report · archived copy
   First round: relied on by 2 readings as counting against it. Second round: relied on by 2 readings as support for the assessment and 3 readings as counting against it.
   • Check of its use as counting against the assessment: Upheld by the second model, run on Gemini 4 Argon (Google).

     Its reason, as the model wrote it: The report examines how long-term cloud and data-center commitments underpin the compute expansion of major AI model developers like OpenAI and Anthropic. This directly relates to factor R4 by highlighting the scale of infrastructure commitments driven by a few leading AI model developers, which counts against the assessment that growth comes mainly from a broad base of enterprise customers.

     • Report 1, quoted: “Who Actually Pays for the AI Data Centers Built for OpenAI and Anthropic”
     • Report 1, quoted: “The analysis outlines how long-term cloud and data-center commitments underpin OpenAI and Anthropic’s compute expansion”
6. 6. First reported 8 October 2026, 1 independent publisher. Linked to this assessment's own causal factors.
   “Microsoft shows off new Windows software, revamped for agentic AI” · fortune.com · 8 October 2026 · Open the report · archived copy
   First round: relied on by 3 readings as support for the assessment. Removed from the second round, because the second model disputed it.
   • Check of its use as support for the assessment: Disputed by the second model, so the second round was read without it, run on Gemini 4 Argon (Google).

     Its reason, as the model wrote it: Factor R6 covers businesses running AI agents in their everyday operations and paying for AI by the amount they use through Azure, Microsoft 365, and Microsoft's other business software. Report 1 describes Microsoft revamping the Windows operating system to embed AI agents, with no mention of business adoption, usage-based pricing, or enterprise software like Azure or Microsoft 365. Embedding agent features into a general operating system is not a change in business usage or usage-based monetization as defined by R6, nor does it separate business adoption from consumer adoption (RX2).

     Its ground: what the reports state is not a change in this causal factor.
     • Report 1, quoted: “Microsoft shows off new Windows software, revamped for agentic AI”
     • Report 1, quoted: “Microsoft is embedding AI agents and hybrid cloud-and-local capabilities into Windows to keep its operating system central as computing shifts toward autonomous AI tools.”
7. 7. First reported 8 October 2026, 1 independent publisher. Linked to this assessment's own causal factors.
   “Microsoft will let Copilot act on local files on Windows PCs – Computerworld” · computerworld.com · 8 October 2026 · Open the report · archived copy
   First round: relied on by 1 reading as support for the assessment. Removed from the second round, because the second model disputed it.
   • Check of its use as support for the assessment: Disputed by the second model, so the second round was read without it, run on Gemini 4 Argon (Google).

     Its reason, as the model wrote it: The only factor shown is R5, which measures consumer AI revenue growth through the Copilot app, Bing search advertising, and consumer Microsoft 365 subscriptions, and where a stronger factor undercuts the assessment. Report 1 describes a technical feature update allowing Copilot to act on local files on Windows PCs and use local AI models to help customers reduce costs and control data, which does not represent a change in consumer AI revenue or advertising growth under R5 that supports the assessment.

     Its ground: what the reports state is not a change in this causal factor.
     • Report 1, quoted: “Microsoft will let Copilot act on local files on Windows PCs”
     • Report 1, quoted: “Copilot will also use AI models running on the PC alongside cloud models to help customers reduce costs and gain more control over their data.”
8. 8. First reported 6 October 2026, 1 independent publisher. Linked to this assessment's own causal factors.
   “Wrtn Azure OpenAI Service Customer Story - The Data Pros” · the-data-pros.net · 6 October 2026 · Open the report · archived copy
   First round: relied on by 2 readings as support for the assessment. Second round: relied on by 5 readings as support for the assessment.
   • Check of its use as support for the assessment: Upheld by the second model, run on Gemini 4 Argon (Google).

     Its reason, as the model wrote it: Report 1 states that Wrtn Technologies uses Azure OpenAI Service to apply advanced models in real products and scale AI workloads on Azure infrastructure. This shows a business running and scaling AI workloads on Microsoft's Azure platform in its operations, which supports the assessment via factor R6 (and business adoption more broadly) even though the report does not explicitly mention South Korea.

     • Report 1, quoted: “The video, “Wrtn Azure OpenAI Service Customer Story,” highlights how Wrtn Technologies uses Azure OpenAI Service to: – Apply advanced models in real products – Scale AI workloads on Azure infrastructure”
9. 9. First reported 5 October 2026, 1 independent publisher. Linked to this assessment's own causal factors.
   “Companies are scaling AI. Do employees know how to use it?” · hcamag.com · 5 October 2026 · Open the report · archived copy
   First round: relied on by 4 readings as support for the assessment. Removed from the second round, because the second model disputed it.
   • Check of its use as support for the assessment: Disputed by the second model, so the second round was read without it, run on Gemini 4 Argon (Google).

     Its reason, as the model wrote it: Factor R6 specifically requires businesses running AI agents and paying for AI 'through Azure, Microsoft 365 and Microsoft's other business software'—that is, 'on Microsoft's products.' Report 1 describes a general KPMG survey about companies scaling AI and developing guidance for AI agents, with no mention of Microsoft, Azure, or Microsoft 365, so it does not reflect a change in R6.

     Its ground: what the reports state is not a change in this causal factor.
     • Report 1, quoted: “Companies are scaling AI. Do employees know how to use it?”
     • Report 1, quoted: “KPMG’s survey of large-company leaders found significant workforce adoption rose to 44%, but only 55% were developing role-specific guidance for AI agents, highlighting a gap between deployment and employee preparation.”
10. 10. First reported 5 October 2026, 1 independent publisher. Linked to this assessment's own causal factors.
    “AWS, HUMAIN to set up unique Saudi AI zone” · gulfconstructiononline.com · 5 October 2026 · Open the report · archived copy
    First round: relied on by no reading. Second round: relied on by no reading.
    No reading relied on it, so it was not checked.

The readings, and their reasons

Readings made before this one

No starting point was read before this one, as the first day records.

Causes of uncertainty

• Microsoft does not report its revenue from AI as a regular separate figure, so the sources of its AI growth have to be read from Azure, Copilot and contracted-business figures that also include other business.
• Microsoft has disclosed its revenue from OpenAI only in its annual report, as a single figure that includes OpenAI's revenue-sharing payments as well as payment for computing.
• From fiscal year 2027, which began in July 2026, Microsoft reports its business in two new segments and counts less of it as Azure, so figures from before and after the change need care when compared.
• Demand for Azure has exceeded the capacity Microsoft has available, so its growth reflects how fast Microsoft can build as well as how much each kind of customer wants.

What would change the judgment

• Microsoft reporting that its contracted business outside OpenAI has stopped growing.
• Microsoft's annual report showing that OpenAI accounted for more than half of the year's growth in Azure.
• Paid Copilot seats holding flat for two quarters while AI model developers sign large new commitments for Azure capacity.
• Several large employers choosing another provider's AI assistant for their whole workforce.

The causal factors, and which way each one cuts

What the assessment assumes, and what would break it

• RA1. Customers other than OpenAI keep adding to the business they have contracted with Microsoft.
  If this fails, the assessment fails.

  Invalidated if, on two consecutive quarterly earnings calls, Microsoft reports that its commercial remaining performance obligation excluding OpenAI, meaning the revenue that customers other than OpenAI have committed to under contract and Microsoft has not yet recognized, was lower than a year earlier. If Microsoft stops reporting the figure excluding OpenAI, the assumption is reviewed rather than invalidated.

  Source: Microsoft Investor Relations, transcript of the earnings call for the fourth quarter of fiscal year 2026, which reports growth in commercial remaining performance obligation excluding OpenAI, Microsoft Investor Relations, earnings releases, webcasts and call transcripts. The criterion’s fingerprint 32034ec6…d881 was fixed on 15 October 2026 and is part of the record’s first day.
• RA2. OpenAI accounts for less than half of the growth in Microsoft's Azure revenue.
  If this fails, the assessment weakens.

  Invalidated if the increase from fiscal year 2026 to fiscal year 2027 in revenue from commercial arrangements with OpenAI, a figure that includes OpenAI's revenue-sharing payments, as disclosed in the notes to Microsoft's annual reports on Form 10-K, is more than half the increase in Azure revenue over the same two fiscal years, with Azure revenue taken as Microsoft reports it under the definition it adopted for fiscal year 2027. Microsoft's fiscal year runs from July to June. If Microsoft stops disclosing either figure, the assumption is reviewed rather than invalidated.

  Source: Microsoft annual report on Form 10-K for fiscal year 2026, note on accounting policies, which discloses revenue from commercial arrangements with OpenAI, Microsoft, Form 8-K of September 2026, presentation on fiscal year 2027 segments and metrics, with Azure revenue restated under the new definition, SEC EDGAR, Microsoft's quarterly and annual reports (CIK 789019). The criterion’s fingerprint 9af3d8e6…2b12 was fixed on 15 October 2026 and is part of the record’s first day.
• RA3. Consumer AI remains a small part of Microsoft's growth from AI.
  If this fails, the assessment is little changed.

  Invalidated if, in two consecutive quarters, revenue in Microsoft's Devices and Consumer segment rises by a larger amount year over year than revenue in its Agents and Infra segment, as reported in the segment information of its quarterly report on Form 10-Q or annual report on Form 10-K. If Microsoft changes these segments, the assumption is reviewed rather than invalidated.

  Source: Microsoft, Form 8-K of September 2026, presentation on the Agents and Infra and Devices and Consumer segments from fiscal year 2027, SEC EDGAR, Microsoft's quarterly and annual reports (CIK 789019). The criterion’s fingerprint 08d1d653…7dec was fixed on 15 October 2026 and is part of the record’s first day.

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.

• RX1. AI model developers, OpenAI above all, are the main source of Microsoft's AI growth, so that growth rises and falls with a few customers.
  No development was recorded on this assessment in the last 7 days.
  Favoured by:
  • Microsoft's contracted business keeps rising quickly while the part outside OpenAI grows slowly.
  • Microsoft's annual report shows revenue from OpenAI growing faster than Azure.
  • AI model developers account for the largest new commitments for Azure capacity.
  Ruled out by:
  • Azure keeps growing quickly while AI model developers move work to other cloud providers.
  • Revenue from OpenAI grows more slowly than Azure over a full fiscal year.
• RX2. Consumers become a major source of Microsoft's AI growth, through the Copilot app, search advertising and AI in consumer subscriptions.
  No development was recorded on this assessment in the last 7 days.
  Favoured by:
  • Microsoft reports that consumer products contributed a large part of its growth in AI revenue.
  • Revenue in the Devices and Consumer segment grows faster than revenue in the Agents and Infra segment, and Microsoft links the difference to AI.
  Ruled out by:
  • Consumer revenue growth slows while business cloud revenue keeps growing quickly.

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 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.

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 rival explanations 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 to this assessment

No change to this assessment yet.

The rest of the judgment

• Main assessment
• The alternative

The question’s text on this page is as sealed on the record’s first day. Fingerprint d6c8f5a4…4dbd. Fingerprint pending. Records are fingerprinted each day at 23:59 UTC.