What is Microsoft's $37 billion AI run rate? Microsoft's $37 billion AI run rate is the annualized revenue from artificial intelligence services that CEO Satya Nadella disclosed during the company's Q3 fiscal 2026 earnings call on April 29, 2026, representing 123% year over year growth. The figure includes revenue from clients running AI workloads on Azure, AI model builders using Microsoft infrastructure, and Microsoft's own AI tools like Microsoft 365 Copilot. For commercial real estate professionals tracking AI commercial real estate demand, this number is the clearest signal yet that hyperscaler AI revenue is materializing fast enough to justify the data center buildout reshaping CRE investment in 2026.
Key Takeaways
- Microsoft's AI business reached a $37 billion annual run rate in Q3 FY26, up 123% year over year, accelerating from prior quarters and validating data center demand.
- Azure cloud revenue grew 40% (39% in constant currency), with management guiding Q4 growth to 39% to 40%, signaling no slowdown in enterprise AI consumption.
- Microsoft expects roughly $190 billion in calendar 2026 capital expenditures, including about $25 billion driven by higher component pricing rather than incremental scope.
- Q3 capex was $31.9 billion, up 49%, while gross margin slipped to 67.6%, the lowest since 2022, due to depreciation from new data center buildout.
- Commercial remaining performance obligations jumped 99% year over year to $627 billion, indicating booked but undelivered cloud and AI demand stretching years into the future.
Microsoft Azure Growth Validates the AI Data Center Thesis
Microsoft reported total Q3 FY26 revenue of $82.9 billion, up 18% (15% in constant currency), and operating income of $38.4 billion, up 20%, beating Wall Street's $81.39 billion revenue estimate and the $4.06 EPS consensus. Diluted EPS came in at $4.27, up 23% on a GAAP basis. The Intelligent Cloud segment, which houses Azure, generated $34.7 billion, up 30%. Microsoft Cloud as a whole reached $54.5 billion, up 29% (Source: Microsoft FY26 Q3 Press Release).
For CRE investors, the most consequential disclosure was the AI run rate itself. Nadella's $37 billion figure represents revenue clients are paying Microsoft today for AI infrastructure and tooling. That is not a forecast, not a forward booking, and not pipeline. It is the actualized run rate on AI workloads as of the quarter that ended March 31, 2026. The 123% year over year growth rate exceeded analyst expectations and is the strongest confirmation yet that AI demand on the supply side, where Microsoft sits, is matched by commercial revenue on the demand side.
The implications mirror the thesis behind Big Tech's $700 billion AI capex surge. Hyperscalers are not building speculative capacity. They are building against contracts and consumption. Commercial remaining performance obligations of $627 billion, up 99% year over year, represent the work Microsoft has booked but has not yet delivered. That backlog drives the megawatt absorption underwriters now use to justify long lease takedowns at facilities like Digital Realty's record 200 MW Charlotte AI lease.
$190 Billion in 2026 Capex and What It Funds
Microsoft expects to invest approximately $190 billion in capital expenditures during calendar year 2026. Roughly $25 billion of that figure reflects higher component pricing rather than incremental scope, but the absolute spend is still historic. Q3 alone delivered $31.9 billion in capex and finance leases, up 49% year over year, and the run rate implies sustained gigawatt scale data center buildout through 2027.
The composition of that spend is what CRE professionals should watch:
- Land and shell construction: Greenfield campuses in Tier 1 and Tier 2 markets, including Wisconsin, Atlanta, Phoenix, San Antonio, and additional Mountain West sites where power interconnects are available.
- Power infrastructure: Substations, transformers, and behind the meter generation. Power timing increasingly determines build timing more than land acquisition.
- GPU clusters and networking: NVIDIA Blackwell and incoming Vera Rubin platforms, plus Microsoft's own Maia accelerators.
- Cooling systems: Liquid cooled racks, immersion cooling, and water management infrastructure that meaningfully changes facility design specifications.
The 67.6% gross margin, lowest since 2022, reflects accelerating depreciation from these new facilities entering service. That tradeoff matters for CRE investors because depreciation pressure incentivizes Microsoft and peers to push utilization higher and faster, which lengthens lease durations and pushes up effective rents per kilowatt across the data center asset class. According to CBRE Global Data Center Research, primary U.S. data center vacancy rates are at multi decade lows, with power constrained markets seeing the steepest rate increases.
What CRE Data Center Investors Should Do Now
Microsoft's Q3 print sharpens four investment theses for commercial real estate:
- Power constrained markets keep pricing power. CBRE and JLL research has emphasized that power availability, not land, is the binding constraint. Microsoft's continued growth at 40% Azure cannot land in markets without immediate megawatts. Sites with shovel ready interconnects are repricing.
- Hyperscaler tenant credit improves the underwriting. A $37 billion AI run rate growing 123% improves the operating coverage on Microsoft leased facilities. Lenders are widening proceeds and tightening cap rate spreads on these assets accordingly.
- Multi cloud is increasing competition for sites. The recent OpenAI Microsoft multi cloud deal amendment means OpenAI workloads will route across AWS, Oracle, and Microsoft simultaneously. Sites that can host multiple anchor tenants gain optionality.
- Component pricing is absorbing capex headroom. The $25 billion component cost layer means hyperscalers will negotiate harder on fee structures and tenant improvement allowances. Operators with in place power should hold pricing leverage.
If you are evaluating data center exposure or repositioning industrial sites for AI workloads, The AI Consulting Network can help you stress test the underwriting against hyperscaler capex curves, power availability data, and current market rents per kilowatt.
Microsoft 365 Copilot Adoption Has CRE Workflow Implications
Beyond infrastructure, Microsoft disclosed that paid Microsoft 365 Copilot seats crossed 20 million in Q3, up from 15 million at the end of January 2026. Year over year seat additions grew 250%. Accenture alone deployed more than 740,000 Copilot seats. Bayer, Johnson and Johnson, Mercedes Benz, and Roche each committed to 90,000 or more seats in the quarter.
For CRE professionals, the practical signal is that enterprise AI adoption inside corporate occupiers is accelerating. As of Q1 2026, 92% of corporate occupiers had initiated AI programs, though only 5% report achieving most program goals. With 20 million paid Copilot seats now in production, AI shaped office demand is becoming a measurable driver of leasing decisions, not a forecast item. CRE investors looking for hands on AI implementation support can reach out to Avi Hacker, J.D. at The AI Consulting Network for tactical guidance on pricing, negotiating, and underwriting these dynamics.
Frequently Asked Questions
Q: What does Microsoft's $37 billion AI run rate include?
A: It captures annualized AI revenue across Azure AI services, AI model providers running on Microsoft infrastructure, Microsoft 365 Copilot subscriptions, GitHub Copilot, and other Microsoft AI tooling, as disclosed by Satya Nadella in the Q3 FY26 earnings call on April 29, 2026.
Q: How does Microsoft's Q3 FY26 print affect CRE data center valuations?
A: Sustained 40% Azure growth and a 99% jump in commercial remaining performance obligations to $627 billion validate hyperscaler tenant demand. That underpins long lease takedowns at Digital Realty, Equinix, and DataBank, which support tighter cap rates on stabilized data center assets.
Q: Will Microsoft's $190 billion 2026 capex slow if AI revenue disappoints?
A: It would be hard to slow in the near term. Capex is largely committed against multi year construction contracts, GPU purchase orders, and power agreements. The component pricing portion may compress, but the underlying buildout is contractually inflexible in the short run.
Q: How should CRE investors price Microsoft tenant risk?
A: Microsoft is one of the highest credit hyperscaler tenants in the market. The 123% AI growth and Q4 guidance of 39% to 40% Azure growth in constant currency suggests stable, growing demand for the leased capacity, supporting long, fixed rate underwriting and lower cap rates.
Q: What is the difference between Azure AI revenue and Microsoft's AI run rate?
A: Azure AI revenue is included in the $37 billion run rate but is not the entire figure. The run rate also includes Microsoft 365 Copilot seats, GitHub Copilot, and other Microsoft AI products that are not delivered through Azure infrastructure billing.