Microsoft’s commercial cloud backlog crossed $678 billion at the close of its fiscal fourth quarter, an 43% year-over-year growth for Azure, alongside 34% for Microsoft cloud services. The July 29, 2026 report landed eight days after Alphabet posted an 32% year-over-year growth for Google Cloud and, one day later, Amazon confirmed AWS grew 17%.7% in its own second-quarter results. For the first time this year, all three hyperscalers reported in the same eight-day window, giving investors and enterprise buyers a rare side-by-side look at how the AI infrastructure boom is actually landing in the numbers.
The timing matters. Microsoft Azure earnings have been the market’s proxy for AI infrastructure demand for two years, and this quarter’s results arrive just months after analysts were warning that capacity limits, not demand, were capping Azure’s growth. Read together, the three reports describe a cloud market that grew faster in the same three months than at almost any point since the ChatGPT-driven boom began.
Don't miss new tech stories on Google
Add Tech Insider once in the Google app and our stories appear in your news suggestions.
Azure’s Backlog Hits $678 Billion: Breaking Down the Numbers
Microsoft doesn’t publish a standalone dollar figure for Azure revenue. It reports Azure and other cloud services as a growth rate against Microsoft Cloud, the umbrella figure that also includes Microsoft 365, Dynamics 365, and other subscription products. For the quarter ended June 30, 2026, Microsoft Cloud revenue reached $59.3 billion, up 27% year over year, while Azure and other cloud services revenue itself grew 43%, according to Microsoft’s own fiscal Q4 2026 earnings materials.
The more striking number sits below the revenue line. Commercial remaining performance obligation, the contracted revenue Microsoft has booked but not yet delivered, reached $678 billion, up 43% from a year earlier. That backlog functions as a forward indicator: it tells investors how much AI and cloud spending is already under contract, even if the data centers to serve it aren’t fully built yet. CEO Satya Nadella tied the quarter to a specific milestone, saying: “This year, Azure revenue surpassed $100 billion for the first time, and Microsoft 365 Copilot reached over 30 million paid seats, reflecting the confidence customers are placing in us to power their AI transformation,” according to Microsoft’s Q4 earnings coverage on Yahoo Finance.
Why Azure Growth Jumped to 43% in the Fourth Quarter
Azure’s growth rate has climbed through fiscal 2026 rather than arriving in one jump. The segment grew 40% in the first fiscal quarter, according to Microsoft’s Q1 FY2026 Intelligent Cloud performance report, which stated plainly: “Azure and other cloud services revenue grew 40% driven by demand for our portfolio of services with continued growth across all workloads.” Growth then eased slightly to the high 30s in the second quarter before climbing back to 43% in the quarter reported in July.
That mid-year dip is where the capacity story comes in. Microsoft’s own Q2 FY2026 materials described an unusual operational fix: shifting older infrastructure originally built for other workloads onto Azure demand. The company’s fiscal Q2 2026 earnings release noted that “revenue grew 39% and 38% in constant currency, slightly ahead of expectations, with ongoing efficiency gains across our fungible fleet enabling us to reallocate some capacity to Azure that was monetized in the quarter.” In plain terms, Microsoft squeezed more Azure revenue out of servers it already owned rather than waiting for new data centers to come online. By the fourth quarter, newer capacity had caught up, and growth accelerated again.
AWS and Google Cloud Report Similar Momentum This Quarter
Azure wasn’t the only hyperscaler having a strong quarter. Alphabet reported Google Cloud revenue of $24.8 billion, up 82% year over year, in results announced July 22, 2026, a jump CNBC covered live as one of the largest cloud growth rates posted by any major provider this cycle. Amazon followed eight days later, reporting AWS revenue of $42.2 billion, up 36.7% year over year, in a report CNBC published July 30, 2026.
The table below lines up all three results side by side. Note that Microsoft’s figure reflects growth in “Azure and other cloud services” rather than a disclosed dollar amount, since the company folds Azure into the broader Microsoft Cloud line rather than breaking it out on its own.
| Provider | Segment | Quarter | Report Date | Segment Revenue | YoY Growth |
|---|---|---|---|---|---|
| Microsoft Azure | Azure and other cloud services | Q4 FY2026 | July 29, 2026 | Not disclosed separately (Microsoft Cloud total: $59.3B) | 43% |
| Google Cloud | Google Cloud | Q2 2026 (calendar) | July 22, 2026 | $24.8 billion | 82% |
| Amazon AWS | AWS | Q2 2026 (calendar) | July 30, 2026 | $42.2 billion | 36.7% |
AWS remains the largest cloud business by revenue even with the slowest growth rate of the three. Google Cloud is the smallest by revenue but grew the fastest, a pattern that has held for several quarters as it expands from a smaller base. Azure sits in between on both measures, though Microsoft’s refusal to disclose a hard dollar figure makes a precise ranking harder to pin down.
From Capacity-Constrained to Capacity-Confident: What Changed
Earlier in fiscal 2026, Azure’s story was about limits, not demand. Coverage throughout late 2025 and early 2026 described Microsoft turning away or delaying some Azure customers because data centers, power hookups, and GPU supply hadn’t caught up with orders. Reuters framed the July 29 report as a turning point, reporting that Microsoft had topped quarterly cloud growth estimates while easing spending concerns that had weighed on the stock for much of the year.
The backlog table below shows how that shift played out quarter by quarter. Backlog growth actually peaked in the second fiscal quarter at 110%, well ahead of revenue growth, meaning Microsoft was signing contracts faster than it could deliver capacity to fill them. By the fourth quarter, backlog growth had cooled to 84% while revenue growth accelerated to 43%, a sign that new capacity was finally catching up with the contracts Microsoft had already signed.
| Fiscal Quarter | Report Date | Commercial Backlog (RPO) | Backlog YoY Growth | Azure YoY Growth | Microsoft Cloud Revenue |
|---|---|---|---|---|---|
| Q1 FY2026 | Late Oct. 2025 | $392 billion | +51% | 40% | Not disclosed |
| Q2 FY2026 | Jan. 28, 2026 | $625 billion | +110% | 38-39% | $50B+ (+26%) |
| Q4 FY2026 | July 29, 2026 | $678 billion | +84% | 43% | $59.3B (+27%) |
Independent trackers had put Microsoft’s 2026 capital spending in the range of $175 billion for the calendar year, most of it aimed at data centers, networking gear, and AI accelerators, though Microsoft does not report that figure as a single line item, so it should be read as an industry estimate rather than a company disclosure.
The OpenAI Concentration Question Inside Microsoft’s Backlog
A meaningful share of that $678 billion backlog traces back to a single customer relationship. Microsoft has disclosed that roughly 45% of its commercial remaining performance obligation ties to OpenAI, a concentration that Reuters has flagged in its coverage of the company’s backlog disclosures. That figure explains why analysts keep pressing Microsoft executives about customer diversification on earnings calls: a backlog that large, tied that heavily to one counterparty, carries different risk than the same number spread across thousands of enterprise accounts.
Microsoft has pushed back on the concentration framing by pointing to sequential growth outside its largest AI lab relationships. The company’s own materials describe broadening demand from traditional enterprise customers running everything from SAP migrations to Copilot deployments, not just frontier AI labs renting raw compute. Still, the OpenAI relationship remains the single largest swing factor in how that backlog eventually converts to recognized revenue, and any change in OpenAI’s compute strategy, including a move to diversify across providers, would show up in Microsoft’s numbers before almost anywhere else.
Inside the Hyperscaler Capex Race
None of this quarter’s growth happens without spending that dwarfs prior cycles. Microsoft, Amazon, and Google have each guided to capital expenditure levels that would have been unthinkable five years ago, funneled into data center shells, power purchase agreements, networking, and GPU or custom silicon orders. That spending has already reshaped adjacent markets. Data center landlords have signed record-setting lease deals with the three hyperscalers, a trend covered in tech-insider.org’s reporting on $850 billion in data center leases, and cloud pricing on GPU-heavy instances has moved up rather than down for the first time in years, a shift tracked in coverage of AWS’s second EC2 GPU price hike of 2026.
The capex race also explains why all three companies keep emphasizing efficiency alongside growth. Building data centers faster than power utilities can deliver electricity has become the binding constraint across the industry, not chip supply alone. That is why Microsoft highlighted its “fungible fleet” strategy this year: reallocating existing capacity is faster than building new capacity, even for a company spending well over $100 billion a year.
What Microsoft Executives Are Saying
Executive commentary this earnings season has shifted from managing expectations about shortages to framing growth as durable. Nadella’s comment about Azure’s $100 billion annual revenue milestone and 30 million paid Copilot seats, cited above, was the headline figure Microsoft chose to lead with, a deliberate pairing of infrastructure revenue with software adoption meant to show that AI spending is translating into recurring product usage, not just one-time compute contracts.
That framing matters for how the market reads the numbers. A backlog built on infrastructure commitments alone is a bet that demand will materialize. A backlog paired with 30 million paid Copilot seats and Azure revenue growth accelerating rather than decelerating is closer to proof that the demand already exists. Investors responded accordingly: shares moved higher in the days following the report, part of why Reuters characterized the results as easing, rather than deepening, the spending concerns that had followed Microsoft through the middle of the fiscal year.
Historical Context: The Two-Year Run-Up to This Moment
None of this happened overnight. The current cycle traces back to the generative AI wave that started in late 2022 and accelerated through 2023, when enterprise interest in large language models turned cloud GPU capacity into the industry’s scarcest resource. Microsoft’s early relationship with OpenAI gave Azure a head start in AI-specific workloads, while Amazon and Google spent 2023 and 2024 building their own answers: AWS through Bedrock and custom Trainium chips, Google through its TPU line and Gemini models.
By 2025, the conversation had shifted from whether AI demand was real to whether the hyperscalers could physically build fast enough to capture it, the capacity-constrained narrative that dogged Azure through the first half of fiscal 2026. The current quarter is best read as the moment that question got a partial answer: demand held up, and the infrastructure investments made over the prior two years started converting into revenue at a faster clip than new orders were arriving.
Competitive Comparison: Azure vs AWS vs Google Cloud for AI Workloads
Where Azure Leads
Azure’s advantage remains its direct line to OpenAI’s models through Azure OpenAI Service, plus deep integration with Microsoft 365, GitHub, and enterprise identity tools that most large companies already run. For organizations standardized on Microsoft’s stack, deploying AI features through Azure requires fewer new vendor relationships than starting fresh with a competitor, and the 30 million Copilot seats Nadella cited show that packaging advantage converting into paid adoption.
Where AWS and Google Cloud Push Back
AWS counters with scale and price competition, backed by Graviton and Trainium silicon designed to undercut Nvidia-based instance pricing, an advantage detailed in tech-insider.org’s comparison of AWS Graviton against Intel and AMD alternatives. Google leans on vertical integration between its TPU hardware and Gemini model family, which helps explain why Google Cloud posted the fastest growth of the three this quarter even from a smaller revenue base. Enterprise buyers increasingly split workloads across more than one provider specifically to avoid depending on any single hyperscaler’s capacity and pricing decisions.
What the Capacity Squeeze Means for Enterprise Buyers
For IT and finance teams managing cloud budgets, this earnings cycle is a signal to revisit capacity planning rather than assume prices and availability will hold steady. Reserved capacity and committed-use pricing have become more valuable as on-demand GPU instances face intermittent availability, a dynamic covered in tech-insider.org’s breakdown of AWS Reserved Instances, Savings Plans, and Spot pricing. Cloud financial operations teams are also under more pressure to justify spend as AI workloads inflate bills faster than traditional compute did, a trend detailed in coverage of cloud waste hitting 29% of AI spend.
The practical takeaway is that enterprises negotiating new cloud contracts in the second half of 2026 have less leverage than they did two years ago. Demand across all three major providers is outpacing supply in specific regions and GPU families, which means buyers who lock in capacity commitments now are trading flexibility for certainty, a trade that made little sense when capacity was abundant and now looks more like standard risk management.
Checking Your Own Cloud Quota and Reserved Capacity
Engineering and platform teams can check their own exposure to regional capacity limits directly from the command line before committing to new deployments. The commands below pull current quota and usage figures from Azure and AWS respectively.
# Check Azure VM quota and current usage by region
az vm list-usage --location eastus --output table
# Check AWS EC2 service quotas
aws service-quotas list-service-quotas --service-code ec2 --output tableRunning these checks against high-demand regions, particularly US East and other data center hubs where hyperscalers have concentrated AI capacity, gives a faster read on where quota increases might take longer to approve than usual. Teams planning large GPU deployments should request quota increases well ahead of launch dates rather than assuming same-week approval, given how tight allocations have run in the regions where AI training and inference workloads cluster.
Market Ripple Effects Across Chips and Data Centers
Strong hyperscaler earnings ripple outward fast. Chipmakers supplying GPUs and custom silicon to all three companies benefit directly from capex guidance that keeps climbing, and memory suppliers have already seen pricing move on the back of AI demand from the same data center buildout. Data center developers and landlords are the other clear beneficiary, locking in multi-year lease commitments from hyperscalers racing to secure power and shell space before competitors do.
The effects reach beyond the AI supply chain narrowly defined. Utilities in regions with heavy data center concentration are renegotiating power agreements and, in some cases, delaying retirements of existing generation capacity to meet hyperscaler demand. That dynamic has drawn regulatory attention in multiple US states, adding a policy dimension to what started as a pure infrastructure spending story.
Five Predictions for Cloud Earnings Through 2027
Based on the trajectory across all three hyperscalers this earnings season, here is how the next several quarters are likely to unfold. These are analyst-style projections, not confirmed guidance from any of the companies involved.
- Backlog-to-revenue conversion becomes the next scrutiny point. Investors who spent early 2026 worrying about capacity will spend late 2026 asking how fast Microsoft can turn $678 billion in signed contracts into recognized revenue, making conversion rate the metric to watch at the fiscal Q1 2027 report expected around October 2026.
- OpenAI concentration disclosure pressure increases. Expect analysts to keep pushing Microsoft for more granular customer-mix data, and expect Microsoft to keep emphasizing enterprise and Copilot growth to dilute the OpenAI concentration story.
- AWS leans harder on custom silicon to defend margins. With the slowest growth rate of the three this quarter, expect Amazon to spotlight Trainium and Graviton price-performance claims more aggressively at its re:Invent conference later this year.
- Google Cloud’s 82% growth pace cools in the second half of 2026. That rate reflects easy comparisons against a smaller prior-year base. Expect deceleration into the 60s or 50s as year-over-year comparisons get tougher, without that implying weaker underlying demand.
- Multi-cloud hedging keeps growing among large AI buyers. Enterprises burned by regional capacity limits in 2025 and early 2026 are likely to keep splitting workloads across two or more providers as standard risk management, not a temporary reaction.
Frequently Asked Questions
How big is Microsoft’s cloud backlog in 2026?
Microsoft’s commercial remaining performance obligation reached $678 billion in the quarter reported July 29, 2026, up 43% from a year earlier.
Why did Azure growth accelerate to 43%?
Growth climbed as new data center and GPU capacity came online through fiscal 2026, following a stretch where Microsoft relied on reallocating existing infrastructure to meet Azure demand while new capacity was still being built.
Is Azure still capacity-constrained in 2026?
Less than it was earlier in the fiscal year. Reuters described the July 2026 results as easing spending and capacity concerns that had weighed on the stock, though Microsoft continues to invest heavily in new capacity rather than describing supply as fully caught up with demand.
How much of Microsoft’s backlog comes from OpenAI?
Microsoft has disclosed that roughly 45% of its commercial remaining performance obligation is tied to OpenAI, a concentration Reuters has covered in its reporting on the company’s backlog.
How does AWS growth compare to Azure in 2026?
AWS revenue grew 36.7% year over year to $42.2 billion in the quarter reported July 30, 2026, slower than Azure’s 43% growth rate but on a larger disclosed revenue base.
How does Google Cloud growth compare to Azure and AWS?
Google Cloud grew 82% year over year to $24.8 billion in the quarter reported July 22, 2026, the fastest growth rate of the three hyperscalers this cycle, though from the smallest revenue base of the group.
What is the difference between Microsoft Cloud revenue and Azure revenue?
Microsoft Cloud revenue, $59.3 billion in the most recent quarter, includes Azure alongside Microsoft 365, Dynamics 365, and other subscription products. Microsoft reports Azure’s performance as a growth percentage rather than a standalone dollar figure.
Will cloud prices rise because of AI capacity constraints?
GPU-heavy instance pricing has already moved higher at some providers in 2026 rather than following the historical trend of falling cloud prices, making reserved capacity and committed-use discounts more valuable for budget planning than in prior years.
Related Coverage
- Google Cloud Hits 82% Growth Ahead of AWS Earnings [2026]
- Data Center Leases Hit $850B, Meta, Microsoft Lead [2026]
- Cloud Waste Hits 29% as AI Spend Breaks Budgets [2026]
- AWS Hikes EC2 GPU Pricing 20%, Second Time in 2026
- AWS Reserved vs Savings Plans vs Spot: 90% Off [2026]
- AWS Graviton vs Intel/AMD: 45% Cheaper, 25% Faster [2026]
- Pinterest Commits $4B to AWS Through 2031 [2026]
