AWS vs Azure vs Google Cloud: What Q3 2026 Earnings Reveal About AI Infrastructure

AWS vs Azure vs Google Cloud has become a more revealing financial comparison following the third-quarter 2026 earnings reports. Amazon Web Services generated $45.984 billion in revenue, Microsoft’s Azure business recorded $33.165 billion, and Google Cloud reached $30.709 billion. The numbers show exceptional demand for computing infrastructure, but the companies’ accounting categories tell different stories.

The quarter ended September 30 delivered a particularly important reporting change: Microsoft disclosed standalone Azure revenue in dollars under its revised financial reporting structure.

That gives researchers a clearer view of the size of Microsoft’s cloud computing business. Yet comparing AWS, Azure and Google Cloud remains more complicated than simply lining up three revenue figures.

Google Cloud includes Workspace, cloud services and TPU-system sales. Azure is now disclosed separately from Microsoft 365 cloud revenue. AWS represents Amazon’s cloud operating segment.

Servers.News reviewed the latest Amazon , Microsoft and Alphabet financial disclosures to distinguish reported revenue, infrastructure investment and the limits of cross-company comparisons.

Revenue for July 1–September 30, 2026, compared with the corresponding 2025 quarter. Microsoft labels this fiscal Q1 2027. Growth rates are calculated from the disclosed dollar figures.

Q3 2026 Cloud Revenue: The Verified Numbers

AWS reported the highest quarterly revenue of the three disclosed businesses. Azure continued to expand rapidly, while Google Cloud recorded the largest percentage and absolute dollar increase.

Source: company Q3 2026 earnings disclosures. Amounts are expressed in US dollars. Year-over-year growth and dollar increases are independently calculated from reported revenue.

Why Google Cloud’s 103% Growth Requires Context

Google Cloud’s quarterly revenue rose from $15.157 billion to $30.709 billion, an increase of approximately 102.6%.

That is a striking expansion in reported revenue, but the figure should not be interpreted as proof that conventional cloud-service consumption doubled organically.

Alphabet identified sales of its Tensor Processing Unit systems as a meaningful contributor to growth. TPU systems sold to customers have different revenue characteristics from recurring usage of hosted computing services.

Google Cloud also includes Workspace and enterprise cloud products, making its segment revenue broader than an infrastructure-only measure.

The Wiz Acquisition Adds Another Complication

Alphabet completed its acquisition of cloud security company Wiz on March 11, 2026, for approximately $29.5 billion.

According to Alphabet’s September 2026 Form 10-Q , Wiz’s financial results are included in the Google Cloud segment following the transaction.

The company has not disclosed sufficient standalone quarterly Wiz revenue to calculate an acquisition-adjusted organic Google Cloud growth rate.

Therefore, a meaningful comparison must acknowledge both TPU-system sales and the contribution of acquired businesses.

AWS: A $46 Billion Quarter With Strong Operating Profit

AWS generated $45.984 billion in quarterly revenue, increasing approximately 39.3% from the same period a year earlier.

Amazon described the quarter as AWS’s fastest growth in nearly five years. The revenue represents a simple annualized run rate of approximately $184 billion.

That annualized figure multiplies the quarter’s sales by four. It is not a forecast of full-year revenue.

AWS operating income reached $17.773 billion, up from $11.434 billion in Q3 2025.

The resulting operating margin was approximately 38.7%, demonstrating the earnings contribution of Amazon’s cloud division despite the company’s substantial infrastructure investments.

AWS accounted for approximately 22.8% of Amazon’s consolidated quarterly revenue but roughly 63.3% of its operating income.

This difference helps explain why cloud infrastructure is strategically important to Amazon even as the company spends heavily on new computing capacity.

Management continues to expect approximately $220 billion in company-wide cash capital expenditures during calendar 2026.

Although AI infrastructure is a major investment driver, the figure applies to Amazon as a whole. It is not a disclosed AWS-only capital budget or a GPU purchasing total.

Servers.News previously examined the implications of Amazon’s investment strategy in its analysis of AWS growth and the company’s capital expenditure plans .

Azure: New Transparency Changes the Comparison

Microsoft introduced a revised financial reporting structure in September 2026, reducing its business reporting to two operating segments: Agents and Infra, and Devices and Consumer.

Crucially, the company now separately discloses Azure revenue within its product and service reporting.

Azure generated $33.165 billion during the September quarter, compared with a recast $22.384 billion in the corresponding 2025 period.

That represents approximately 48.2% year-over-year growth, compared with Microsoft’s rounded reported figure of 48%.

Microsoft Cloud revenue, a much broader category, reached $63.9 billion. It includes other services and should not be substituted for Azure’s standalone revenue in a direct cloud-provider comparison.

Similarly, Microsoft’s Agents and Infra segment generated $76.8 billion, but that segment combines multiple businesses beyond Azure.

The new Azure disclosure makes revenue analysis more transparent. However, Microsoft does not separately disclose Azure operating income in a format directly comparable with AWS segment profit.

How Quickly Are the Three Businesses Growing?

Profitability: AWS and Google Cloud Are Not Identical Businesses

The comparison becomes more challenging when moving from revenue to profitability.

AWS and Google Cloud report operating income at the segment level. Microsoft does not publish an equivalent standalone Azure operating profit figure.

Operating margins calculated from reported operating income and revenue. Alphabet records some shared AI research and development expenses outside the Google Cloud segment, further limiting a normalized margin comparison.

Google Cloud operating income rose from $3.594 billion to $11.121 billion. AWS operating income increased approximately 55% year over year.

Both results indicate strong segment profitability. Neither provides a direct measure of GPU utilization, data-center efficiency or the cost of processing an AI workload.

The AI Infrastructure Spending Race

The earnings reports reveal capital investment on a remarkable scale.

Amazon maintains guidance for approximately $220 billion in company-wide cash capital expenditures in 2026. Alphabet expects between $195 billion and $205 billion for the year.

Microsoft reported approximately $45.2 billion in capital expenditures for its latest quarter, including finance-lease commitments. Its quarterly spending included extensive investment in CPUs, GPUs, memory, storage and data-center infrastructure.

These figures cannot be added together and described as a comparable cloud AI capital expenditure total without additional adjustments.

These figures deliberately show their different time periods and accounting definitions. They are not directly comparable annual AI spending totals.

Alphabet spent approximately $51.4 billion on capital expenditures during Q3. Management said most spending supported technical infrastructure and expected higher depreciation and energy-related operating costs.

Microsoft’s capital expenditure was below the company’s earlier quarterly expectations partly because some third-party computing capacity deliveries were delayed.

That illustrates an important distinction: financial willingness to invest does not guarantee that electrical connections, physical facilities and compute hardware will become available on schedule.

What Hyperscale Investment Actually Builds

Much of the debate focuses on how much the major cloud companies spend. The technical question is what those investments purchase and when that infrastructure becomes productive.

Simplified Servers.News infrastructure model. Actual cloud-provider architectures and financial recognition practices differ.

Capital spending on land, buildings and electrical infrastructure may support operations over many years.

Accelerators, storage and networking equipment have different replacement cycles and depreciation profiles.

Electricity and cooling create continuing operating costs once equipment enters service.

Consequently, a dollar of capital expenditure cannot be converted directly into cloud revenue, GPU hours or installed megawatts.

Electricity, Cooling and Networking Are the Real Constraints

Cloud growth is increasingly influenced by the availability of physical infrastructure.

Major AI facilities require grid connections capable of supporting large sustained electrical loads. High-density accelerator clusters demand cooling systems, appropriate power distribution and increasingly sophisticated networking.

Key constraints include:

• Grid capacity: suitable electrical connections and transmission infrastructure may take years to deliver.
• Cooling performance: high-density AI systems require thermal designs suited to their hardware power characteristics.
• Accelerator availability: equipment deliveries and cluster commissioning determine when purchased hardware becomes productive.
• Network efficiency: AI training performance depends on interconnect capacity, topology and communication overhead.
• Utilization: installed compute capacity is not equivalent to productive customer workload hours.

These issues are becoming more visible as electricity demand rises. In a recent investigation, Servers.News examined Norway’s scenarios for rapidly increasing data-center power consumption .

The fundamental point is that financial commitments cannot eliminate constraints in construction, electricity supply or hardware delivery.

Why Cloud Revenue Is Not the Same as AI Revenue

Cloud platforms serve many workloads beyond artificial intelligence.

Revenue may come from virtual machines, databases, networking, cybersecurity, storage, productivity applications and other managed services.

AI-related demand can influence all of these categories, but the companies do not publish standardized disclosures that isolate infrastructure revenue specifically from training and inference across all three businesses.

The same caution applies to long-term compute commitments.

Agreements to purchase future capacity are not equivalent to revenue already recognized or computing resources already installed.

That distinction is important when examining the scale of contractual demand. Servers.News previously reported on Anthropic’s disclosed long-term cloud computing obligations .

Servers.News Assessment: Three Different Growth Models

Amazon Web Services: AWS remains the largest of the three reported revenue lines, combining rapid sales growth with substantial operating profitability. Its key financial challenge is maintaining productive utilization and durable returns as investment increases.

Microsoft Azure: Azure’s new standalone revenue disclosure makes its expansion easier to evaluate. Microsoft’s results also illustrate how cloud demand can outrun the delivery schedules of supporting infrastructure.

Google Cloud: Google reports the strongest percentage growth and rapidly expanding operating income. But its revenue mix, TPU-system sales and inclusion of acquired Wiz operations limit comparisons with conventional cloud consumption.

None of these financial results establishes which provider offers the best performance, pricing, reliability or energy efficiency for a specific customer workload.

Those questions require workload-level benchmarking, transparent pricing and operational performance measurements.

Five Indicators to Watch in the Next Earnings Cycle

• AWS profitability: whether segment operating margins remain strong as depreciation and infrastructure investment expand.
• Azure disclosure: whether Microsoft’s standalone reporting continues to provide consistent comparisons over time.
• Google Cloud revenue composition: whether Alphabet discloses more detail about TPU-system sales and other components of growth.
• Capital efficiency: whether higher spending translates into installed, utilized and commercially productive compute capacity.
• Power and delivery constraints: whether grid connections, cooling capacity and hardware supply keep pace with customer demand.

The Bottom Line

The third-quarter 2026 earnings results show extraordinary expansion across the largest US cloud infrastructure providers.

AWS reported $45.984 billion in revenue, Azure generated $33.165 billion, and Google Cloud reached $30.709 billion.

But the most important conclusion is not a simple ranking.

Cloud financial results increasingly reflect a complex combination of computing services, hardware sales, software products, acquisitions and physical infrastructure investment.

As AI systems demand more electrical power, cooling and high-performance processors, the decisive question is whether providers can turn unprecedented capital spending into reliable, efficiently utilized computing capacity and sustainable earnings.

That will be a more meaningful measure of long-term AI infrastructure performance than headline revenue growth alone.