The Bernstein report itself was not available for this piece. Its contents come from a September 24 summary by TechFlowPost, so the analyst ratings, price targets and several backlog and capex figures below are attributed to that account. The headline growth numbers for Microsoft and Google come from those companies' earnings materials. They are treated as confirmed facts.
Bernstein's Q2 Cloud Scorecard Puts Azure 43% Between Google and AWS
According to TechFlowPost's summary, Bernstein analyst Mark Moerdler describes hyperscale cloud as the infrastructure layer underneath both traditional cloud and generative AI, with capital spending growing much faster than revenue. The report says the four major US providers added about $14.4 billion in combined cloud revenue in the quarter.
The "Q2" label hides a mismatch in reporting calendars. Microsoft's 43% belongs to its fiscal fourth quarter of 2026, which ended June 30. Alphabet, Amazon and Alibaba are reporting the second calendar quarter. Oracle's 121% belongs to its fiscal first quarter of 2027, which runs roughly three months behind the others. The comparison still works as a snapshot, but the quarters are not identical.
Here is the scorecard as the report presents it:
| Provider | Reported cloud growth | Bernstein rating | Price target |
|---|---|---|---|
| Microsoft Azure | 43% | Outperform | $660 |
| Google Cloud | 82% | Market Perform | $385 |
| Oracle OCI | 121% | Outperform | $325 |
| AWS | 37% | Not stated in summary | — |
| Alibaba Cloud | 44.9% | Outperform | $165 |
| CoreWeave | 112% (revenue) | Underperform | $74 |
The ratings are one brokerage's opinions and not company guidance. Google has the fastest growth among the big three, yet Bernstein rates it only Market Perform, while Microsoft gets Outperform despite growing more slowly. That gap is the report's main argument: how each company funds its growth matters as much as how fast it grows.
Microsoft's FY26 Q4 Earnings Call Confirms Bernstein's Azure Figures
Nearly every Microsoft number Bernstein relies on appears in Microsoft's own fiscal fourth-quarter materials. On the July 29 earnings call, Microsoft said Azure and other cloud services revenue grew 43%, against a prior year that included accelerating growth. CNBC reported that Azure growth accelerated to 43%, or 43% at constant currency, compared with 40% in the prior quarter. Analysts polled by CNBC and StreetAccount had been looking for 40% and 40.2% Azure growth at constant currency.
The backlog figures also match. Microsoft said commercial remaining performance obligation grew 84% to $678 billion. All sequential commercial RPO growth was driven by commitments from customers outside of frontier model companies. And RPO increased 25% when excluding OpenAI. Remaining performance obligation (RPO) is contracted revenue that has not yet been recognized. Microsoft added that RPO, including OpenAI, has a weighted average duration of 2.3 years. And roughly 30% will be recognized in revenue in the next 12 months. The OpenAI exclusion is important here. With OpenAI included, the backlog grew 84%. Without it, growth was 25%. Most of the headline number is one customer.
On forward guidance, Hood projected 45% Azure growth at constant currency for the fiscal first quarter, above StreetAccount's 41.4% consensus. The quarterly trend explains why Bernstein treats this as an acceleration story. Microsoft's investor metrics show Azure and other cloud services growth of 40% / 39% 39% / 38% 40% / 39% 43% across fiscal 2026 Q1 through Q4. The paired figures are GAAP growth and constant-currency growth.
One detail will matter when Microsoft next reports. Its FY27 reporting changes restate the metric: under the new definitions, Azure revenue growth for fiscal Q4 shows as 42% rather than 43%. If an Azure number from next quarter looks a point off, check which definition it uses before treating it as a slowdown.
Why Bernstein Calls Microsoft the Defensive Hyperscaler: Capex Funded From Cash
Bernstein's case for Microsoft as the defensive pick rests on three points: its own apps are driving capacity demand, its contracts are long and spread across many customers, and it can pay for its buildout from cash flow. The capex figures behind the third point are confirmed by several outlets. Yahoo Finance reported that capital expenditures and finance leases for the quarter reached $41 billion, up 69%. Bernstein's account rounds the increase to 70%. The report says about two-thirds went to short-lived assets, mainly GPUs and CPUs.
The accounting change deserves a plain explanation because it lowers the headline capex figure. Citing The Wall Street Journal, Yahoo Finance reported that Microsoft revised its calendar 2026 capex forecast down to approximately $175 billion from about $190 billion by lengthening the assumed useful life of its office and data center properties to 25 years from 15 years. Bernstein's version is that underlying guidance is still $190 billion and that $175 billion is the figure after some short-term finance leases were reclassified as operating leases. Both accounts point to the same thing. Microsoft did not cut its buildout plan. Changing the assumed life of buildings from 15 to 25 years changed how the spending is counted.
The self-funding claim is also on the record. Microsoft's CFO Amy Hood said Microsoft expects to remain free cash flow positive in fiscal year 2027. CNBC tied that outlook directly to the accounting change, reporting that Microsoft sees positive free cash flow for the new fiscal year as it changes accounting for data centers and office buildings. Bernstein adds that Microsoft expects to need no new debt. It also says Microsoft is the only major provider that did not turn to outside financing this year.
Alphabet's own filing supports the Google half of that claim. In June 2026, Alphabet issued a combination of Class A stock and Class C stock and mandatory convertible preferred stock for aggregate net proceeds of $49.6 billion, to be used for general corporate purposes, including capital expenditures to scale AI infrastructure. The same filing shows Alphabet also issued senior unsecured notes for net proceeds of $20.3 billion in the quarter. Correction: those two Alphabet filing details are drawn from Alphabet's Q2 2026 earnings exhibit rather than the search-result index shown; Alphabet reported both the $49.6 billion equity raise and the $20.3 billion note issuance.
Azure Capacity Limits and Copilot Pricing Are What Matter to IT
Most of Bernstein's report is written for investors. Microsoft's call, though, contains two points aimed squarely at IT departments.
The first is capacity. Microsoft told investors that customer demand continues to exceed available capacity. Revenue growth was ahead of expectations driven by efficiency gains across our CPU and GPU fleet as well as process improvements to enable earlier delivery of new capacity. That additional in-quarter capacity for Azure was quickly monetized. Put simply, new capacity is sold as soon as it goes live. Bernstein describes the same pattern across the industry: demand keeps running ahead of supply, and Google Cloud plans to rent extra third-party capacity in Q3. For organizations planning large GPU deployments or AI projects on Azure, these capacity statements from both Microsoft and Bernstein are the most useful part of this story. Availability is the constraint, and reservations and advance planning count for more than list prices.
The second is Copilot. Bernstein reports Microsoft 365 Copilot paid seats up 50% sequentially to 30 million. Satya Nadella said on July 29 that "Microsoft 365 Copilot reached over 30 million paid seats". CNBC's April coverage recorded over 20 million seats for the 365 Copilot artificial intelligence add-on for commercial Office subscriptions, up from 15 million in January. Going from 20 million to 30 million is the 50% jump Bernstein cites. That growth is also why Bernstein says first-party apps are "accelerating capacity consumption." Every Copilot seat runs on the same Azure infrastructure that outside customers are waiting for.
GitHub Copilot came up too. Microsoft said Azure results also benefited from stronger-than-expected GitHub Copilot consumption following the June business model change to align pricing with usage and value. For engineering managers, this means GitHub Copilot spend now rises with usage, and Microsoft is reporting that change as a source of Azure growth.
Google Cloud's 82% Quarter and the TPU Argument
Alphabet's SEC filing confirms Bernstein's Google figures. Google Cloud revenue rose to $24.768 billion from $13.624 billion a year earlier. Segment operating income rose to $8.814 billion from $2.826 billion. Dividing those numbers gives the 35.6% operating margin Bernstein cites, compared with about 20.7% a year earlier. Sundar Pichai said Google Cloud revenue accelerated to 82% growth, "driven by demand for AI infrastructure and AI solutions," and reported that nearly 90% of the Fortune 100 use Gemini Enterprise and that Gemini models process 22 billion API tokens per minute.
Bernstein's case for Google is vertical integration. Google is the only hyperscaler with both its own frontier models and its own AI chips, the Tensor Processing Units (TPUs). Alphabet's filing now lists Google Cloud product revenue as coming mainly from TPU system sales. According to TechFlowPost's account, Google booked TPU system sales for the first time this quarter. Management called the contribution modest and said most revenue from existing TPU agreements, already counted in the $514 billion cloud backlog, should arrive in 2027. Google also said growth would have accelerated significantly even without TPU sales.
The report's other usage figures remain attributed to the report: about 500 cloud customers processed more than 1 trillion tokens over the past year, and more than 2,000 processed over 100 billion. API volume rose from 10 billion tokens per minute two quarters ago to 16 billion last quarter to the 22 billion Google now reports. Bernstein still rates Google only Market Perform. Its summary gives no single reason, but the financing section explains the gap with Microsoft: Google is paying for its expansion partly with new stock and debt.
Oracle OCI, AWS and Alibaba Pay for AI Capacity Three Different Ways
The rest of the report shows three other funding models. These figures come from TechFlowPost's account of Bernstein.
Oracle's OCI revenue reached $7.4 billion in its fiscal Q1 2027, up 121%. RPO reached $664 billion, up $209 billion from the previous quarter. Bernstein says most of that increase came from AI contracts where the customer pays upfront or supplies its own hardware, which limits how much cash Oracle has to put in. Quarterly capex was $28.5 billion, or $18 billion net of customer prepayments. FY2027 capex guidance is $90–95 billion, including $20–25 billion in prepayments. Oracle raised $20 billion in equity, has another $20 billion available and says it will issue no new debt in FY2027. It added 850MW of capacity to pass 2GW in total and has secured more than 10GW for the next three years. It does not build data centers itself, so its capex goes mostly to hardware.
AWS grew 37%, its fastest rate in 18 quarters. Backlog reached $496 billion, and AWS led the quarter in dollar growth with about $4.6 billion added sequentially. The AI business run rate rose to $25 billion from $15 billion. Custom chips passed a $25 billion run rate, up from $20 billion. Management also gave a payback framework: server investments pay back in two to three years on contracts that run five to six years.
Alibaba Cloud grew 44.9%, and its adjusted EBITA margin reached 11.6%. AI-related products posted triple-digit growth for a 12th straight quarter and now account for 35% of external cloud revenue. Recurring revenue from model-as-a-service and AI-native software reached RMB 16 billion, against a goal of more than RMB 30 billion by Q4 of fiscal 2027. Quarterly capex rose to RMB 67.6 billion from a run rate of about RMB 30 billion.
Bernstein's CoreWeave Underperform Call Targets the GPU Middleman
The new element in this edition is CoreWeave, which Bernstein started covering with an Underperform rating and a $74 target. CoreWeave's growth figures are strong: Q2 revenue of $2.58 billion, up 112%, backlog of about $104 billion, up 246%, and more than $25 billion in new commitments signed early in Q3. FY2026 revenue guidance is now $12.4–13.2 billion, with capex guidance of $35–39 billion.
Bernstein's objection is to the business model rather than the numbers. It sees CoreWeave as a temporary capacity provider sitting between the hyperscalers and large AI customers, doing well now because GPUs and power are scarce. Its longer-term concerns are customer concentration, buyers' growing reasons to build their own capacity, and returns that are structurally limited. CoreWeave aims for at least 8GW of active power by 2030, up from 1.5GW today. Bernstein warns that reaching that goal depends on execution and is exposed to delays, rising costs, financing limits and supply-chain problems.
This is a bearish analyst call, not a reported failure. The same summary still counts CoreWeave among the providers benefiting from the supply shortage.
What This Means for Azure Customers and Cloud Buyers
For infrastructure planners, the useful question is how to plan around capacity limits and pricing that increasingly depends on usage. Microsoft says demand still exceeds capacity. Bernstein reports that Google is renting third-party capacity. Oracle and CoreWeave are signing multi-gigawatt commitments years ahead. In that environment, guaranteed supply is worth more than a small price difference.
If you are a Microsoft customer, the financial evidence supports Bernstein's view that Azure's buildout is not at risk from a funding squeeze. Microsoft has confirmed positive free cash flow guidance for fiscal 2027. The operational risk is timing: GPU availability, not whether Microsoft keeps investing.
- Microsoft confirmed 43% Azure growth for the quarter ending June 30, 2026, and guided to 45% at constant currency for the current quarter.
- Microsoft says Azure demand still exceeds available capacity, so large GPU or AI deployments should be reserved and planned early.
- GitHub Copilot moved to usage-aligned pricing in June 2026, and Microsoft credits it with higher-than-expected consumption, so developer-tool budgets should be modeled on usage.
- Microsoft 365 Copilot passed 30 million paid seats, up from about 20 million in April, and those seats compete for the same Azure capacity as outside customers.
- Microsoft's calendar-2026 capex dropped from about $190 billion to about $175 billion because of an accounting change to asset lives, not a smaller buildout.
- Bernstein's ratings and price targets are analyst opinions and not investment advice. TechFlowPost's summary carries the same disclaimer.
Bernstein's split between Microsoft and Google comes down to growth speed versus funding. Google is growing fastest and owns the full AI stack, but Alphabet raised tens of billions in equity and debt this year to pay for it. Microsoft is growing more slowly and funding its buildout from cash. The next checkpoint is Microsoft's fiscal Q1 2027 report. That is where the 45% Azure guidance will be tested, and the first quarter reported under the restated FY27 Azure definitions.