Cloud Market Share Statistics 2026
By Axis Intelligence Research
Co-authors: Elena Rodriguez & James Porter | Last updated: September 1, 2026 | License: CC BY 4.0
Five cloud providers reported dollar revenue for the quarter ended June 30, 2026, and together they booked $86.6 billion. AWS took 48.8% of that reported pool, Google Cloud 28.6%. Microsoft reported no Azure dollar figure at all — only a 43% growth rate and a fiscal-year statement that Azure passed $100 billion.
Quick Answer
According to Axis Intelligence Research, AWS generated $42.23 billion in the June 2026 quarter and Google Cloud $24.77 billion — a Reported Disclosure Share (RDS™) of 48.8% and 28.6% of the $86.6 billion pool of cloud revenue that providers actually reported in dollars. Microsoft is absent from that pool by its own choice: it discloses Azure growth (43%) and an annual threshold (over $100 billion in FY2026), never a quarterly dollar amount. Every widely quoted “Azure holds 21%” figure is therefore a third-party estimate, not a reported number.
Key Findings
- According to Axis Intelligence Research, AWS held a Reported Disclosure Share of 48.8% and Google Cloud 28.6% of the $86.6 billion in cloud revenue that providers reported in dollars for the quarter ended June 30, 2026.
- Google Cloud added $11.14 billion of incremental year-over-year quarterly revenue against AWS’s $11.36 billion — a gap of $215 million, or under 2%, despite AWS operating at 1.7 times Google Cloud’s revenue base.
- According to Axis Intelligence Research, Oracle carries a Backlog Coverage Multiple of 16.1x — sixteen years of contracted revenue at its current cloud run rate — against 5.2x for Google Cloud and 2.9x for AWS.
- Microsoft reported Azure and other cloud services revenue growth of 43% for the quarter ended June 30, 2026, and stated that Azure revenue passed $100 billion for the fiscal year; it published no quarterly Azure dollar figure, which is why Azure cannot enter a reported-revenue share table.
- Eurostat’s 2025 enterprise survey, published in February 2026, found that 52.7% of EU enterprises bought paid cloud services, but only 28.2% bought cloud computing power to run their own software — the gap between cloud adoption and cloud infrastructure adoption.
Who Has the Largest Cloud Market Share in 2026?
Start with what is verifiable. For the quarter ended June 30, 2026, Amazon reported AWS segment net sales of $42,232 million, up 36.7% from $30,873 million, with segment operating income of $16,621 million on a 39.4% margin. Alphabet reported Google Cloud revenue of $24,768 million, up 81.8% from $13,624 million, with segment operating income of $8,814 million. Both figures come from 8-K exhibits filed with the SEC.
Now the part almost every cloud market share article skips. Microsoft’s FY26 Q4 release discloses Azure and other cloud services revenue growth of 43% and Intelligent Cloud segment revenue of $39,306 million — but Intelligent Cloud also contains server products and enterprise services, and Microsoft has never broken Azure out as a standalone quarterly dollar line. Satya Nadella’s statement that “Azure revenue surpassed $100 billion for the first time” is a fiscal-year threshold, not a quarterly figure.
That asymmetry is the whole story of cloud market share measurement in 2026. Three of the four largest providers publish auditable dollars. One publishes a percentage and an annual milestone. Any table that puts all four in the same column has silently mixed reported revenue with modelled revenue — and readers almost never learn which cells are which.
Elena Rodriguez: The procurement version of this problem is familiar to anyone who has run a cloud RFP. You ask three vendors for a per-seat number and two give you a rate card while the third gives you a “typical customer sees” percentage. You do not average those. You put the rate cards in one column, flag the third, and negotiate accordingly. Market share analysis deserves the same discipline, and it rarely gets it.
Reported cloud revenue, quarter ended June 30, 2026
| Provider | Revenue reported | Prior-year period | Growth | Period end | Source |
|---|---|---|---|---|---|
| AWS | $42,232M | $30,873M | +36.7% | 2026-06-30 | Amazon Form 8-K, Ex. 99.1 |
| Google Cloud | $24,768M | $13,624M | +81.8% | 2026-06-30 | Alphabet Form 8-K, Ex. 99.1 |
| Oracle (total cloud) | $9,913M | $6,737M | +47% | 2026-05-31 | Oracle Q4 FY26 release |
| Alibaba AI Cloud & Compute | $7,139M | not disclosed in USD | +45% (RMB) | 2026-06-30 | Alibaba Form 6-K |
| CoreWeave | $2,575M | $1,212M | +112.5% | 2026-06-30 | CoreWeave Form 8-K |
| Microsoft Azure | not disclosed | not disclosed | +43% | 2026-06-30 | Microsoft FY26 Q4 release |
Source: company filings and earnings releases as listed; compiled by Axis Intelligence Research, retrieved September 1, 2026. Oracle’s quarter ends May 31, a three-week offset from the calendar quarter.
Reported Disclosure Share (RDS™): Ranking Providers by What They Actually Publish
Most cloud provider market share rankings blend reported revenue with analyst estimates and present a single clean percentage. Axis Intelligence Research built the Reported Disclosure Share to do the opposite: measure each provider’s share of the revenue that providers themselves put in writing, and leave the estimates out entirely.
RDS™ (Reported Disclosure Share) = provider’s reported cloud revenue for its most recent reported quarter ÷ the sum of all provider-reported cloud revenue for the same period.
Inputs: AWS $42,232M · Google Cloud $24,768M · Oracle cloud $9,913M · Alibaba AI Cloud and Compute Services $7,139M · CoreWeave $2,575M. Denominator: $86,627M.
Worked example: Google Cloud RDS = 24,768 ÷ 86,627 = 0.286, or 28.6%.
| Provider | Reported revenue | RDS™ | Annualized run rate |
|---|---|---|---|
| AWS | $42,232M | 48.8% | $168.9B |
| Google Cloud | $24,768M | 28.6% | $99.1B |
| Oracle (cloud) | $9,913M | 11.4% | $39.7B |
| Alibaba AI Cloud & Compute | $7,139M | 8.2% | $28.6B |
| CoreWeave | $2,575M | 3.0% | $10.3B |
Source: Axis Intelligence Research calculation from the filings listed above; snapshot as of September 1, 2026. Formula and inputs disclosed for reproduction.
Scope. RDS is not a claim about the total cloud market. It is a claim about the disclosed portion of it, and it is deliberately biased toward companies that report. Providers that publish no dollar figure — Microsoft, Tencent, Huawei, IBM’s cloud line, most sovereign and regional operators — are excluded rather than estimated. The number a reader gets is smaller in scope and larger in reliability, and it can be recomputed from six public documents in about ten minutes.
Adding Azure at its disclosed floor
Microsoft’s own FY2026 disclosure supports one arithmetic operation and no more. If Azure revenue passed $100 billion across the four quarters ended June 30, 2026, the average quarter was at least $25.0 billion. Because Azure accelerated through the year — 40% growth in the March quarter, 43% in the June quarter — the June quarter sat above that average. So $25.0 billion is a floor, not an estimate of the actual figure.
| Provider | Basis | Revenue | RDS™ (floor basis) |
|---|---|---|---|
| AWS | Reported | $42,232M | 37.8% |
| Microsoft Azure | Disclosed floor | $25,000M | 22.4% |
| Google Cloud | Reported | $24,768M | 22.2% |
| Oracle (cloud) | Reported | $9,913M | 8.9% |
| Alibaba AI Cloud & Compute | Reported | $7,139M | 6.4% |
| CoreWeave | Reported | $2,575M | 2.3% |
Source: Axis Intelligence Research, September 1, 2026. Azure row derived arithmetically from Microsoft’s stated FY2026 threshold ($100B ÷ 4) and labelled as a floor.
On the floor basis, the top three providers hold 82.4% of reported cloud revenue. That figure is higher than the 60-something percent that circulates in most coverage, and the reason is definitional rather than substantive: this denominator counts only companies that report, while broader trackers include hundreds of regional providers that never file a cloud line. Both numbers can be correct. Only one of them can be checked.
Which Cloud Provider Is Gaining Fastest in 2026?
Percentage growth rates flatter small bases, so read the dollars instead. AWS added $11,359 million of year-over-year quarterly revenue. Google Cloud added $11,144 million. The gap is $215 million on bases that differ by a factor of 1.7.
That is the most consequential single comparison in cloud infrastructure right now, and it is invisible in any table that only shows growth percentages, where 36.7% and 81.8% look like different sports.
| Provider | Incremental YoY quarterly revenue | Growth rate | Revenue base |
|---|---|---|---|
| AWS | $11,359M | +36.7% | $42,232M |
| Google Cloud | $11,144M | +81.8% | $24,768M |
| Oracle (cloud) | $3,176M | +47% | $9,913M |
| CoreWeave | $1,363M | +112.5% | $2,575M |
Source: Axis Intelligence Research calculation from company filings; retrieved September 1, 2026.
Elena Rodriguez: Watch what happens to the migration conversation if this holds for two more quarters. Enterprise buyers have spent a decade treating AWS as the default and everything else as the alternative that needs justifying. Dollar-for-dollar parity in net new spend does not flip that overnight — data gravity and egress economics are stubborn — but it changes who has to justify what in the room. The question I would put to any vendor pitching me in Q4: how much of your incremental revenue is a handful of frontier-model contracts, and how much is the mid-market renewing and expanding?
Backlog Coverage Multiple (BCM™): How Much Cloud Revenue Is Already Contracted
Revenue tells you about the quarter that ended. Contracted backlog tells you about the quarters that have not started. Axis Intelligence Research built the Backlog Coverage Multiple to make those disclosures comparable.
BCM™ (Backlog Coverage Multiple) = disclosed backlog or remaining performance obligation ÷ (most recent quarterly cloud revenue × 4). The result reads as years of current-run-rate revenue already under contract.
Worked example: Google Cloud BCM = 514,000 ÷ (24,768 × 4) = 514,000 ÷ 99,072 = 5.19x.
| Provider | Disclosed backlog | Annualized revenue | BCM™ | Backlog scope |
|---|---|---|---|---|
| Oracle | $638B | $39.7B | 16.09x | Company-wide RPO |
| CoreWeave | $104B | $10.3B | 10.10x | Company-wide revenue backlog |
| Google Cloud | $514B | $99.1B | 5.19x | Google Cloud only |
| AWS | $496B | $168.9B | 2.94x | AWS only |
| Microsoft | $678B | $237.2B | 2.86x | All commercial, not Azure alone |
Sources: Oracle Q4 FY2026 release ($638B RPO); CoreWeave Q2 2026 release ($104B); Alphabet Q2 2026 earnings call ($514B); Amazon Q2 2026 earnings call, July 30, 2026 ($496B); Microsoft FY26 Q4 release ($678B commercial RPO, up 84%). Axis Intelligence Research calculation.
Two scope mismatches sit in that table and neither can be engineered away. Microsoft’s $678 billion is commercial remaining performance obligation across Microsoft 365, Dynamics and Azure together, so its 2.86x is not an Azure figure and should not be read as one. Oracle’s RPO is company-wide, and Oracle disclosed that most of its recent increase came from large AI contracts where customers prepaid for GPUs or supplied the GPUs themselves — $75 billion of prepaid and customer-supplied hardware. A backlog with hardware prepayment inside it is not the same instrument as a consumption commitment.
Axis Intelligence Research computes and publishes both anyway, with the scope stated per row, because the alternative — averaging them into a single “hyperscaler backlog” number — would produce a figure that means nothing. Documented refusal to merge incompatible disclosures is part of the method, not a gap in it.
What the ratios do support is a ranking of forward visibility. Oracle and CoreWeave have sold years of capacity they have not yet built. AWS, at 2.94x, has the largest backlog in absolute dollars and the shortest coverage relative to its revenue — which is what maturity looks like on this metric, not weakness.
Cloud Capex Efficiency (CCE™): What Each Capex Dollar Is Currently Returning
The 2026 cloud buildout is the largest private infrastructure program in corporate history, and it is running ahead of the revenue it supports. Amazon’s trailing-twelve-month purchases of property and equipment reached $173.0 billion. Alphabet’s trailing-twelve-month capital expenditure hit $132.4 billion, pushing quarterly free cash flow to negative $5.9 billion. Microsoft spent $115.9 billion on property and equipment in FY2026. Oracle’s fiscal-year capital expenditure was $55.7 billion against negative $23.7 billion free cash flow.
CCE™ (Cloud Capex Efficiency) = incremental year-over-year cloud revenue in the quarter ÷ capital expenditure in the same quarter, expressed in cents of new revenue per capex dollar.
Worked example: AWS CCE = 11,359 ÷ 54,208 = 0.210, or 21.0 cents.
| Provider | Incremental quarterly revenue | Quarterly capex | CCE™ |
|---|---|---|---|
| Oracle | $3,176M | $7,413M | 42.8¢ |
| Microsoft (Intelligent Cloud) | $9,428M | $35,802M | 26.3¢ |
| Google Cloud | $11,144M | $44,924M | 24.8¢ |
| CoreWeave | $1,363M | $6,422M | 21.2¢ |
| AWS | $11,359M | $54,208M | 21.0¢ |
Source: Axis Intelligence Research calculation from company cash flow statements and segment tables, retrieved September 1, 2026. Oracle quarterly capex derived from its trailing-four-quarter disclosure ($55,663M less $48,250M).
Definitions and method. Capital expenditure at Amazon, Alphabet and Microsoft is company-wide, not cloud-only, and all three have said the overwhelming majority of current spend is AI and cloud infrastructure. CCE therefore understates cloud-only efficiency at those three by an amount none of them discloses. It is a comparative ratio across a consistent definition, not an accounting return. Capacity built this quarter also earns revenue in later quarters, so a low reading during a buildout phase is expected rather than alarming — the value of the metric is watching the same number across quarters. Oracle’s high reading reflects a smaller revenue base and a capex quarter that fell below its own recent peak.
Amazon raised 2026 capital expenditure guidance to approximately $220 billion on its July 30 earnings call, citing higher memory costs; Alphabet raised its 2026 range to $195–205 billion. Neither company committed that capital speculatively — the backlog figures above are the demand-side counterpart. Whether the returns land is a 2027 and 2028 question, and CCE is where it will show up first. Our AI capex tracker follows the announcements as they land, and the AI data center tracker follows the physical builds.
How Fast Is Enterprise Cloud Adoption Actually Growing?
Vendor revenue measures spend by the customers who already buy. It says nothing about how broadly cloud has penetrated the enterprise base, and on that question the best source is not a vendor at all.
Eurostat’s 2025 enterprise ICT survey, published February 2026, found 52.7% of EU enterprises with ten or more employees bought paid cloud services, up 7.4 percentage points from 2023 and up from 17.8% in 2014. The detail underneath that headline is where it gets interesting.
| Measure, EU enterprises, 2025 | Share | Basis |
|---|---|---|
| Bought any paid cloud service | 52.7% | All enterprises, 10+ employees |
| Large enterprises | 84.7% | All large enterprises |
| Medium enterprises | 66.8% | All medium enterprises |
| Small enterprises | 49.3% | All small enterprises |
| Bought cloud computing power to run own software | 28.2% | All enterprises |
| Bought application development platforms | 26.1% | All enterprises |
| Bought at least one IaaS service | 77.3% | Of cloud-buying enterprises |
| Highest national rate: Finland | 79.2% | All enterprises |
| Lowest national rate: Bulgaria | 17.8% | All enterprises |
Source: Eurostat, Cloud computing statistics on use by enterprises (dataset isoc_cicce_use), 2025 survey published 2026.
Half of EU enterprises buy cloud. Just over a quarter buy the compute layer that the hyperscaler revenue tables are actually measuring. The rest are buying email, office software and file storage — the SaaS surface, delivered by cloud, which never appears in an infrastructure market share table. When a market forecast quotes cloud adoption at “over half of enterprises” and then applies that to IaaS spending, this is the number it has skipped.
Elena Rodriguez: The 61-point spread between Finland and Bulgaria is the number I would put in front of anyone modelling European cloud demand. It is not a technology gap. It is a broadband, skills and firm-size gap, and it moves at the speed of national industrial policy rather than product releases. The IaaS growth in Europe over the next three years comes disproportionately from the medium-sized enterprise band in the countries currently sitting under 40%.
What Is Driving Cloud Growth in 2026?
AI workloads, and the providers now say so explicitly rather than implying it. Amazon disclosed that its AI business and its chips business each exceeded a $25 billion annualized run rate, both growing at triple-digit rates. Alphabet attributed Google Cloud’s acceleration to enterprise AI solutions and AI infrastructure, and disclosed that Gemini models process 22 billion API tokens per minute. Oracle attributed its RPO surge to AI training and inference demand. Alibaba reported that AI-related product revenue grew at triple-digit rates for a twelfth consecutive quarter.
The composition shift matters for anyone forecasting from these numbers. A cloud business whose growth comes from thousands of enterprises expanding steadily behaves differently from one whose growth comes from a handful of AI labs signing multi-gigawatt commitments. The second is larger and faster and concentrates counterparty risk in a way the first never did. CoreWeave is the clearest case: $104 billion of backlog against $2.6 billion of quarterly revenue, 1.5 GW of active power and roughly 3.7 GW contracted, and a $626 million quarterly net loss driven by $640 million of net interest expense.
Capacity, not demand, is the current constraint. Amazon said explicitly it will not have enough capacity to meet 2026 demand even at $220 billion of capex. Alphabet said it will rent third-party compute to meet its backlog. That is a supply-limited market, and supply-limited markets do not reallocate share quickly — customers take the capacity that exists. When we see outages ripple through that constrained capacity, they land in our cloud outage tracker; the power side of the constraint shows up in the data center electricity demand tracker.
Methodology
Collection. Every figure in this analysis comes from a document fetched and read on September 1, 2026: SEC Form 8-K exhibits (Amazon, Alphabet, CoreWeave), a Form 6-K exhibit (Alibaba), investor-relations earnings releases (Microsoft, Oracle), company earnings-call disclosures where a figure was stated only verbally (AWS backlog, Google Cloud backlog, capex guidance), and Eurostat’s published enterprise survey. No figure is carried over from prior editions of this page, and no figure is sourced from a third-party market tracker.
Period alignment. Four providers report a quarter ended June 30, 2026. Oracle’s fourth fiscal quarter ended May 31, 2026, and Microsoft’s fourth fiscal quarter ended June 30, 2026. Where periods differ, the period end is stated in the row.
Currency. Alibaba’s US dollar figure is the company’s own translation, published in its results announcement. Its growth rate is computed by the company in RMB (RMB48,437M against RMB33,418M, +45%) to avoid an FX artefact; Axis Intelligence Research reproduces the RMB-based growth rather than deriving a USD growth rate from a prior-period figure Alibaba did not publish in dollars.
Original metrics. RDS™, BCM™ and CCE™ are Axis Intelligence Research constructions. Each formula, its inputs and a worked example are disclosed inline above so that any reader can recompute the readings from the same public filings. All three are snapshot readings dated September 1, 2026 and carry no implied history: this is their baseline publication.
What this reading does not capture. Provider-reported cloud segments are not identically defined. AWS is an infrastructure and platform segment. Google Cloud includes Workspace and, from 2026, TPU system sales. Oracle’s cloud line includes SaaS applications, so the OCI-only figure ($5,787M) is stated separately for readers who want infrastructure alone. Alibaba’s segment was restructured during the quarter to combine Cloud Intelligence Group with T-Head, which affects comparability against its own prior disclosures. Private cloud, hybrid deployments, managed hosting and every non-reporting provider sit outside all figures on this page.
About This Dataset
The accompanying CSV contains every number in this article as an individual row with its value, unit, as-of date, source organisation, source document, source URL, retrieval date, and flags for whether the figure is primary-sourced and whether Axis Intelligence Research calculated it. Rows carrying an original metric include the method note pointing to the formula disclosed above.
License: CC BY 4.0. Free to use, redistribute and build on with attribution to Axis Intelligence Research.
Cite this research:
- APA: Axis Intelligence Research & Rodriguez, E. (2026). Cloud market share statistics 2026: What the filings actually show. Axis Intelligence. https://axis-intelligence.com/cloud-market-share-statistics/
- MLA: Axis Intelligence Research and Elena Rodriguez. “Cloud Market Share Statistics 2026: What the Filings Actually Show.” Axis Intelligence, 1 Sept. 2026, axis-intelligence.com/cloud-market-share-statistics/.
- Chicago: Axis Intelligence Research and Elena Rodriguez. “Cloud Market Share Statistics 2026: What the Filings Actually Show.” Axis Intelligence. Last modified September 1, 2026. https://axis-intelligence.com/cloud-market-share-statistics/.
Frequently Asked Questions
Why do cloud market share figures differ so much between sources?
Because the denominators differ and the disclosure quality differs. A tracker that includes hundreds of regional providers and estimates Azure produces a smaller share for AWS than a table built only from reported dollars. According to Axis Intelligence Research, AWS holds 48.8% of provider-reported cloud revenue for the June 2026 quarter, and 37.8% when Azure is included at the $25.0 billion floor implied by Microsoft’s own annual disclosure. Neither figure is wrong; they answer different questions, and a share figure without its denominator stated is not usable.
Does Microsoft disclose Azure revenue in dollars?
No. Microsoft reports Azure and other cloud services as a growth percentage — 43% for the quarter ended June 30, 2026 — inside the Intelligent Cloud segment, which also contains server products and enterprise services. The only dollar disclosure is annual and threshold-shaped: Azure revenue passed $100 billion in fiscal 2026. Any quarterly Azure dollar figure in circulation is a model output, and the honest arithmetic floor derived from Microsoft’s own statement is $25.0 billion per quarter.
How do you compare providers whose cloud segments include different things?
You state the composition in the same table as the number. Google Cloud includes Workspace and TPU system sales; Oracle’s cloud line includes SaaS applications; AWS is infrastructure and platform. Axis Intelligence Research publishes Oracle’s OCI-only figure ($5,787 million) beside its total cloud figure ($9,913 million) so readers can pick the definition that matches their question, rather than reconciling them into a single number that matches neither.
What does a cloud backlog figure actually commit a customer to?
Less than the headline implies, and the details vary by provider. Google Cloud’s $514 billion is contracted revenue Alphabet expects to recognise as customers consume, with just over half expected within 24 months. Oracle’s $638 billion includes large AI contracts where customers prepaid for GPUs or supplied them — $75 billion of that hardware component — which is a materially different instrument from a consumption commitment. Contracts can also be renegotiated. Read backlog as contracted intent with a stated conversion window, not as booked revenue.
Is the cloud market becoming more concentrated or less?
More concentrated among reporting providers, on the current evidence. AWS, Azure at its disclosed floor, and Google Cloud together account for 82.4% of provider-reported cloud revenue for the June 2026 quarter. The counterweight is real but early: CoreWeave, at 3.0% of the reported pool, grew 112.5% and holds ten years of contracted coverage, and Oracle’s cloud infrastructure line grew 93%. Concentration among incumbents and rapid entry at the edge are happening simultaneously.
Which provider is winning the AI cloud workload?
On disclosed dollars, no single provider. AWS added $11.36 billion of incremental quarterly revenue and Google Cloud $11.14 billion — a $215 million gap. Amazon disclosed an AI business above a $25 billion run rate; Alphabet attributed its acceleration to enterprise AI infrastructure; Oracle attributed a $638 billion RPO to AI training and inference. The workload is being split across providers rather than consolidating, which is consistent with a capacity-constrained market where buyers take what is available.
How much of the enterprise base actually buys cloud infrastructure?
Far less than headline adoption suggests. Eurostat’s 2025 survey found 52.7% of EU enterprises bought paid cloud services, but only 28.2% bought cloud computing power to run their own software and 26.1% bought application development platforms. The majority of cloud adoption at the enterprise-count level is email, office software and file storage — SaaS delivered over cloud, which does not appear in infrastructure revenue tables.
Should cloud capex guidance change how I read market share?
It should change your time horizon. Capacity built in 2026 earns revenue in 2027 and 2028, so today’s share figures reflect capacity decisions made eighteen months ago. Amazon guided to roughly $220 billion of 2026 capital expenditure and Alphabet to $195–205 billion, both citing demand exceeding supply. According to Axis Intelligence Research, current Cloud Capex Efficiency runs between 21 and 43 cents of incremental quarterly revenue per capex dollar across the reporting providers — the ratio to watch as those builds come online.
Where can I get the underlying data?
Every figure on this page ships as a CSV row with its source URL and retrieval date, licensed CC BY 4.0. Recomputing RDS™, BCM™ and CCE™ from the original filings requires six public documents and the formulas disclosed above.