AWS vs Azure vs Google Cloud: The Cloud Wars

AWS vs Azure vs Google Cloud market dominance in cloud infrastructure is not a race anymore. It is an oligopoly, and the three players are no longer competing on the number everyone quotes.

Three companies now hold roughly 63% of the global cloud infrastructure market: AWS at 28%, Microsoft Azure at 21%, and Google Cloud at 14% (Synergy Research Group, Q1 2026). That headline share looks stable. Underneath it, the growth ranking is being reshuffled by the AI build-out, and the real contest has moved from price to unit economics and committed backlog.

The short version: AWS leads on share and disclosed operating margin, Azure leads on enterprise distribution and rides Microsoft’s AI bundle, and Google Cloud is the fastest-growing challenger with backlog that nearly doubled in a single quarter. Growth now runs in reverse order of size, Google Cloud +63%, Azure +40%, AWS +20% (Amazon Form 10-K FY2025; Microsoft Form 8-K and Alphabet Form 8-K, Q1 2026).

This piece reads the three-way race through the filings, not the keynote stage. Every number ties to a specific SEC filing or named market-share report and its period. The framing is analytical: how to think about the cloud oligopoly’s structure, not what to do about any stock.

Key takeaways

• The cloud is a three-firm oligopoly: AWS 28%, Azure 21%, Google Cloud 14%, about 63% of the global infrastructure market combined (Synergy Research Group, Q1 2026).
• Growth runs inverse to size: Google Cloud +63% YoY, Azure +40% YoY, AWS +20% YoY, so the smaller players are gaining share points even though AWS stays largest (Amazon Form 10-K FY2025; Microsoft Form 8-K and Alphabet Form 8-K, Q1 2026).
• AWS holds the disclosed margin lead: $45.6B operating income on $128.7B revenue in FY2025, roughly a 35.4% operating margin (Amazon Form 10-K, FY2025).
• Azure’s revenue is not separately disclosed: Microsoft reports Intelligent Cloud at $30.9B (+28%) and Microsoft Cloud at $49.1B (+26%) instead (Microsoft Form 8-K, Q1 FY2026).
• Google Cloud’s backlog reached $462B in Q1 2026, nearly double the prior quarter, with more than half expected to convert within 24 months (Alphabet Form 8-K, Q1 2026).
• The AI capex arms race is compressing margins across all three even as revenue accelerates, which is why this is a race on unit economics now, not headline price.

AWS vs Azure vs Google Cloud: the cloud infrastructure market by the numbers

The first thing to fix is the shape of the market. Cloud infrastructure is not fragmented. Three firms hold about 63% of it, and the rest of the field, Alibaba, Oracle, IBM, Tencent, and a long tail, splits the remainder (Synergy Research Group, Q1 2026).

That concentration is the defining fact. When three vendors control nearly two-thirds of a market this large, the competitive question stops being “who wins customers” and becomes “how do three entrenched players grow against each other without a price war that destroys everyone’s margin.”

Here is the market-share picture from the most-cited infrastructure tracker, alongside each provider’s most recent disclosed growth.

Provider	Market share (Q1 2026)	Most recent growth	Source for growth
AWS	28%	+20% YoY (FY2025)	Amazon Form 10-K, FY2025
Microsoft Azure	21%	+40% YoY (Q1 FY2026)	Microsoft Form 8-K, Q1 FY2026
Google Cloud	14%	+63% YoY (Q1 2026)	Alphabet Form 8-K, Q1 2026

Market share: Synergy Research Group, Q1 2026 (IaaS + PaaS), published April 29, 2026. Growth rates are each company’s own most recent disclosure; the periods differ, see the methodology note below.

Read the table in two passes. By size, the order is AWS, Azure, Google. By growth, the order reverses completely. That inversion is the entire story of the current cloud wars: the leader is growing slowest, and the smallest of the three is growing fastest.

A caution on the comparison. The growth rates come from different fiscal periods and different reporting structures, so they are directionally comparable, not line-item identical. The methodology note at the end spells out exactly what is and is not apples-to-apples here.

Methodology: how these growth numbers compare

• Inputs: AWS growth from Amazon’s FY2025 10-K (full year). Azure growth from Microsoft’s Q1 FY2026 8-K (a single quarter, Azure growth rate only). Google Cloud growth from Alphabet’s Q1 2026 8-K (a single quarter). Market share from Synergy Research, Q1 2026.
• Assumptions: that each company’s disclosed growth rate is representative of its current trajectory, and that Synergy’s IaaS + PaaS definition is a fair proxy for “cloud infrastructure.”
• Sensitivity: quarterly figures are noisier than annual ones. A single quarter of Azure or Google Cloud growth can overstate or understate the run-rate. AWS’s annual figure is the most stable input.
• What this misses: the periods are not aligned (one annual, two quarterly), and Synergy’s category excludes SaaS and managed databases. Treat the ranking as robust and the exact percentages as period-dependent.

How is the cloud market structured as an oligopoly?

It is structured as a three-firm oligopoly where roughly 63% of cloud infrastructure spend flows to AWS, Azure, and Google Cloud, and no single firm has the share to dictate terms to the other two (Synergy Research Group, Q1 2026). Each holds a defensible base, so competition runs through growth rate and contract commitment rather than open price war.

That structure has consequences. In a true oligopoly, the players are large enough that a price war would be mutually destructive, so they compete on adjacent axes: switching cost, bundled software, AI capability, and the size of pre-committed enterprise contracts. The pricing-power side of that dynamic is the through-line in AWS margin pressure and the cloud reset, which reads the same structure from the leader’s margin angle.

The concentration also reshapes who captures value up the stack. When three firms own the infrastructure layer, every model lab, SaaS vendor, and AI startup pays rent to one of them, the dependency mapped in the AI infrastructure market map.

AWS: scale and margin dominance, but growth is slowing

AWS is still the largest and most profitable of the three by disclosed numbers, and it is growing the slowest. In FY2025 AWS posted $128.7B of revenue, up 20% year over year, and $45.6B of operating income, which implies roughly a 35.4% operating margin (Amazon Form 10-K, FY2025).

That margin is the headline AWS advantage. No other hyperscaler discloses a cleanly comparable cloud operating margin, and AWS’s is structurally high because of two things: a multi-year head start in building out infrastructure, and scale economics on a base of long-lived enterprise workloads.

But the 20% growth rate is the tension. AWS is compounding off the largest base, so a 20% increase is a larger absolute dollar gain than Google Cloud’s 63% on a smaller base. The trap is reading the growth percentage as a verdict on competitiveness. On absolute revenue added, AWS is still adding more than either rival.

The strategic question for AWS is whether its installed base of existing workloads keeps growing fast enough to offset the fact that a disproportionate share of new AI infrastructure decisions are landing on Azure and Google Cloud. That is the durability question, and it is exactly the kind of “the lead is real but the marginal customer is shifting” dynamic that decides oligopoly contests.

Azure: enterprise lock-in without standalone revenue transparency

Azure is the hardest of the three to measure, by design. Microsoft does not disclose Azure revenue as a standalone dollar figure. Instead it reports two broader buckets and an Azure growth rate.

In Q1 FY2026, Microsoft reported the Intelligent Cloud segment, which includes Azure plus server products and other services, at $30.9B, up 28% year over year (Microsoft Form 8-K, Q1 FY2026). It separately reported Microsoft Cloud, a wider commercial measure spanning multiple units, at $49.1B, up 26% (Microsoft Form 8-K, Q1 FY2026). And it disclosed that “Azure and other cloud services” grew 40% year over year, 39% in constant currency (Microsoft Form 8-K, Q1 FY2026).

So the analyst is left with a growth rate and two overlapping aggregates, but no clean Azure revenue line. This opacity is a feature, not an accident. It blunts direct comparison and lets Microsoft frame the story around the bundle.

And the bundle is the real Azure weapon. Microsoft sells Azure into an enterprise base already locked into Windows, Office, Active Directory, and increasingly Copilot, so the cloud decision rides on top of relationships that already exist. The mechanics of that pull-through are the subject of Microsoft Copilot and enterprise lock-in. Azure does not have to win on raw price; it wins on “you are already here.”

That distribution advantage is why Azure can grow 40% while disclosing the least. The growth is real, but its source is partly structural lock-in rather than head-to-head infrastructure wins, and the filings do not let you separate the two.

Google Cloud: the fastest challenger, and why the backlog matters more than revenue

Google Cloud is the growth story, and the most interesting number is not its revenue. In Q1 2026, Google Cloud posted $20.0B of revenue, up 63% year over year, and $6.6B of operating income (Alphabet Form 8-K, Q1 2026). That is the fastest growth of the three by a wide margin.

But the figure that reshapes the competitive read is the backlog. Google Cloud’s backlog reached $462B in Q1 2026, nearly double the prior quarter’s roughly $240B, and Alphabet disclosed that more than half is expected to convert to revenue within 24 months (Alphabet Form 8-K, Q1 2026).

Backlog is contractually committed future cloud spend. A backlog that nearly doubles in one quarter, with more than half landing inside two years, implies a large block of revenue is already signed rather than competed for. That is a structural shift away from spot purchasing toward locked-in, multi-year commitment.

The backlog read: why committed revenue changes the contest

• Reduced uncertainty. A backlog this size means a meaningful chunk of Google Cloud’s near-term growth is contracted, not forecast. That lowers the risk that a single quarter’s growth rate is a head-fake.
• Evidence of consolidation. Enterprises do not commit hundreds of billions in future spend casually. The jump suggests large customers are standardizing significant AI and data workloads on Google Cloud.
• A different competitive signal than revenue. Revenue tells you what was billed last quarter. Backlog tells you what has already been promised. In an oligopoly, the promise is the leading indicator.

The caution: backlog conversion is not guaranteed at the disclosed pace, contracts can be renegotiated, ramped slowly, or include flexible terms. The figure is a strong signal, not a settled outcome. Google’s broader strategy of using owned distribution and AI to lock in demand is the wider argument in Google’s AI strategy as a distribution war.

The Cloud Three-Way Scorecard

The cleanest way to hold the three providers side by side is to score them on the axes that actually decide an oligopoly: position, growth, disclosed margin, AI posture, and the single thing each one wins on. Call it the Cloud Three-Way Scorecard, a citeable framework for comparing AWS, Azure, and Google Cloud on the dimensions that matter rather than the headline revenue alone.

The Scorecard is the asset to cite when someone reduces the cloud wars to “who’s biggest.” Biggest is one row. The race is decided across all five.

Axis	AWS	Microsoft Azure	Google Cloud
Market share (Q1 2026)	28%	21%	14%
Disclosed growth	+20% YoY (FY2025)	+40% YoY (Q1 FY2026)	+63% YoY (Q1 2026)
Revenue transparency	Reported: $128.7B (FY2025)	Not standalone; bundled in Intelligent Cloud $30.9B	Reported: $20.0B (Q1 2026)
Disclosed operating margin	~35.4% (FY2025, implied)	Not broken out	~33% annualized (Q1 2026, one quarter)
AI position	Largest installed base; broad model and chip lineup	Bundled with Copilot and enterprise stack	Fastest growth; record backlog; in-house silicon
What it wins on	Scale, margin, breadth of services	Enterprise distribution and lock-in	Growth rate and committed backlog

Sources: market share, Synergy Research Group, Q1 2026; AWS figures, Amazon Form 10-K, FY2025; Azure/Microsoft figures, Microsoft Form 8-K, Q1 FY2026; Google Cloud figures, Alphabet Form 8-K, Q1 2026. Margin for AWS and Google Cloud is implied from disclosed operating income and revenue; Microsoft does not break out a comparable cloud margin. Periods differ across providers, see the methodology note above.

Read the Scorecard across rows and the oligopoly’s internal logic appears. AWS wins the position and margin rows. Azure wins distribution but loses transparency. Google Cloud loses on size but wins the two forward-looking rows, growth and backlog. No provider sweeps the card, which is precisely why none can dictate to the other two.

The AI build-out is reshuffling margins and growth

The reason the growth ranking inverted is the AI capex arms race. All three providers are pouring capital into AI infrastructure, GPUs, custom silicon, data centers, and power, and that spend is hitting their cost structures at the same time enterprises are consolidating AI workloads onto cloud platforms.

Two effects run in parallel. On the revenue side, AI workloads are driving the fast growth at Azure (+40%) and Google Cloud (+63%) as enterprises move new AI projects onto whichever platform offers the best models, tooling, and committed pricing (Microsoft Form 8-K and Alphabet Form 8-K, Q1 2026). On the cost side, the capex required to serve those workloads is enormous, and it compresses margin even as revenue climbs.

That dual effect, accelerating revenue alongside compressing margin, is the defining feature of the current cloud cycle. A provider can post a record growth rate and a record capex bill in the same quarter, and the second can quietly erode the unit economics that made the first worth chasing. The scale and structure of that spending is the subject of the AI capex arms race.

The competitive consequence is the reshuffle. AWS’s lead was built in an era when cloud meant general compute and storage, where its head start and scale dominated. The AI era resets some of that advantage because new AI infrastructure decisions are fresh decisions, and a meaningful share of them are landing on Azure and Google Cloud rather than defaulting to the incumbent. AWS keeps the largest base; it no longer owns the marginal AI workload by default.

Why are AWS, Azure, and Google Cloud no longer competing on price?

Because they form an oligopoly where an open price war would be mutually destructive, so the three compete on switching cost, bundled capability, and committed-spend contracts instead. Google Cloud’s $462B backlog is the clearest evidence: the contest has moved to locking in multi-year commitments, not winning the cheapest spot price (Alphabet Form 8-K, Q1 2026).

Cloud used to be a price story. Providers cut per-unit compute and storage prices in lockstep, and customers chased the cheapest credible option. That phase is largely over among the big three.

The shift is structural. Once general compute and storage commoditized, matching a rival’s list price gained nothing and gave up margin, so the three stopped racing each other to zero and started competing on the dimensions that raise switching cost.

This matters for anyone buying cloud. The lever is no longer “negotiate a lower list price.” It is “negotiate committed-spend discounts, portability terms, and exit costs,” because the providers are competing for your commitment, not your spot purchase. The vendor that captures a large committed backlog has converted a price-sensitive buyer into a locked-in one, which is worth far more than a one-quarter discount.

The deeper point: the absence of a price war is itself a sign of the oligopoly maturing. Three firms with defensible bases do not race each other to zero. They compete on the dimensions that raise switching cost, which is exactly what the Scorecard’s “what it wins on” row captures.

The operator’s playbook: what this means for customers, startups, and investors

The three-way structure changes what a buyer, a founder, and an analyst should actually do. The headline ranking is the least useful part of it. Here is the playbook.

• Negotiate on commitment, not list price. The providers compete on committed-spend contracts now, so the leverage is in multi-year commitment discounts, ramp flexibility, and portability terms, not in haggling over per-hour compute. Bring a credible multi-cloud alternative to the table even if you intend to consolidate.
• Price portability before you sign. The backlog dynamic exists because committed spend raises switching cost. Before locking into a large commitment, model the cost of leaving: data egress, re-architecture, and retraining. The discount is only real net of the exit cost.
• For startups, treat cloud cost as a unit-economics decision, not a procurement one. If your product passes through cloud-compute cost, the provider’s pricing is your gross margin. Read your cloud bill the way you read COGS, the discipline at the heart of the AI infrastructure market map.
• Watch the margin line, not just the growth line. A provider growing 40% to 60% while capex compresses margin is a different investment-grade story than one growing the same rate with stable margin. The growth headline can mask the unit-economics reality, the exact tension in AWS margin pressure and the cloud reset.
• Read backlog as the leading indicator. For Google Cloud specifically, the $462B backlog tells you more about the next two years than the current revenue line does (Alphabet Form 8-K, Q1 2026). When a provider discloses backlog, weigh it alongside revenue, not under it.
• Do not confuse bundle growth with infrastructure wins. Azure’s growth is partly enterprise lock-in pulling Azure along, not head-to-head infrastructure victories. Separate “they chose Azure on merit” from “they were already a Microsoft shop” before drawing competitive conclusions.

The bear case: what the skeptics get right

The strongest objection to the three-way oligopoly framing is that it imposes neat structure on a market that is messier, less comparable, and less stable than the Scorecard implies. The skeptic has three good points.

The numbers are not comparable, so the ranking is partly an artifact. AWS’s growth is a full-year figure, Azure’s and Google Cloud’s are single quarters, and Azure’s revenue is not even disclosed (Amazon Form 10-K FY2025; Microsoft Form 8-K and Alphabet Form 8-K, Q1 2026). Comparing a 20% annual rate to a 63% quarterly rate is not clean. The skeptic is right that the precise growth ordering rests on mismatched periods, even if the directional ranking holds.

Market share depends entirely on the definition. Synergy’s 28/21/14 split measures IaaS and PaaS only. Include managed databases, SaaS, or different service buckets and the percentages move (Synergy Research Group, Q1 2026). The “63% combined” figure is real under one definition and different under another. Anyone citing a single share number without the definition is overstating its precision.

Backlog is a promise, not revenue. Google Cloud’s $462B backlog is a powerful signal, but it is contracted future spend, not booked revenue (Alphabet Form 8-K, Q1 2026). Contracts get renegotiated, ramped slowly, or include flexibility that softens the headline. Treating backlog as locked revenue overstates the certainty.

Weighing it: the bear case bounds the thesis, it does not break it. The exact growth percentages are period-dependent and the share figures are definition-dependent, so the right read is the ranking and the direction, not the decimal points. AWS leads on share and margin, Azure leads on distribution, Google Cloud leads on growth and committed backlog. That structure holds even after you discount every comparability problem the skeptic correctly raises.

Where this oligopoly is vulnerable: the edges and alternatives

A credible read names the holes, and there are three real ones.

The disclosure asymmetry undermines clean comparison. AWS reports a clean revenue and operating-income line; Microsoft bundles Azure into Intelligent Cloud and Microsoft Cloud; Google Cloud reports revenue and operating income but at a single-quarter cadence (Amazon Form 10-K FY2025; Microsoft Form 8-K and Alphabet Form 8-K, Q1 2026). Any three-way comparison is built on three different reporting structures, so the precision is lower than the tables suggest.

The capex bet is unhedged across all three. Every provider is spending heavily on AI infrastructure on the assumption that demand sustains. If AI workload growth slows or efficiency gains cut the compute required per unit of value, the capex becomes a multi-year margin drag for all three simultaneously. The oligopoly’s shared bet is also a shared risk.

The edges are where the disruption lives. The 37% of the market outside the big three is not nothing, and specialized providers, neocloud GPU specialists, sovereign-cloud regional players, and open-source-adjacent stacks, can take specific high-value workloads even without challenging the overall ranking. The oligopoly is durable at the center and contestable at the edges, and AI infrastructure is creating new edges faster than in the prior cloud era.

None of these is fatal on today’s evidence. All three are why this is an oligopoly under load, with the leadership stable but the margins and the marginal workload genuinely in play.

How the pieces fit together

The cloud wars resolve into a single structure once you read the filings instead of the keynotes.

1. Three firms hold about 63% of cloud infrastructure: AWS 28%, Azure 21%, Google Cloud 14% (Synergy Research Group, Q1 2026).
2. Growth runs inverse to size, Google Cloud +63%, Azure +40%, AWS +20%, so the smaller players gain share points while AWS stays largest (Amazon Form 10-K FY2025; Microsoft Form 8-K and Alphabet Form 8-K, Q1 2026).
3. AWS wins on disclosed margin (~35.4% operating margin, FY2025), Azure wins on enterprise distribution, and Google Cloud wins on growth and a record $462B backlog (Amazon Form 10-K FY2025; Microsoft Form 8-K, Q1 FY2026; Alphabet Form 8-K, Q1 2026).
4. The AI capex arms race is accelerating revenue and compressing margin at the same time, which is why the contest has moved off price.
5. The race now runs on committed backlog, bundle lock-in, and unit economics, not headline list price, which is the signature of a maturing oligopoly.

The firms reducing the cloud wars to “who is biggest” are reading one row of the Scorecard. The market is decided across all five, and right now the leader is growing slowest while the challenger is locking in the most future revenue. That tension, not the share ranking, is the story.

---

Analysis, not investment advice. Figures are drawn from public SEC filings cited inline by company and fiscal period (Amazon Form 10-K FY2025; Microsoft Forms 8-K and 10-Q, Q1 FY2026; Alphabet Form 8-K, Q1 2026) and from Synergy Research Group’s Q1 2026 cloud market share report. Frameworks here, including the Cloud Three-Way Scorecard, are for understanding business strategy and tradeoffs, not for making buy or sell decisions.

Want the full toolkit for reading filings like this, the segment-margin worksheet, the market-share comparison framework, and the Cloud Three-Way Scorecard template used above? It’s in the Tech Business Analysis Playbook.