Did hyperscalers hit their 2025 AI cloud revenue targets?

For healthcare organizations planning 2026–2027 AI workloads, the useful question is not whether the AI cloud market was “hot” in 2025. It was. The procurement question is narrower: did the major hyperscalers actually produce enough 2025 AI cloud revenue momentum to justify treating them as durable infrastructure partners, and can those numbers be compared cleanly?

The short answer on ai cloud revenue targets 2025 is yes, with important qualifications. AWS, Microsoft, and Google Cloud all exited 2025 with stronger-than-expected cloud growth tied heavily to AI demand. But “AI cloud revenue” is not a standardized reporting category. AWS disclosed the clearest AI run-rate signal, Microsoft disclosed Azure growth inside a broader Intelligent Cloud segment, and Google Cloud showed the fastest growth rate while also signaling that available compute constrained revenue.

That table is the best starting point because it separates three things that are often blurred together: total cloud scale, AI-labeled run rate, and deployable capacity. A hospital system does not buy a growth rate. It buys access to GPUs or specialized accelerators, storage and networking, compliance-ready services, engineering support, contractual remedies, and enough vendor durability to make a five-year architecture bet survivable.

AWS: the clearest scale story, and the clearest capex pressure

AWS remains the provider whose 2025 numbers are hardest to dismiss. It ended 2025 with $128.7 billion in total revenue and a $142 billion annualized run rate. More striking for anyone comparing vendor capacity is the incremental revenue figure: AWS added $21.2 billion during the year, more than Google Cloud’s entire annual revenue base cited in the same comparison [1][2].

The AI-specific disclosure also matters. Amazon said AWS AI services had crossed a $15 billion annualized run rate by early 2026 [3]. That is not the same thing as audited full-year AI revenue, and it should not be read as a clean apples-to-apples number against Microsoft or Google. But it is a useful signal: AI demand is large enough inside AWS to be described as a major business line rather than an experimental overlay.

For healthcare buyers, the other side of that signal is spending pressure. AWS’s free cash flow fell from $26 billion to $1.2 billion as Amazon invested heavily in AI infrastructure [2]. That does not make AWS fragile; it shows the scale of the buildout required to serve AI demand. It also means procurement teams should assume that premium AI capacity, especially for accelerator-heavy workloads, will remain a capacity-managed resource rather than an endlessly elastic commodity.

That distinction changes contract conversations. If a health system is planning ambient documentation, imaging model development, payer-provider automation, or large-scale clinical summarization, it should ask not only whether the vendor offers the relevant AI service, but whether the requested region, instance class, accelerator type, data residency posture, and support model can be reserved on the implementation timeline. Revenue scale is reassuring only if it maps to available infrastructure for the workloads being contracted.

Microsoft: strong Azure growth inside a less clean reporting boundary

Microsoft’s 2025 cloud performance is commercially important, but its disclosure structure requires more care. Intelligent Cloud reached about $131 billion in annualized run rate, and Azure grew 39% in Q4 2025 [1]. For a healthcare organization already standardized on Microsoft identity, productivity, security, analytics, or developer tooling, that Azure growth is a real infrastructure signal.

The caution is that Intelligent Cloud is not Azure alone. It includes server products and enterprise services, so the segment’s ARR should not be treated as direct Azure revenue. The Azure growth rate is Azure-specific, but the absolute Azure revenue base is not disclosed in isolation. This is exactly the kind of reporting boundary that can mislead a procurement committee if a slide converts segment size into assumed Azure AI capacity.

Microsoft’s AI business had already crossed a $13 billion run rate in late 2024, according to the same earnings comparison [1]. That confirms substantial AI monetization before 2025 closed, but it does not define AI revenue in the same way as AWS’s AI-services run rate. A health system comparing the two should resist ranking them by those two numbers alone. The better question is whether Microsoft can allocate the needed Azure capacity, governance controls, model access, and enterprise support within the buyer’s existing Microsoft estate.

Microsoft’s advantage in healthcare procurement is often not a single AI revenue number. It is the installed enterprise relationship: identity, productivity workflows, endpoint management, security tooling, and data platforms already sitting inside the organization. Its revenue momentum supports the case that Azure AI infrastructure is a strategic priority, but the reporting format still leaves buyers needing more contract-level proof.

Google Cloud: faster growth, smaller base, real capacity questions

Google Cloud’s 2025 numbers make it more difficult to treat the market as a two-provider story. It ended 2025 with a $71 billion annualized run rate and grew 48% year over year in Q4 2025, the fastest growth rate among AWS, Microsoft, and Google Cloud in the CRN comparison [1].

That faster growth can matter in negotiations. A provider gaining share may be more aggressive on pricing, migration support, co-development, or workload-specific commitments. Google’s AI research position and cloud growth trajectory may be particularly relevant for organizations building analytics-heavy or model-development environments rather than simply consuming packaged AI features.

The limitation is practical availability. Google Cloud acknowledged that revenue would have been higher if demand could have been met, reflecting compute constraints through 2025 [1]. For hospitals and health systems, that means growth rate alone should not be treated as proof that the required accelerator capacity will be available in the preferred region and contract window.

The market is large enough to justify investment, but not clean enough for lazy comparison

The broader cloud market confirms why these numbers matter. Gartner estimated global public cloud spending at $723.4 billion in 2025, up 21.5% from 2024, with AI described as the primary catalyst in the earnings comparison [1]. That scale helps explain why hyperscalers are still building aggressively even while infrastructure costs pressure margins and cash flow.

Healthcare is a smaller but consequential part of that demand. Mordor Intelligence estimated the healthcare cloud computing market at $54.69 billion in 2025, with North America holding 48.3% share and the market growing at an 11.09% CAGR toward $102.77 billion by 2031 [4]. Those figures describe healthcare cloud computing broadly, not AI cloud revenue specifically. They are useful for sizing the procurement environment, not for claiming that healthcare AI alone is driving hyperscaler growth.

Healthcare-specific generative AI spend is narrower still. Menlo Ventures put healthcare generative AI spending at $1.4 billion in its 2025 state-of-the-market analysis [5]. That number covers healthcare-specific generative AI solutions; it does not include every general-purpose AI tool used by health systems, every non-LLM AI system, or the underlying cloud infrastructure embedded in broader enterprise contracts. Blending it with total healthcare cloud spend would create a false precision that procurement teams cannot use.

The same caution applies to “AI in healthcare” market estimates more broadly. Different firms use different boundaries: software only or software plus services; generative AI only or all AI; provider, payer, pharma, and life sciences together or separately. Those forecasts can help frame urgency, but 2025 hyperscaler actuals are more useful when the decision is whether a vendor has the operating scale and capital commitment to support infrastructure contracts already being negotiated.

What the 2025 actuals should change in healthcare cloud decisions

The first change is vendor viability assessment. None of the three major providers looks like a speculative AI infrastructure bet based on 2025 cloud performance. AWS has the largest scale signal and a disclosed AI run rate above $15 billion. Microsoft has substantial Azure growth and deep enterprise integration. Google Cloud has the fastest growth rate and continued AI infrastructure momentum. For a health system, this supports negotiating with all three as serious counterparties rather than reducing the field too early on perceived market strength alone.

The second change is negotiating posture. Strong hyperscaler growth does not automatically improve buyer leverage. In constrained AI infrastructure markets, demand can strengthen vendor pricing power for scarce capacity. The relevant negotiation is less about headline cloud discounts and more about reserved capacity, regional availability, accelerator commitments, egress terms, support response, model governance, audit rights, business associate agreements, and remedies if capacity is not delivered when clinical or operational programs depend on it.

The third change is capacity planning. AI workloads do not behave like ordinary SaaS expansion. A pilot may run acceptably on limited capacity, then hit cost or availability constraints when scaled across a health system. If all three providers were compute-constrained through 2025, a hospital should not assume that an approved budget automatically converts into deployable AI infrastructure on the desired timeline [2][3].

The fourth change is architecture discipline. A healthcare organization may decide to use one hyperscaler for enterprise AI services, another for data science workloads, and a third-party or private environment for particularly sensitive or latency-bound use cases. Revenue numbers cannot decide that architecture, but they can identify which vendors are investing at a scale likely to support mature AI service roadmaps.

Capex is the underappreciated procurement signal

Revenue shows demand. Capex shows what vendors are doing about supply. CloudZero reported that the big five hyperscalers collectively spent more than $600 billion on capex in 2026, with about 75% tied to AI infrastructure [6]. That figure is not limited to healthcare, and it is not a direct measure of available clinical AI capacity. It does, however, confirm that the largest providers are treating AI infrastructure as a capital-intensive race rather than a software-margin add-on.

For healthcare buyers, that cuts both ways. Heavy capex makes it more plausible that the vendor can support large AI workloads over a multi-year contract. It also increases the likelihood that vendors will protect margins through capacity tiering, committed-use structures, premium support models, and workload-specific pricing. A procurement team that asks only for a discount may miss the more important clause: who receives scarce capacity when demand exceeds supply.

Q1 2026 confirms momentum, not certainty

The early 2026 data does not overturn the 2025 picture. It extends it. Google Cloud accelerated to 63% growth in Q1 2026, according to MindStudio’s analysis of the AI infrastructure race [7]. That continuation strengthens the case that AI infrastructure demand did not fade after 2025 budget cycles closed.

It should still be treated as forward context rather than a new procurement answer. A single quarter can show momentum, but healthcare infrastructure decisions need evidence of service fit, contractual reliability, compliance controls, integration pathways, and regional capacity. Q1 growth is encouraging; it does not remove the need for workload-level due diligence.

How to use the numbers without overusing them

The disciplined use of 2025 hyperscaler actuals is as a benchmark layer, not a vendor-selection shortcut. They can help a healthcare organization answer whether the provider is investing at scale, whether AI demand is material to the provider’s cloud business, whether competitive dynamics are shifting, and whether supply constraints should be written into the procurement plan.

They cannot determine which hyperscaler is best for a given healthcare AI workload. That requires comparing data architecture, HIPAA and business associate agreement posture, identity and access controls, model governance, observability, region strategy, migration cost, internal skills, and ecosystem alignment. For that layer of the decision, revenue analysis should sit alongside a workload-level comparison such as ClinicalMind’s AWS, Azure, and GCP for healthcare AI infrastructure comparison.

The procurement conclusion is intentionally limited. The 2025 actuals show that AWS, Microsoft, and Google Cloud all had real AI-driven cloud demand and continued investment capacity. They also show why healthcare organizations should not treat AI cloud revenue as a standardized scoreboard. Use the numbers to pressure-test vendor durability and market momentum, then make the contract decision on reserved capacity, compliance fit, workload architecture, service commitments, and the consequences if promised compute is not available when the clinical program is ready to scale.

References

1. AWS Vs. Microsoft Vs. Google Cloud Earnings Q4 2025 Face-Off, CRN
2. Think AWS Is Losing To Azure and Google Cloud? You Need To Hear This Quote From Amazon CEO Andy Jassy, Motley Fool
3. Amazon CEO reveals AI revenue, dismisses spending doubts in annual letter, Reuters
4. Healthcare Cloud Computing Market Size Report 2026-2031, Mordor Intelligence
5. 2025: The State of AI in Healthcare, Menlo Ventures
6. 100+ Cloud Computing Statistics: A 2026 Market Snapshot, CloudZero
7. Google Cloud vs AWS vs Azure Q1 2026 — Which Hyperscaler Is Winning the AI Infrastructure Race?, MindStudio