AWS vs Azure vs GCP — Comparison for the DACH Mid-Market 2026

The decision between AWS, Microsoft Azure and Google Cloud Platform is, for many mid-market companies in Germany, Austria and Switzerland, not a technical but a strategic one — it shapes architecture, personnel profiles, licence costs and vendor dependency for the next five to ten years. Over the past years we have accompanied more than 60 cloud projects in the DACH mid-market — from greenfield platforms to classic lift-and-shift migrations — and we see three recurring mistakes: first, choosing a provider purely by market share without reference to the company's existing IT landscape; second, underestimating the GDPR implications for data residency and key ownership; third, the absence of a migration strategy that cements vendor lock-in for years. This article provides a sober assessment of the three hyperscalers, highlights their strengths and weaknesses from a DACH perspective, compares the most important service categories and delivers a sector mapping. For the full picture on cloud strategy see our Cloud & DevOps Guide for the Mid-Market.

Why the choice of provider is a board-level decision — market shares and DACH regions 2026

The global cloud infrastructure landscape in 2026 continues to be dominated by three providers. According to the latest analysis by Synergy Research and Canalys, Amazon Web Services holds approximately 30 to 32 percent of the global market share in the infrastructure segment, Microsoft Azure 24 to 26 percent and Google Cloud Platform 11 to 13 percent. The remaining 30 percent is distributed among Alibaba Cloud, Oracle Cloud, IBM Cloud and a range of regional providers such as OVH, Hetzner and STACKIT.

More informative for mid-market companies is the distribution in the German-speaking region. Here the picture looks different: Azure leads with around 38 percent according to Bitkom surveys from 2025, AWS follows with 32 percent, and Google Cloud reaches 14 percent. Azure's lead is explained primarily by the high penetration of Microsoft 365 in the DACH mid-market and the ease of bundling through existing Enterprise Agreements. If you already have Office 365, Azure credits, identity integration and licence consolidation come along for the ride.

For data residency, regional locations are decisive. Frankfurt is the most mature DACH hub across all three providers. AWS has operated eu-central-1 since 2014 with three Availability Zones. Azure offers Germany West Central (Frankfurt) plus Germany North (Berlin) for disaster recovery. Google Cloud operates europe-west3 in Frankfurt with three zones, supplemented by europe-west10 (Berlin, since 2023).

Zurich has gained considerably over the past three years, driven by Swiss banks and compliance requirements that call for data storage outside the EU. AWS opened the eu-central-2 region in 2023, Azure operates Switzerland North and Switzerland West (Geneva), and Google Cloud has europe-west6. Vienna is not yet a fully-fledged cloud hub in 2026 — Azure has Austria East partially in operation with a full rollout planned for 2027. AWS and GCP serve Austria primarily from Frankfurt with around 20 milliseconds of latency, which is sufficient for most workloads.

AWS — Strengths and weaknesses from a DACH perspective

Amazon Web Services is the oldest and broadest hyperscaler. The service portfolio spans more than 240 individual services — from classic virtual machines (EC2) and object storage (S3) through databases (RDS, Aurora, DynamoDB) to highly specialised services such as satellite ground stations or quantum computing test environments. This breadth is simultaneously the most important strength and a significant weakness.

AWS's strengths lie in three areas. First, the service portfolio: virtually every conceivable use case has a matching AWS service, often in multiple variants. Second, documentation and community maturity — the official documentation is widely regarded as excellent, complemented by a massive Stack Overflow, GitHub and conference community. Third, the maturity of the core building blocks: EC2, S3, IAM and VPC have grown over fifteen years and are production-proven at a scale no other provider matches.

The weaknesses are equally clear. The complexity means that even simple architectures quickly encompass twenty or thirty interconnected services. Pricing is notoriously opaque — egress charges, regionally varying hourly rates, tiered discounts and reservation models make reliable budget planning difficult. For DACH mid-market companies there is the added factor that AWS support is primarily English-language, and German-speaking account managers are available to varying degrees depending on the contract tier. AWS is regularly the right choice when a company has no strong Microsoft ties, Linux workloads predominate and sufficient in-house cloud know-how is available or can be hired.

Azure — Strengths and weaknesses

Microsoft Azure is the de-facto standard in the DACH region for companies with an existing Microsoft landscape. The platform strategy is explicitly designed as an extension of the Microsoft ecosystem — Active Directory, Office 365, Windows Server, SQL Server and Dynamics connect seamlessly through Azure identity and networking services.

The strengths lie clearly in integration. Microsoft Entra ID (formerly Azure AD) is the standard identity platform for most DACH mid-market companies and can be coupled to the on-premises Active Directory with minimal effort. Windows Server licence mobility through Azure Hybrid Benefit reduces licence costs measurably — companies with existing Software Assurance agreements often run Azure VMs 30 to 40 percent cheaper than comparable AWS EC2 instances with Windows. The hybrid strategy via Azure Arc, which brings resources running on-premises or in other clouds under unified management, is in 2026 the most mature hybrid solution on the market.

The weaknesses lie primarily in tooling consistency and platform maturity. The Azure CLI (az) and PowerShell modules are functional but not as consistent as their AWS counterparts — the same operations across different services often follow different syntax conventions. Some new services are released in preview status and mature more slowly than expected. Outages are rare, but when they occur they frequently affect multiple coupled services simultaneously — the prominent Frontdoor incident in 2024 and several Entra incidents in 2025 remain in memory. For pure Linux workloads with no Microsoft connection, Azure tends to be less optimised than AWS or GCP, even though the gap has narrowed in recent years.

Google Cloud Platform — Strengths and weaknesses

Google Cloud Platform is the specialist with clear focal points. Its global market share among the three hyperscalers is the smallest, but GCP leads in two areas that are decisive for certain use cases: data and networking.

The data strengths lie in BigQuery, the analytical data warehouse that handles many analytics workloads faster and more cost-effectively than the AWS and Azure equivalents. The ML platform Vertex AI, the seamless integration with TensorFlow and proximity to DeepMind research make GCP the natural choice for companies whose strategy is heavily oriented around data analytics or generative AI. The global backbone network — the same one that powers Google Search and YouTube — delivers measurably lower latencies and higher inter-region bandwidth than the standard tiers of the two competitors. Kubernetes workloads are also frequently more comfortable on GCP because Google invented Kubernetes and Google Kubernetes Engine (GKE) is considered the most mature managed Kubernetes platform.

The weaknesses lie in breadth and DACH market share. The service portfolio at around 120 services is considerably smaller than AWS — many niche services are absent or less mature. Enterprise sales presence in the DACH mid-market is weaker than AWS and Azure, and the partner channel is less densely populated. For companies without a strong data or Kubernetes focus, there are rarely compelling reasons to choose GCP over the two market leaders. A pragmatic recommendation from our practice: GCP is an excellent second cloud for analytics and ML workloads alongside an Azure or AWS primary platform, but rarely the obvious first choice for a complete mid-market IT setup.

Side-by-side comparison of the most important categories

The table below maps the central service categories side by side. It is intended as orientation — selecting the right service for a specific project requires a brief architecture workshop, because suitability is highly dependent on the concrete use case.

Category	AWS	Azure	GCP
Compute (VM)	EC2, broadest instance portfolio	Virtual Machines, Hybrid Benefit for Windows	Compute Engine, good custom machine types
Object Storage	S3, de-facto standard	Blob Storage, multiple tiers	Cloud Storage, simpler pricing
Managed SQL	RDS, Aurora (very broad)	Azure SQL, strong MSSQL integration	Cloud SQL, Cloud Spanner (global)
NoSQL	DynamoDB	Cosmos DB (multi-model)	Firestore, Bigtable
Data Warehouse	Redshift	Synapse Analytics	BigQuery (market leader)
Networking	VPC, Transit Gateway	VNet, ExpressRoute	VPC, Premium-Tier Global Backbone
Kubernetes	EKS	AKS	GKE (most mature platform)
Serverless	Lambda	Azure Functions	Cloud Run, Cloud Functions
Identity	IAM	Entra ID (formerly Azure AD)	Cloud IAM
Pricing	Complex, many options	Bundles via EA, Hybrid Benefit	Sustained-use discounts automatic
Free Tier	12 months + Always Free	12 months + USD 200 credit	90 days + USD 300 credit, Always Free
EU regions DACH	Frankfurt, Zurich	Frankfurt, Berlin, Zurich, Geneva, Vienna	Frankfurt, Berlin, Zurich

Noteworthy is the regional density: Azure leads in the DACH region with five locations, GCP follows with three and AWS with two. For companies with hard data residency requirements — such as Austrian health workloads or Swiss financial data — this density quickly becomes a selection criterion.

Which cloud for which sector

Sector patterns are not deterministic, but they help with the initial narrowing-down. From more than 60 accompanied projects in our practice the following tendencies emerge:

• Mechanical engineering and manufacturing mid-marketPredominantly Azure, because the historical Microsoft landscape (AD, Exchange, SharePoint, SQL Server) sets the migration path. AWS is the choice when IoT platforms, edge compute or cloud-native greenfield projects are the priority.
• Retail and e-commerceMixed — AWS is often the first choice for shop platforms due to the scalability of Lambda, DynamoDB and CloudFront. Azure is strong when ERP (Dynamics, SAP) and Office 365 integration are central. GCP is brought in for recommendation engines and personalisation via BigQuery.
• Financial services and insuranceHeavily regulated; in Switzerland frequently Azure with Switzerland regions, in Germany Azure or AWS Frankfurt. Data residency and key ownership (customer-managed keys, HSM) are the decisive selection criteria.
• Healthcare and pharmaData residency and data protection are the top priority — typically Azure with Germany West Central or AWS with eu-central-1 plus additional data processing clauses. GCP is rare except for research and genomics workloads.
• Software and technology companiesOften AWS or GCP — AWS for service breadth, GCP for Kubernetes and data strengths. Azure plays a smaller role unless enterprise customers expect Microsoft stacks.
• Public sector and government agenciesGrowing importance of sovereign solutions such as STACKIT or sovereign cloud offerings from Azure and AWS. Pure hyperscaler migrations are increasingly difficult to push through without an explicit sovereignty concept.

Hybrid scenarios — Azure Arc, AWS Outposts, GCP Anthos

Hybrid architectures — in which part of the IT remains in the company's own data centre or at a colocation provider and another part runs in the cloud — are the norm in the DACH mid-market in 2026. All three hyperscalers offer their own hybrid platforms, which differ significantly in maturity and concept.

Azure Arc pursues the most comprehensive approach. With Arc, local servers, Kubernetes clusters, databases and even workloads from other cloud providers are brought under a unified Azure management plane. Configuration management, compliance policies, monitoring and patching all run through the same tooling as native Azure resources. For companies with distributed sites and mixed environments, Arc is the most mature hybrid platform in 2026.

AWS Outposts is conceptually the opposite: AWS delivers a pre-configured rack with AWS hardware into the customer's data centre, where it runs as an extension of an AWS region. The advantage is a genuine AWS experience on-premises — the same APIs, the same tools. The downsides are hardware costs, space requirements and dependency on the AWS hardware roadmap. Outposts makes sense for latency-critical workloads or specific compliance requirements, but less so for classic hybrid consolidation.

Google Anthos bets entirely on Kubernetes as the abstraction layer. Anthos runs on GKE, on local clusters, on AWS and on Azure — all managed centrally. For companies that are already committed to Kubernetes-native architecture, Anthos is the most consistent multi-cloud platform. For classic VM or legacy workloads Anthos is less suitable, because it is conceptually designed for containerised applications.

Migration effort between clouds and vendor lock-in

The question of vendor lock-in is one of the most common topics in board discussions — and simultaneously one of the most frequently misunderstood. Some degree of lock-in is unavoidable with any cloud platform. The question is not whether it exists but how significant it is and which components are affected.

Migration effort broadly breaks down into three layers. Infrastructure-level building blocks such as virtual machines, block and object storage and standard network structures can be moved between providers with manageable effort — typically EUR 1,500 to 4,000 per VM workload including testing, cut-over and documentation. Managed database migrations are more demanding, costing between EUR 8,000 and 60,000 per database depending on size and downtime tolerance. Provider-specific services such as serverless functions, proprietary identity platforms or native ML platforms are generally complete re-implementations that cost between 20 and 60 percent of the original build effort, depending on complexity.

In practical terms: anyone who wants to minimise lock-in sticks to open standards wherever possible in architecture decisions — Kubernetes instead of a native container orchestrator, Postgres or MySQL instead of vendor-specific database dialects, OpenTelemetry instead of proprietary monitoring. This incurs somewhat higher costs in the initial phase and forgoes some of the convenient managed-service advantages, but pays off when switching providers or pursuing multi-cloud strategies. A deeper treatment of this trade-off can be found in our cluster on multi-cloud versus single-cloud.

GDPR and data residency

GDPR compliance of the three hyperscalers is fundamentally given in 2026, but needs to be assessed in detail. All three providers supply data processing agreements under Article 28 GDPR, operate EU regions and offer encryption options with customer-managed keys via hardware security modules. The solid minimum configuration for sensitive data typically includes: choosing an EU region, activating customer-managed keys via the respective key management service, supplementary contractual clauses and a documented data protection impact assessment.

The remaining dispute concerns transfers to the USA and governmental access under the US CLOUD Act. The EU-US Data Privacy Framework of July 2023 established a new legal basis — all three hyperscalers are certified under the Framework. Individual German supervisory authorities continue to assess residual risks differently. For companies with particularly sensitive data — health, financial, children's or law enforcement data — sovereign cloud offerings such as Microsoft Cloud for Sovereignty or the AWS European Sovereign Cloud (planned for 2026) are increasingly a supplementary option. For most mid-market use cases, a carefully configured standard hyperscaler region remains the pragmatic choice.

Reepa's experience with all three providers

In our consulting and project practice we have run workloads on all three hyperscalers since 2018 and have accumulated a mixed set of positive and negative experiences. From this experience a number of practical heuristics have crystallised.

First: the provider decision is too often treated as purely technical, but is roughly two thirds a personnel and organisational strategy decision. What skills exist in-house? Which can be found cost-effectively on the market? What vendor partnerships are in place? These questions determine long-term success more strongly than technical feature comparisons.

Second: the cost range between the three providers for comparable workloads is usually smaller than the internal operations and staffing effort. Anyone switching provider because of an 8 percent list-price difference between AWS and Azure typically has not yet leveraged the bigger cost levers (reservations, rightsizing, egress optimisation, FinOps discipline). Systematic cloud cost optimisation across all hyperscalers is the subject of our article on cloud costs and FinOps.

Third: migration projects without a clear architecture target fail regardless of the provider. Migrating "to the cloud" without a strategy typically results in building a virtual data centre with the disadvantages of both worlds — cloud pricing with a data-centre architecture. A structured approach from assessment to cut-over is described in our cloud migration guide.

Frequently asked questions