Azure is Microsoft's cloud platform, and the cloud most large enterprises default to in 2026. If you've ever wondered why Azure shows up in every Fortune 500 architecture diagram even when AWS would technically be a better fit, the answer is mostly procurement and integration with Microsoft's ecosystem — Office 365, Active Directory, GitHub, Visual Studio, .NET. This post is for developers thinking about Azure for a real project, with an honest read of where it wins and loses.
What Azure is
Azure offers a service catalog comparable to AWS in breadth: virtual machines, managed databases, containers, serverless functions, AI, storage, networking, identity, DevOps tooling. Launched in 2010 as "Windows Azure," renamed to Microsoft Azure in 2014, and now the #2 cloud globally by revenue.
Where Azure wins
- Microsoft-ecosystem integration. If you live in Office 365, Active Directory, or .NET, Azure is the path of least resistance. Single sign-on with Entra ID (formerly Azure AD), seamless connection to Power Platform, M365 admin centers, etc.
- Hybrid cloud. Azure Arc lets you manage on-premises servers and other clouds from the Azure control plane. Real strength for enterprises with significant on-prem footprints.
- Compliance and government. Azure Government, Azure Stack, sovereign clouds. Required posture in many regulated industries.
- Azure OpenAI Service. Privileged access to OpenAI's models — GPT-5, Codex, image gen, voice — with enterprise controls (data isolation, regional deployment) the OpenAI public API doesn't offer.
- .NET / C# / Visual Studio integration. Best-of-class.
- GitHub integration. Microsoft owns GitHub. The DevOps loop from GitHub Actions to Azure deployment is among the smoothest.
- Enterprise sales. Microsoft's enterprise relationships pull Azure into deals AWS doesn't even see.
Where Azure trails
- Developer experience. The portal is widely considered the most cluttered of the three major clouds. Concepts overlap (App Service vs Container Apps vs Functions vs Container Instances) and discovery is hard.
- Documentation. Quantity is high, signal-to-noise is lower than AWS or GCP. Many docs lag behind product changes.
- Service breadth at the bleeding edge. AWS still ships niche services Azure doesn't match.
- Reliability incidents. Azure has had several high-profile multi-region incidents in the past few years. AWS and GCP haven't been incident-free either, but the perception has stuck.
- Naming. Service names change frequently — Azure AD became Entra ID; "Azure Web Apps" became "App Service"; etc.
The core services you'll actually use
- App Service — managed PaaS for web apps. Closest analog to AWS App Runner or Railway. Easy "push code, get URL" experience.
- Azure Functions — serverless functions, equivalent to AWS Lambda.
- Container Apps — managed container runtime, similar to AWS Fargate or Google Cloud Run.
- AKS — managed Kubernetes.
- Virtual Machines — the IaaS classic.
- Azure SQL Database / Cosmos DB / Database for PostgreSQL — managed databases. Cosmos DB is the multi-model NoSQL service.
- Blob Storage — equivalent to S3.
- Azure CDN / Front Door — content delivery and edge routing.
- Entra ID — identity. Critical if your customers are enterprises that already use Microsoft for SSO.
- Application Insights — observability, part of the broader Azure Monitor.
- Azure OpenAI Service — covered separately below.
- Azure DevOps — CI/CD, work tracking. Largely overlaps with GitHub now.
Azure OpenAI Service
This is Azure's most differentiated AI offering. Through a commercial agreement with OpenAI, Azure provides hosted versions of GPT-5, GPT-4o, GPT-4o-mini, Codex, Image gen, and voice models — deployed in Azure regions, billed through Azure, with enterprise compliance posture (data residency, no model training on your data, private networking).
Why use it vs the OpenAI API directly:
- Data stays in your chosen Azure region.
- Inherits Azure's compliance certifications (HIPAA, SOC 2, FedRAMP, etc.).
- Single Azure invoice, integration with Azure cost management.
- Private network access (peering, ExpressRoute).
- RBAC via Entra ID rather than API keys.
Why use OpenAI directly:
- Newest models are available on OpenAI before they reach Azure (sometimes weeks-to-months lag).
- Quotas and rate limits are sometimes more permissive on OpenAI's direct platform.
- Simpler if you're not otherwise an Azure shop.
For iOS app proxies running on Railway/Vercel, OpenAI direct is fine. Azure OpenAI is the right pick for enterprise customers requiring compliance posture.
Entra ID (formerly Azure AD)
Microsoft's identity service. Most enterprise customers in 2026 use Entra ID as their corporate identity provider. If your iOS app serves businesses, supporting "Sign in with Microsoft" via Entra ID is a real feature, and it's easier when your backend is on Azure.
For consumer apps without enterprise customers, Entra ID is overkill — use Apple Sign In + Google + email/password via your auth provider.
Pricing and credits
Azure pricing is in the same league as AWS — usage-based across many services. Two practical notes:
- Free tier is real. $200 of free credit for 30 days, plus 12 months of free tiers on selected services, plus always-free quotas on others. Good for testing.
- Reserved Instances / Savings Plans offer significant discounts (30-60%) for committing to one- or three-year usage. Useful only after you've established a stable workload.
- Cost surprises are common in Azure too. The "Cost Management + Billing" page in the portal is your friend.
Set a budget alarm on day one, same as on AWS.
When to pick Azure
Reasonable reasons to pick Azure:
- You're a .NET / C# shop.
- Your customers are large enterprises that mandate Azure.
- You need Azure OpenAI Service's enterprise OpenAI access.
- You're integrated tightly with Microsoft 365, GitHub Enterprise, or Power Platform.
- You have Azure credits from a Microsoft partnership (these are common).
- You need hybrid cloud or specific regulatory compliance Azure offers.
Less compelling reasons (where AWS, GCP, Railway, or Vercel are usually better):
- You want the "best developer experience" cloud (GCP and Railway are friendlier).
- You want the broadest service catalog (AWS still wins on breadth).
- You're a solo iOS developer with no Microsoft footprint (Railway).
A pragmatic Azure starter stack
If you've decided Azure is the right call:
- Compute: App Service (simplest) or Container Apps (if you want containers).
- Database: Azure Database for PostgreSQL.
- Storage: Blob Storage + Azure CDN or Front Door.
- Identity: Entra ID (if enterprise), or third-party (Auth0, Clerk) for consumer.
- Observability: Application Insights + Sentry.
- Secrets: Azure Key Vault.
- AI: Azure OpenAI Service if compliance requires; Anthropic direct otherwise.
- CI/CD: GitHub Actions deploying to Azure (Microsoft-native and clean).
- Budget alarm on day one.
See also: Backend Servers Explained, AWS Deep Dive, GCP Deep Dive.
- Microsoft — Azure documentation
- Microsoft — Azure OpenAI Service