What the exam is and how it is scored

Microsoft Azure Fundamentals (AZ-900) is Microsoft's entry-level Azure exam. The skills measured are those published as of July 20, 2026. You need a score of 700 or greater to pass.

The exam is split into three skill areas, each with a weight:

• Describe cloud concepts: 25–30%
• Describe Azure architecture and services: 35–40%
• Describe Azure management and governance: 30–35%

The weight shows the relative share of the exam. The higher the percentage, the more questions that area contains. This guide covers the first area, 'Describe cloud concepts'. It is the smallest area. Technical IT experience is not required, though general IT knowledge helps.

The material here is based on Microsoft Learn, which is licensed CC BY 4.0.

What this area asks

The area breaks into three groups of objectives:

• Cloud computing basics: what it is, shared responsibility, cloud models, and the consumption-based pricing model.
• The benefits of cloud services: availability, scalability, reliability, predictability, security, governance, manageability and sustainability.
• The cloud service types: IaaS, PaaS and SaaS, and when each fits.

Much of this area comes down to telling two similar ideas apart. The sections below focus on those distinctions.

Cloud computing and the consumption-based model

Cloud computing is the delivery of computing services over the internet. Those services include virtual machines, storage, databases and networking. They also extend to Internet of Things (IoT), machine learning (ML) and artificial intelligence (AI).

Because the services arrive over the internet, you are not limited by a physical datacenter you must build yourself. If you need more capacity quickly, you expand in the cloud instead of waiting for new hardware.

The consumption-based model. You pay for the resources you use, and nothing more. You rent compute power and storage, then release them when finished.

The key distinction here is between two kinds of spending:

• Capital expenditure (CapEx): up-front spending on physical infrastructure such as servers, network hardware and datacenter space.
• Operational expenditure (OpEx): ongoing spending on services over time.

Cloud services are paid for as you consume them, so cloud computing is classified as an operational expense. If a question describes moving from buying servers to renting capacity, the shift is from CapEx to OpEx.

The benefits that follow from this model:

• No upfront cost for hardware or datacenter infrastructure.
• No need to buy and manage capacity that may sit underused.
• Resources can be added when demand rises.
• Resources can be released when demand falls.

Compare this with capacity planning in a traditional datacenter. There you estimate future needs in advance. Overestimate and you pay for idle equipment. Underestimate and performance suffers until more hardware is ordered, installed and provisioned. In the cloud you adjust to actual demand. Cloud providers use a pay-as-you-go pricing model, so you typically pay only for what you consume.

Shared responsibility

In an on-premises datacenter, you are responsible for everything: the physical space, security, servers, patching and software. In the cloud, responsibility is split between the provider and you.

Three lists are worth memorising exactly.

Always the provider's:

• The physical datacenter
• The physical network
• The physical hosts
• Infrastructure

Always yours:

• The information and data stored in the cloud
• The devices allowed to connect to your cloud (phones, computers and so on)
• The accounts and identities of people, services and devices in your environment

Depends on the service type:

• Operating systems
• Network controls
• Applications
• Identity and access

The third list is where questions try to catch you. The Learn example is a database. If you use a cloud SQL database, the provider maintains the database itself, but you are still responsible for the data you put into it. If you install SQL on a virtual machine you deployed, you are also responsible for database patches and updates.

Shared responsibility is tied directly to the service types. IaaS puts the most responsibility on you. SaaS puts most of it on the provider. PaaS sits in the middle.

Cloud models

The cloud model describes the deployment type. There are three main models, plus multicloud.

• Private cloud: used by a single entity. It gives your IT team much greater control, but at greater cost and with fewer of the benefits of public cloud. It can be hosted in your own datacenter, or in a dedicated datacenter offsite, even by a third party.
• Public cloud: built, controlled and maintained by a third-party provider. Anyone who wants to buy the services can use the resources. General public availability is the key difference from private cloud.
• Hybrid cloud: public and private clouds used together in an interconnected environment. A private cloud can surge into public resources for temporary demand, and you choose which services stay private.
• Multicloud: you use two or more public cloud providers, perhaps for different features or while migrating from one provider to another. You manage resources and security in each.

The comparison points the material gives:

• Public: no capital expenditure to scale up, quick provisioning and deprovisioning, pay only for what you use, but you do not have complete control over resources and security.
• Private: complete control over resources and security, and your data is not collocated with other tenants' data, but hardware must be bought for startup and maintenance, and you maintain and update it.
• Hybrid: the most flexibility, and you decide where applications run and control security, compliance or legal requirements.

Two Azure offerings attach to this topic:

• Azure Arc is a set of technologies for managing your environment whether it is public cloud on Azure, a private cloud in your datacenter, a hybrid setup, or a multicloud environment.
• Azure VMware Solution lets you run VMware workloads in Azure. It fits an organisation already established with VMware in a private cloud that wants to move to public or hybrid.

Benefits of cloud services

This group is a set of terms that sound alike. Learn what each one means.

High availability is about keeping resources available when needed, regardless of disruptions. Azure has uptime guarantees that depend on the service, and these guarantees are part of the service-level agreements (SLAs).

Scalability is the ability to adjust resources to meet demand. It also means you are not overpaying when demand drops. There are two kinds:

• Vertical scaling changes the capability of a resource. Scale up to add CPUs or RAM to a virtual machine, or scale down to lower them.
• Horizontal scaling changes the number of resources. Scale out to add virtual machines or containers, or scale in to remove them. Either can be automatic or manual.

Reliability is the ability of a system to recover from failures and keep functioning. It is one of the pillars of the Microsoft Azure Well-Architected Framework. Resources deployed in regions around the world mean that if one region has a catastrophic event, others keep running.

Predictability comes in two forms:

• Performance predictability: predicting the resources needed to deliver a good customer experience. Autoscaling, load balancing and high availability support it.
• Cost predictability: forecasting cloud spend. You can track use in real time, monitor efficiency and use analytics. The Azure Pricing Calculator gives an estimate of potential spend.

Security and governance. Templates help ensure deployed resources meet your technical and regulatory standards. Cloud-based auditing flags resources that fall out of compliance. Depending on your operating model, patches and updates may be applied automatically. Cloud providers are typically well suited to handling distributed denial of service (DDoS) attacks.

Manageability has two sides, and the exam likes to separate them:

• Management of the cloud: autoscaling, deploying from templates, monitoring health and replacing failing resources, and alerts on configured metrics.
• Management in the cloud: how you work with your environment: a web portal, a command line interface, APIs, or PowerShell.

Sustainability. The practices listed are scaling down when demand drops, turning off or deallocating unused resources, choosing efficient configurations to avoid overprovisioning, and using governance and monitoring to track usage. A development environment that shuts down overnight and at weekends is the example given.

Cloud service types

The three service types differ in how much you manage.

IaaS is essentially renting hardware in a cloud datacenter; what you do with it is up to you. Typical scenarios are lift-and-shift migration and testing and development environments.

PaaS is the middle ground. You do not worry about licensing or patching operating systems and databases. Typical scenarios are a development framework and analytics or business intelligence.

SaaS is a fully developed application you rent or use. Email, messaging, productivity and finance applications are the examples. It is the easiest to get running and needs the least technical knowledge.

Common traps

• Calling cloud CapEx. Cloud is OpEx, because you pay as you consume.
• Mixing up up/down and in/out. Up and down belong to vertical scaling (bigger or smaller resource). Out and in belong to horizontal scaling (more or fewer resources).
• Assuming private means on-site. A private cloud serves one entity but can be hosted offsite, even by a third party.
• Confusing hybrid and multicloud. Hybrid combines public and private. Multicloud uses multiple public providers.
• Believing the provider owns your data. Data, devices and identities are always yours, in every model.
• Treating operating systems or identity as fixed. These depend on the service type, so read which type the question names.
• Treating a managed database like one you install yourself. A cloud SQL database shifts maintenance to the provider. SQL on your own virtual machine leaves patching with you.
• Merging manageability's two meanings. Autoscaling and template deployment are management of the cloud. Portal, CLI, API and PowerShell are management in the cloud.
• Swapping reliability and high availability. Reliability is recovering from failure and continuing. High availability is maximum availability, backed by SLAs.

How to study

1. Build the three lists first. Write the always-provider, always-you and depends-on-service lists from memory until you can do it without checking. 2. Draw the service-type table yourself. Fill in what you and the provider manage for IaaS, PaaS and SaaS. Then check it against the material. 3. Study in pairs. Learn CapEx with OpEx, vertical with horizontal, private with hybrid, and reliability with predictability. The material presents each pair side by side. 4. Make up scenarios. For each cloud model and service type, write one sentence describing a case where it is the right choice. The objectives ask you to identify appropriate use cases. 5. Practise under question conditions. Reading is not enough. Answer questions, read the explanation for every wrong answer, and go back to the section it points to. 6. Keep the weight in mind. This area is 25–30%, so give it steady attention, but do not neglect the two larger areas.

Practise

When you can state the distinctions above without notes, test yourself with questions. You can practise this area, and the rest of the exam, at AZ-900 Azure Fundamentals practice.

Practise this with Microsoft Azure Fundamentals (AZ-900) questions

More of this guide