The posture management tier is free, and most Azure subscriptions have never switched it on.
The product scores your cloud security posture, measures it against a built-in benchmark that reaches across AWS and GCP as well as Azure, and defends servers, containers, storage, databases and APIs through separate plans you switch on only where each one earns its keep. That first part costs nothing whatsoever, and almost nobody has enabled it.
- Free tierFoundational CSPM, including secure score
- AWS and GCPMulticloud, agentless
- Attack pathsHow an attacker would actually get in
- Ten plansEnabled selectively, not all at once
Four capabilities are in the free foundational tier.
The documentation says plainly that free foundational CSPM capabilities are included. In our experience most American Azure estates have never enabled a single one of them, which makes this the cheapest security improvement any of them could make.
- Secure score summarizes your posture out of the security recommendations and rises as you remediate them. It gives you a figure to report upward and a ranked list to work down, both running on subscriptions you are already paying for.
- Centralized policy management turns your security conditions into recommendations naming the resources that violate them, with the Microsoft cloud security benchmark sitting there as a built-in standard spanning Azure, AWS and GCP alike.
- Multicloud coverage connects AWS and GCP environments agentlessly for posture insight, letting a mixed estate be assessed against a single benchmark rather than turning into three separate conversations.
- Code pipeline insights connect GitHub, Azure DevOps and GitLab to surface infrastructure as code misconfigurations and exposed secrets. Beyond the free tier, paid Defender CSPM adds attack path analysis, the cloud security explorer, regulatory compliance, governance and data security posture management. Those are the capabilities worth taking to a budget conversation, once the free tier has already shown you where you stand.
Eight capabilities, and the line between free and paid.
Secure score, in the free foundational tier
Free foundational CSPM capabilities are included, and secure score sits among them. It summarizes your posture from the security recommendations and climbs as you remediate them. Any business holding Azure subscriptions with no posture management in place at all gets a genuine, immediate improvement here, costing nothing beyond the effort of acting on what it tells you.
AWS and GCP in the same view, also free
Multicloud coverage lives in that foundational tier as well, connecting AWS and GCP environments agentlessly for posture insight. Plenty of American businesses ended up multicloud through an acquisition, a particular workload requirement, or a decision a developer made that nobody ever reviewed. For those, a single posture view spanning all three providers is usually the first occasion anybody has seen the complete picture.
The Microsoft cloud security benchmark, applied across providers
Centralized policy management is another free tier capability, turning security conditions into recommendations that name the resources violating your policy. Behind it sits the Microsoft cloud security benchmark, a built-in standard applying security principles with detailed technical implementation guidance for Azure and, importantly, for other providers including AWS and GCP. What that hands a mixed estate is one consistent yardstick instead of three separate ones.
Attack path analysis is the paid capability that genuinely justifies its cost
It models traffic moving across your environment to identify potential risks before you change anything, with the cloud security explorer letting you query a map of that environment directly. This is what turns a list of eight hundred recommendations into the small handful that genuinely chain together into a route somebody could walk. Those two documents produce entirely different conversations at board level.
Workload plans you enable selectively
Individual plans cover servers, containers, storage, databases, App Service, Key Vault, Resource Manager, APIs and AI services. The design intent is that you enable whatever your estate genuinely contains rather than the full set, and the plan by plan decision is exactly where the cost conversation belongs. Most businesses need two or three of them. Almost none need ten.
Storage protection, including SAS token misuse
Protection here covers malware, threats specific to storage, leakage of sensitive data and misuse of Shared Access Signature tokens. That final item deserves emphasis. SAS tokens get handed around freely during projects, are almost never revoked afterwards, are frequently scoped far wider than they needed to be, and nobody monitors them. They come up as a finding in Azure estate after Azure estate.
DevOps security, from code rather than after deployment
Code pipeline insights connect your GitHub, Azure DevOps and GitLab repositories, surfacing infrastructure as code misconfigurations and exposed secrets, then correlating those findings against cloud security context so remediation can be prioritized where the code lives. This appears under the foundational free tier as well as under Defender CSPM, which makes it an unusually inexpensive place to begin.
AI workload security, which is newly relevant
AI security posture management discovers the generative AI applications running in your environment and identifies their vulnerabilities, producing what is termed an AI bill of materials, while AI threat protection watches for threats aimed at those workloads. Anyone building on Azure OpenAI needs this simply to know which AI workloads exist, and that list is regularly longer than anything the security team was told about.
Four things that stop this turning into a very expensive dashboard.
The free tier goes on and gets interpreted before we propose anything you pay for
At no cost, foundational CSPM hands you a secure score, recommendations driven by policy, multicloud posture and code pipeline insights. Working through that first shows you the shape of the problem, establishes a baseline you can measure future progress against, and ensures every plan decision afterwards rests on your own findings rather than on a sales conversation.
We prioritize by attack path, not by recommendation count
A large estate throws off hundreds of recommendations, and working through them by severity alone is both slow and demoralizing for the people doing it. Attack path analysis picks out the chains that genuinely lead somewhere, and that list is invariably far shorter. Break one link and the entire path disappears, which is a considerably better use of a limited engineering budget.
We enable plans selectively and say so
Ten plans exist and most estates warrant two or three of them. We map what you genuinely run against the plans covering it, and we name the ones we would leave switched off just as clearly, because recommending you buy everything is not advice. Where the free tier is sufficient for a given workload, we say so.
We settle ownership before turning on alerting
Every workload protection plan produces security alerts, and an alert needs somebody to respond to it. Before enabling any of them we establish who receives an alert, who owns remediating a misconfiguration inside a given subscription, and what happens when one arrives at two in the morning. Across a multi-subscription estate, ownership is the genuinely difficult question, and it is the one determining whether anything improves at all.
Six US situations where cloud posture is the actual gap.
An estate with subscriptions nobody can fully account for
Some were created for projects, some inherited through an acquisition, and some spun up by a developer holding a corporate card and formalized long afterwards. Turning the free posture tier on across every subscription is the fastest route to seeing what genuinely exists and how exposed each part is. The inventory on its own usually justifies the exercise.
A regulated firm needing cloud compliance evidence
When a SOC 2 auditor, a HIPAA assessor, an NYDFS Part 500 examiner or an enterprise customer asks how your cloud workloads are secured and against which standard, the regulatory compliance capability inside Defender CSPM maps your posture onto recognized standards and generates the evidence in a form that goes straight into the audit pack.
A development team shipping infrastructure as code
Connecting GitHub, Azure DevOps and GitLab through code pipeline insights surfaces infrastructure as code misconfigurations and exposed secrets while they are still text rather than deployed resources. Catching one in a pull request costs a few minutes of an engineer attention. Catching the same thing in production, after an auditor found it first, costs a great deal more than that.
An organization that ended up multicloud without deciding to
One workload on AWS because a vendor insisted, something on GCP that arrived with an acquisition, and everything else on Azure. Connecting AWS and GCP agentlessly for posture insight sits in the free tier, and assessing all three against the Microsoft cloud security benchmark is usually the first genuinely consistent view anybody in the business has had.
A business running production databases and storage in Azure
Defender for Databases reaches Azure SQL, SQL running on machines, the open-source relational databases and Cosmos DB. Defender for Storage handles malware, leakage of sensitive data and misuse of SAS tokens. Wherever the data effectively is the business, those two plans justify themselves faster than any of the others.
An organization building on Azure OpenAI
The posture side discovers your generative AI applications and identifies their vulnerabilities, producing an AI bill of materials in the process, while threat protection watches for attacks aimed at those workloads. For most businesses the first piece of value is simply establishing which AI workloads exist at all, since that list routinely runs longer than anything the security team was ever told about.
What organizations actually know about their cloud security posture.
| Feature | Defender for Cloud in use | Free tier only | Nothing enabled |
|---|---|---|---|
A posture score you can track over time | Yes | Yes | No |
Misconfigurations surfaced against a benchmark | Yes | Yes | No |
AWS and GCP in the same view | Yes | Yes | No |
Attack paths identified across resources | Yes | No | No |
Sensitive data located across cloud storage | Yes | No | No |
Regulatory compliance reporting | Yes | No | No |
Threat detection on servers and containers | Yes | No | No |
Storage and SAS token misuse detected | Yes | No | No |
Infrastructure as code issues caught before deployment | Yes | Yes | No |
Could answer a SOC 2 cloud question from a report | Yes | Partly | No |
Ten plans, and what each one is actually protecting.
Plan
Foundational CSPM, free
- What it protects
- Posture management, secure score, policy, the multicloud connection and code pipeline insights
Plan
Defender CSPM
- What it protects
- Attack path analysis, the cloud security explorer, governance, regulatory compliance, and posture for data and AI
Plan
Defender for Servers
- What it protects
- Windows and Linux machines wherever they run, whether Azure, AWS, GCP or your own building
Plan
Defender for Containers
- What it protects
- Kubernetes hardening, vulnerability assessment and runtime protection
Plan
Defender for Storage
- What it protects
- Malware, threats particular to storage, leakage of sensitive data, and misuse of SAS tokens
Plan
Defender for Databases
- What it protects
- Azure SQL, SQL running on machines, the open-source relational databases and Cosmos DB
Plan
Defender for App Service
- What it protects
- Attacks aimed at the web applications and APIs you run on App Service
Plan
Defender for Key Vault
- What it protects
- Unusual or hostile attempts to reach or exploit a Key Vault account
Plan
Defender for Resource Manager
- What it protects
- Unusual and potentially harmful resource management operations
Plan
Defender for APIs
- What it protects
- Visibility across the APIs your business depends on, with threat detection in real time
Plan
AI Services
- What it protects
- Threats to generative AI applications, detected in real time
Five steps, and the first one is free.
- 1
Enable foundational CSPM across every subscription
That includes the subscriptions nobody remembered existed, plus connecting AWS and GCP wherever those are in play. All of this is the free tier. It produces a secure score and a ranked set of recommendations, and it sets the baseline everything afterwards gets measured against.
- 2
Interpret the findings rather than forwarding them
A large estate generates a great many recommendations, and handing over the raw list accomplishes nothing at all. We go through them alongside your team, separating what is genuinely exposed from what is merely theoretically imperfect, and produce a remediation list somebody could realistically finish inside a quarter.
- 3
Decide which workload plans your estate warrants
Plans get mapped against what you actually run, whether that is servers, containers, storage, databases, App Service, Key Vault, APIs or AI services. We name the ones we would leave off with the same clarity as the ones we would switch on, because enabling selectively is how this product was designed to be used.
- 4
Add attack path analysis and prioritize by chain
Wherever Defender CSPM is justified, attack path analysis and the cloud security explorer between them cut hundreds of findings down to the handful of chains genuinely leading somewhere. That is the moment the reporting stops being a compliance artifact and starts driving actual engineering decisions.
- 5
Establish ownership and the response path
Four questions get answered: who owns remediation within each subscription, how a recommendation turns into a ticket somebody works, who receives a security alert from a workload plan, and what happens to that alert outside working hours. Governance rules can assign tasks to resource owners and track their progress, and that mechanism is what keeps the score from drifting back down again.
What organizations ask about Defender for Cloud.
Fifteen questions worth answering first.
What you actually run
- How many Azure subscriptions do you have?Including the ones a project team created and forgot.
- Do you also have AWS or GCP workloads?Multicloud posture is in the free tier.
- Do you run containers or Kubernetes?A separate plan, and a common blind spot.
- Do you have storage accounts with SAS tokens issued?SAS misuse is specifically covered.
- Are you building anything on Azure OpenAI?AI posture management applies.
What you already have
- Has foundational CSPM ever been enabled?It is free and frequently untouched.
- Do you know your current secure score?If not, that is the first number to get.
- Are your repositories connected?Code pipeline insights are in the free tier.
- Is the Microsoft cloud security benchmark applied?Built in, and covers AWS and GCP too.
- Do you have Defender for Servers Plan 2?It now includes suspicious DNS activity alerts.
Would findings be acted on
- Who owns remediation of a cloud misconfiguration?Frequently nobody, which is the real finding.
- Is there a change process for production subscriptions?Recommendations become tickets or they become noise.
- Would attack path findings reach a decision maker?They are the ones worth escalating.
- Do you need regulatory compliance reporting?That sits in the paid Defender CSPM plan.
- Do you have anybody watching alerts out of hours?Workload plans generate alerts that need a responder.
The pages around this one.
Cybersecurity audit and compliance
An independent point-in-time assessment, and where it differs from continuous posture management running inside the platform.
Cloud migration services
The wider platform work: architecture, migration and the estate this posture applies to.
Defender for Endpoint
The endpoint half of this same product family, which Defender for Servers builds on to protect servers.
Go and find your secure score. It costs nothing and an afternoon covers it.
Foundational CSPM costs nothing, reaches every subscription including the ones nobody remembers creating, and connects AWS and GCP without agents. Should that number come back badly, you will know exactly what to do next. Should it come back well, you have just saved yourself a purchase.
Related Services
Explore more solutions that work great with this service
Microsoft Defender for Servers
Defender for Servers engagements for US organizations: estate
Learn moreMicrosoft Defender for Storage
Defender for Storage deployment for US organizations: storage
Learn moreMicrosoft Defender for Containers
Container and Kubernetes security for US organizations using Defender
Learn moreMicrosoft Defender for SQL
Defender for SQL deployment for US organizations: the full SQL estate
Learn moreMicrosoft Defender for APIs
API security for US organizations using Defender for APIs: APIs
Learn moreMicrosoft Defender XDR Services
One incident queue across endpoint, email and identity
Learn more