AWS vs Azure vs GCP 2026: Which Cloud to Learn First, by the Numbers

People ask me some version of the same question every month: "I want to get into cloud, should I learn AWS, Azure or GCP first?" I have shipped production systems on all three over fourteen years, and my honest first answer used to be "it doesn't matter much, the concepts transfer". That is still mostly true, but it is lazy advice when real numbers exist. So this post puts AWS vs Azure vs GCP side by side for someone choosing which cloud to learn in 2026: market share from Synergy Research, what US job postings ask for, what developers actually use, the free tiers (AWS rebuilt its own in 2025), the entry certifications, what each AI platform is called this year, and a service mapping table so that learning one makes the other two cheaper.

AWS vs Azure vs GCP market share in 2026

Synergy Research Group is the source most people quote for cloud infrastructure share, and its Q2 2026 release, published on 30 July 2026, puts Amazon at 28%, Microsoft at 20% and Google at 15% of a quarterly market worth about $143.4 billion, growing 43% year over year, the fastest rate in eight years. Synergy's chief analyst attributes the acceleration to generative AI, and says GenAI-specific cloud services grew 165%. Two years earlier, in Q2 2024, the split was 32%, 23% and 12% of a $79 billion quarter.

Read those carefully. AWS is still the largest by a clear margin, but it has lost about four points of share while the market nearly doubled, so its absolute revenue still grew a lot. Google is the one gaining share. Microsoft's figure in Synergy's data is also much smaller than the "Microsoft Cloud" revenue Microsoft reports, because Synergy measures cloud infrastructure services, not SaaS such as Microsoft 365. Market share tells you where the money is. It does not tell you where the jobs near you are, which is a different measurement.

Which cloud has the most jobs?

The best recent dataset I found is an Oxylabs analysis of roughly 850,000 US tech job postings from January 2025 to March 2026, built on Coresignal's jobs database. AWS appeared in 30% of all postings, Azure in 24% and GCP in 14%; nearly 42% mentioned at least one of the three, and 15% mentioned both AWS and Azure. Split by role, DevOps and SRE postings asked for AWS 58% of the time and Azure 45%. The closest races were cybersecurity (AWS 39%, Azure 35%, GCP 20%) and data engineering (AWS 47%, Azure 42%, GCP 24%). The authors are clear about limitations: it is a keyword-matched sample, US only, and some companies are overrepresented.

Developer usage tells a similar story. The Stack Overflow 2026 Developer Survey, released on 6 October 2026 with 30,903 responses, has AWS at 41.8% of respondents who answered the cloud question, Azure at 25.1%, and Google Cloud at 24.5%, with Cloudflare (24.6%) now level with the second tier, according to Virtualization Review's write-up of the data. In the "admired" column, which measures whether current users want to keep using a platform, the three are almost tied: 45.8%, 43.2% and 42.6%.

Measure	AWS	Azure	Google Cloud	Source
Cloud infrastructure share, Q2 2024	32%	23%	12%	Synergy Research
Cloud infrastructure share, Q2 2026	28%	20%	15%	Synergy Research
Share of US tech job postings, Jan 2025 to Mar 2026	30%	24%	14%	Oxylabs
DevOps and SRE postings mentioning it	58%	45%	not reported	Oxylabs
Developers who used it in the past year (2026)	41.8%	25.1%	24.5%	Stack Overflow via Virtualization Review
Users who want to keep using it (2026)	45.8%	43.2%	42.6%	Stack Overflow via Virtualization Review

Is AWS still the best cloud to learn first?

For most people, yes, and the reason is boring: AWS leads every measure above, and the concepts it forces you to learn early (IAM policies, VPCs and subnets, regions and availability zones, the shared responsibility model) are the hardest to pick up later and the easiest to carry to the other two. AWS's console and naming are also the least friendly of the three. That sounds like a reason to avoid it. In practice it means that once you can read an IAM policy and reason about a security group, Azure RBAC and GCP's project-based IAM feel tidy rather than alien.

There are clear exceptions, and they matter more than the averages:

• You work in, or want to work in, a Microsoft shop. If your target employers run Entra ID, Microsoft 365 and Windows Server, Azure is the more direct route, and the job data shows Azure close behind AWS in data and security roles. Identity is where Azure is genuinely different: Entra ID sits under everything, and learning it is half the job.
• You want data, analytics or ML engineering. BigQuery, GKE and Google's AI stack are where GCP is strongest, and its 15% share is the one rising. GCP's project model and its console are also the most pleasant for a beginner, which counts for something when you are learning alone at night.
• Your local market says otherwise. The Oxylabs data is US-only. Spend twenty minutes searching job boards in your city for each platform name before you commit three months to one of them.

AWS vs Azure vs GCP services: the mapping table

Learning a second cloud is mostly learning new names for the same ideas. Google publishes an official AWS, Azure and Google Cloud service comparison, which I used for the core rows below. The AI row reflects 2026 renames: Azure AI Foundry became Microsoft Foundry around Ignite 2025 (the name used throughout Microsoft's Ignite announcements), and Google folded Vertex AI into the Gemini Enterprise Agent Platform in April 2026, saying all Vertex AI services will be delivered through it rather than as a standalone product. If a tutorial from 2025 says "Vertex AI" or "Azure AI Foundry", it is the same product family under its old name.

What it does	AWS	Azure	Google Cloud
Virtual machines	EC2	Virtual Machines	Compute Engine
Object storage	S3	Blob Storage	Cloud Storage
Serverless functions	Lambda	Functions	Cloud Run functions
Managed Kubernetes	EKS	AKS	GKE
Managed relational database	RDS, Aurora	Azure SQL Database, Azure Database for PostgreSQL/MySQL	Cloud SQL
Identity and permissions	IAM, IAM Identity Center	Entra ID with Azure RBAC	Cloud IAM, Cloud Identity
Private networking	VPC	Virtual Network (VNet)	VPC (global)
Metrics and logs	CloudWatch	Azure Monitor	Cloud Monitoring, Cloud Logging
Managed foundation models and agents	Amazon Bedrock	Microsoft Foundry	Gemini Enterprise Agent Platform

The three structural differences worth learning early are not in the table. AWS isolates things by account, Azure by subscription and resource group, and GCP by project. GCP's VPC is global while AWS and Azure networks are regional. And Azure's identity plane is a separate product (Entra ID) that also runs your company's email login, which is why Azure administration jobs often overlap with Microsoft 365 work.

Here is the same first exercise (a private storage bucket in London) on all three, which is a decent way to feel the difference in mental models. The CLI versions are the ones current at the time of writing.

# AWS CLI 2.39.3: buckets are global names, created in one Region.
# Block Public Access is on by default for new buckets.
aws s3api create-bucket \
  --bucket ajayk-learn-aws-0001 \
  --region eu-west-2 \
  --create-bucket-configuration LocationConstraint=eu-west-2

# Azure CLI 2.91.0: everything lives in a resource group first,
# and blobs live inside a storage account.
az group create --name rg-learn --location uksouth
az storage account create \
  --name ajayklearnaz0001 \
  --resource-group rg-learn \
  --location uksouth \
  --sku Standard_LRS \
  --allow-blob-public-access false

# Google Cloud CLI 591.0.0: the bucket belongs to your active project.
gcloud storage buckets create gs://ajayk-learn-gcp-0001 \
  --location=europe-west2 \
  --uniform-bucket-level-access \
  --public-access-prevention

Free tiers compared: AWS credits, Azure $200, GCP $300

AWS rebuilt its free tier on 15 July 2025. New accounts now get $100 in credits at sign-up and can earn up to $100 more by completing five starter activities (launching an EC2 instance, an RDS database, a Lambda function, a Bedrock prompt, and setting up a budget, $20 each). You choose a Free plan or a Paid plan. According to the Billing User Guide, the Free plan cannot incur charges, closes after six months or when credits run out, and AWS keeps your content for 90 days so you can upgrade. That is the first time AWS has offered a beginner account that genuinely cannot run up a bill, and it removes the main reason I used to steer nervous learners elsewhere. I wrote up the secure first-hour setup for a new AWS free tier account separately.

Azure's free account gives $200 of credit for 30 days, 12 months of free monthly amounts on 20+ popular services, and 65+ always-free services; after 30 days you must move to pay-as-you-go or the account is disabled. Google Cloud's Free Trial is $300 over 90 days, with no GPUs, no quota increases and no Windows VMs during the trial, plus an always-free tier that includes one e2-micro VM a month in three US regions and two million Cloud Run requests.

For a learner, the deciding variable is time, not dollars. Thirty days of Azure credit goes fast if you study on weekends. GCP's 90 days and AWS's six months suit a part-time pace better.

Cloud certifications for beginners: which one first?

Each provider has a foundational exam aimed at people who have never run a workload. AWS charges $100 for foundational exams (Cloud Practitioner and AI Practitioner) and $150 for associate ones such as Solutions Architect Associate. Google's Cloud Digital Leader is $99, 90 minutes and 50 to 60 questions, valid for three years. Microsoft's AZ-900 Azure Fundamentals is a 45-minute exam priced by country, and Microsoft Learn's free modules cover the whole syllabus.

My view, after interviewing a lot of candidates: foundational certs help you learn the vocabulary and get past a CV filter for a first role, and that is all they do. Hiring managers for engineering roles look for an associate-level cert at most, and they look much harder at whether you have built something and can explain it. A public repository with Terraform that creates a VPC, a container service and a database, plus a README explaining your IAM choices, is worth more than a second foundational badge. The DevOps roadmap for 2026 goes through which certifications are worth the money at each stage.

AWS vs Azure vs GCP for AI work

If you are choosing a cloud because of AI, the platforms are converging on the same shape: a model catalogue, an agent runtime, retrieval, evaluation and governance, each sold as one console. Amazon Bedrock has been AWS's managed model service since 2023, and it is one of the five AWS starter activities. Microsoft's pitch for Foundry is model choice: its Ignite 2025 recap claims Azure is the only cloud offering both Claude and GPT frontier models. Google describes its Agent Platform as an evolution of Vertex AI, and the old Vertex AI documentation URL now redirects there.

The skill that transfers is not the model API, which is a few lines in any SDK. It is the platform work around it: least-privilege roles for the thing calling the model, private networking to the endpoint, cost alerts on token spend, logging every call. That is DevOps discipline applied to a new kind of workload, and it is identical in shape across all three clouds. If you learn workload identity federation properly on one provider, you understand how an agent should authenticate on the other two.

Should you learn multiple clouds?

Eventually, and less than you think. The Oxylabs data shows 15% of postings ask for both AWS and Azure, which is meaningful but not dominant. The practical path is depth in one provider until you can design, deploy, secure and debug a small production system on it, then breadth: a weekend each on the other two, using the mapping table, to learn their account model, identity and networking. Terraform or OpenTofu makes the second and third cloud cheaper still, because the workflow is identical and only the resource names change. Kubernetes helps too, since EKS, AKS and GKE all run the same API.

FAQ

Which cloud is easiest for beginners?

Google Cloud usually has the gentlest console and project model, and Azure feels familiar to people who already administer Microsoft 365. AWS is the steepest at first. The new AWS Free plan, which cannot incur charges, has made AWS much less risky to experiment on.

Is AWS still worth learning in 2026?

Yes. AWS still holds the largest share of cloud infrastructure spend (28% in Synergy's Q2 2026 data), appears in more US tech job postings than Azure or GCP, and is used by more developers in the Stack Overflow 2026 survey. Its share is slipping, but from a large lead in a fast-growing market.

Should I learn Azure or AWS first for a job?

Learn whichever your target employers use, which a job-board search in your city will tell you in twenty minutes. In the US data, AWS leads overall, but Azure is close in data engineering and security roles and dominant in Microsoft-centric enterprises. The core concepts transfer either way.

Do I need a cloud certification to get hired?

No, but a foundational cert helps with CV filters for a first role, and an associate-level cert signals real study. What gets you hired is a project you built, deployed and can explain, including why its permissions and network are set up the way they are.

If I were starting from zero this quarter, I would open an AWS account on the Free plan, set a budget and MFA in the first hour, and spend six weeks building one small, real system on it: a container behind a load balancer, a managed Postgres database, everything in Terraform, every permission least-privilege. Then I would sit the Cloud Practitioner or Solutions Architect Associate exam, rebuild the same system on GCP or Azure over a weekend using the mapping table, and write down what was different. That last note, more than the cert, is what I would bring to an interview.

---

Related: DevOps Roadmap 2026: Skills, Certifications and What AI Changed, Serverless in 2026: Lambda vs Cloud Run vs Workers, and When It Costs More and Workload Identity Federation Explained (2026): OIDC Instead of Cloud Keys.