The AWS Free Tier changed for new customers in 2025. Under the current offer, an eligible new AWS customer can open a Free account plan for up to six months, receive $100 in AWS credits at sign-up and earn up to another $100 by completing qualifying activities. The Free plan expires after six months or when credits are exhausted, whichever occurs first. This is different from older instructions claiming that every new account receives twelve months of the same free allowances.
Always Free offerings are a separate part of AWS Free Tier: selected services have ongoing monthly no-charge allowances subject to defined usage limits. A Free plan is not a promise that every AWS service or configuration is available, and a Paid plan can generate normal charges. Anyone using a tutorial should verify the account’s actual plan and service eligibility before deploying resources.
Understand Free and Paid account plans
AWS’s current Free account plan is intended for learning and proof-of-concept projects. It restricts access to some products and features and, according to AWS, does not ordinarily incur charges while active. A learner must not interpret that protection as a production hosting guarantee or unlimited free access to every service. The plan ends when the credits run out or the six-month limit passes.
A Paid account plan has broader service access and standard pay-as-you-go billing for usage not covered by applicable credits or free allowances. Before upgrading, inspect persistent resources such as virtual machines, managed databases, retained log groups and storage. Determine which resources will continue running and what they may cost. An old course’s assumption about a universal Free Tier should not dictate a billing decision.
Eligibility matters. AWS states that the introductory Free plan and credits are for new customers, and people who previously had an AWS account can be ineligible. Existing AWS customers should check service-specific Always Free offers and any credits already granted to them rather than assume that creating another account reproduces the new-customer terms.
How AWS credits work
New eligible accounts generally receive $100 in introductory credits. Additional credits, up to another $100, depend on completing qualifying activities displayed by AWS. Thus the advertised $200 is a maximum potential credit amount, not a cash payment or a credit balance automatically given to every user. Credits are associated with the account and cannot be transferred to another account.
AWS documents a credit-expiration date in the account’s current billing or settings experience. Review that date and the remaining credit balance rather than relying on a remembered expiration rule from a tutorial written before the account-plan redesign. A credit can cover eligible usage, but some service configurations or plan-restricted features may not be available on Free accounts at all.
Even when credits are present, every planned activity deserves a resource inventory. An experiment that uses a virtual machine, an S3 bucket, a managed database, logging and data transfer involves distinct sources of usage. Understanding which component uses the credits is more useful than treating the balance as permission to launch an unreviewed infrastructure estate.
What Always Free means in practice
AWS offers more than thirty services with specified ongoing free-usage allowances. The individual allowance can relate to storage volume, API requests, execution time or another meter. Free usage within one service does not cancel charges generated by a second service, and going beyond an allowance can consume remaining credits or result in charges on a Paid plan.
Service availability and usage limits change, so check the Free Tier offering associated with the specific product, account type and Region. A test service that produces free compute invocations may still write data to storage or generate transfer and logging usage. AWS’s service catalog is a collection of separate pricing models, not a single unlimited no-cost subscription.
A learner should write down expected monthly resources before creating them: compute hours, capacity, database size, requests, inbound and outbound data, log retention and any snapshot or backup copies. This makes it possible to identify which service is responsible if usage differs from the estimate. The intention is to learn cloud operations without being surprised by the operational consequences.
Build one bounded S3 learning project
Begin with a small exercise containing only synthetic data. Create or use one private S3 bucket and define a limited object prefix where a test application may upload its own files. Identify the Region, encryption expectations, public-access settings, object naming convention and how the test objects will be deleted at the end. Do not grant public read access merely to make a demonstration easier.
Amazon S3 and Amazon EC2 solve different problems. S3 manages objects; EC2 runs virtual machines that need lifecycle and patch management. The difference between EC2 and S3 matters when selecting the simplest tool for a file-handling exercise. A persistent server is often unnecessary when the learning goal is to understand object storage permissions.
After a successful upload, test a second operation that should be denied, such as writing to a different prefix with an intentionally restricted role. Record the expected error, inspect relevant logging and verify that the original operation remains authorized. Clean up both the test objects and any temporary identities. The result should demonstrate what the access rule means, not just that the console displayed a green notification.
Security applies to temporary accounts
A Free Tier account is still a real AWS account. Its root user should be protected with strong authentication, and programmatic access should use appropriately scoped roles or other supported short-term credentials where possible. Never place access keys in source repositories, browser-side scripts, screenshots or shared chat logs. Use disposable synthetic content instead of employer or customer data.
AWS identity and access management helps separate a person’s administrative permissions from the application’s permissions. A process that needs to upload one object should not automatically be allowed to delete all buckets, and a learner should test both successful and forbidden operations. Security is proved through expected behavior under controlled conditions.
Logs, encryption and retention also require consideration. An account that is temporary may still collect sensitive information in filenames, object tags or logs. Set appropriate cleanup policies and periodically review what remains. Temporary projects are useful for learning disciplined operations precisely because they include a safe teardown procedure.
Monitoring and controlling expenses
Use the management console’s cost and usage information to view the Free plan’s remaining days and credits. AWS also sends alerts associated with the credit balance and plan ending. These alerts should supplement, not replace, a written list of resources and an independent check of what has been deleted.
On a Paid plan, budget alerts and usage reports help reveal unexpected spending but generally should not be assumed to be instantaneous hard caps. Charges may be generated before a notification reaches the owner. The AWS cost-optimization guide offers useful context for idle compute, storage lifecycles and unneeded resources, provided the recommendations are applied to the actual pricing model.
When a project is finished, verify deletion across compute, volumes, snapshots, objects, data-transfer settings and log groups. A stopped compute instance may still have billable storage or attached resources on a Paid plan. A good cloud operations habit is to check the resulting environment, not assume that deleting the most visible application icon removed every dependency.
What happens when the Free plan ends
AWS says that once the Free plan expires, the account and its resources become inaccessible until the user upgrades. AWS currently retains data for 90 days after expiry, during which an upgrade can restore access; if the account is not upgraded within that window, the account content is permanently deleted. This is a serious constraint for anyone using the Free plan to hold a portfolio or unique datasets.
Export any work needed for learning or professional documentation before the plan ends. Keep a local copy of source code, synthetic datasets and written notes, and document how the environment could be recreated. Do not place unique business records or a sole backup in a temporary account whose lifecycle ends automatically.
AWS’s Free plan FAQ and official billing documentation control the details. The correct next action near expiry is to review whether the project should be deleted, exported or deliberately upgraded to a Paid plan with the resulting costs understood. No educational resource should pressure a learner into an unexpected subscription merely to preserve a sample exercise.
Relate lab work to current certification skills
Older versions of this page mixed Free Tier instructions with the AWS Certified SysOps Administrator exam. AWS has since updated that certification family, and the current catalog includes AWS Certified CloudOps Engineer – Associate. A study plan should use the active issuer objective guide for the chosen credential rather than rely on retired SOA-C02 instructions or historical Free Tier policies.
Operations concepts such as identity, logging, networking, capacity and recovery are still useful to practice. A small AWS Lambda versus EC2 test can show why an event-driven task behaves differently from an always-running virtual server. Choose a service that the account actually permits and measure the application rather than assume the free allowance covers every combination of features.
The official AWS Free Tier FAQ and AWS Billing Free Tier documentation explain current plan eligibility, service availability and credit rules. In 2026 the most useful Free Tier outcome is not the number of services created; it is a tested, secure, cost-aware project that can be shut down and rebuilt without lost data.