NitronEdge TechnologiesHelping Enterprises Harness Innovation
Cloud · AWS · Buyer guide

AWS pricing explained: what really drives the cost

How AWS pricing works, what moves the bill (compute, region, data transfer, support, AI tokens) and how to choose an AWS partner, with a buyer checklist.

AWS pricing is mostly pay as you go: you pay for the compute, storage, data transfer and AI tokens you use, billed monthly, with no licence fee to start. The bill comes down through commitments such as Savings Plans, volume tiers and credits, and goes up with support plans, data transfer out and the region you run in. For a real number, model your workload in the AWS Pricing Calculator.

That is the short answer. The longer one matters, because two companies running the same application on AWS can see very different bills. The difference is rarely the list price. It is architecture, commitment and discipline.

How AWS charges for its services

AWS describes four pricing approaches on its official pricing page:

  • Pay as you go. Most services charge per unit used: per second or hour of compute, per gigabyte stored per month, per request, per million tokens.
  • Save when you commit. Savings Plans give a lower rate in exchange for a commitment, measured in dollars per hour, over one or three years.
  • Pay less by using more. Some services, storage in particular, get cheaper per unit as volume grows.
  • Flat rate. A smaller number of offers have fixed pricing.

Data transfer into AWS is free. Data transfer out to the internet is charged, with the first 100 GB a month free, aggregated across services and regions (except China and GovCloud).

There is no single "AWS price". Every service has its own price list, and each varies by region.

What affects the cost

These are the levers that move an AWS bill, roughly in order of impact for a typical enterprise workload.

Compute choices. Instance family, size and how many hours it runs. An always-on server sized for peak load is the most common source of waste. Autoscaling, scheduled shutdown of non-production environments and serverless options all cut this.

Commitment. According to AWS, Compute Savings Plans can reduce costs by up to 66% and EC2 Instance Savings Plans by up to 72%, against on-demand rates, for a one or three year commitment. The catch is that you pay the commitment whether you use it or not, so commit only to the steady baseline you are sure of.

Region. The same service costs different amounts in different regions. Running in Middle East (UAE), Middle East (Bahrain), Asia Pacific (Mumbai or Hyderabad), Africa (Cape Town) or Canada (Central) may be priced differently from US East. Data residency rules often decide the region for you, so price the region you are allowed to use, not the cheapest one.

Data transfer. Moving data out to users, between regions or between Availability Zones adds up. Architectures that chat across zones or copy data between regions for backup can produce a transfer line that surprises finance.

Storage tiers. Keeping logs and backups in a high-performance tier when an archive tier would do is a slow, steady leak.

Generative AI usage. Amazon Bedrock bills mainly per token. The official Bedrock pricing page lists Standard, Flex, Priority and Reserved tiers: Flex is priced at a 50% discount to Standard, Priority at a 75% premium, and batch inference is 50% cheaper than on-demand for selected models. Model choice, prompt length and caching often matter more than any discount.

Support plan. AWS support is charged as the greater of a minimum fee or a tiered percentage of your monthly AWS charges. On the AWS support pricing page, checked on 22 November 2026, Business Support+ starts at $29 a month per account, Enterprise Support at $5,000 a month and Unified Operations at $50,000 a month. Basic Support is included for everyone. For a large estate the percentage, not the minimum, is what you pay.

Credits and free tier. New accounts can get up to $200 in credits over six months under the current AWS Free Tier, and more than 30 services have always-free monthly allowances. Useful for a pilot, irrelevant at production scale.

Implementation and running costs. None of the above includes the people who design, migrate, secure and operate the environment. These costs sit outside the AWS bill but usually decide whether the bill stays under control.

How AWS pricing applies in the Middle East, India and Africa

For regulated sectors in the Gulf, the region choice usually comes first. Banks and government entities in the UAE often need data held in the country, which points to the Middle East (UAE) region. AWS has also announced a region in Saudi Arabia, which matters for organisations working under the Saudi PDPL and sector rules. Until it is live, check which services and instance types are available in the region you plan to use, because not every service launches everywhere on day one.

In India, the Mumbai and Hyderabad regions give two in-country options, which helps with resilience designs under the DPDP Act and RBI expectations for regulated entities. In Africa, Cape Town is the in-continent region. Organisations elsewhere on the continent should compare it with other regions on latency and data protection grounds.

Billing currency, tax treatment and payment terms can also vary by country and by whether you buy direct from AWS or through a partner. Ask about this early, because it affects how finance compares quotes.

A checklist for choosing an AWS partner

If you are buying AWS through a partner, or hiring one to design and run your environment, ask these questions:

  1. Do they price your workload with the AWS Pricing Calculator, line by line, and share the model? A single total with no breakdown tells you nothing.
  2. How do they size Savings Plans? Look for a plan based on your actual usage history, with headroom, not a target discount.
  3. What tagging and cost reporting do they set up on day one? Without tags by team and application, nobody can own a cost.
  4. Which region do they recommend, and why? The answer should cover your regulator and data classification as well as latency.
  5. How do they estimate generative AI costs? They should talk about tokens per request, request volume and caching, as well as the model's list price.
  6. What happens after go-live? A monthly cost review with named actions is worth more than a one-off optimisation exercise.

How to get started

  1. List the workloads you plan to move or build, with rough compute, storage and data transfer for each.
  2. Model them in the AWS Pricing Calculator for the region you are allowed to use.
  3. Run a short pilot on credits or a small budget to replace estimates with real usage.
  4. After two or three months of steady usage, decide on Savings Plans for the baseline.
  5. Set budgets and alerts with AWS Budgets, track spend in Cost Explorer, and review both monthly.

For help designing an AWS environment for AI, data or security workloads, see our AWS partner page. If your plans include agents built on Amazon Bedrock, our agentic AI services cover design, security and rollout.

Is AWS cheaper than running servers on premises?

It depends on how steady and how well sized the workload is. AWS removes hardware purchase and data centre costs, and suits variable or growing workloads well. A large, flat, always-on workload with no right-sizing can cost more in the cloud. Model both with real usage before deciding.

Can AWS pricing be negotiated?

Published rates apply to everyone, but large commitments can qualify for private pricing agreements with AWS, and Savings Plans and volume tiers apply automatically. The AWS pricing page also has a route to request a pricing quote.

Sources

  1. AWS Pricing (Amazon Web Services) (2026-11-22)
  2. Compute Savings Plans pricing (AWS) (2026-11-22)
  3. AWS Support plan pricing (AWS) (2026-11-22)
  4. Amazon Bedrock pricing (AWS) (2026-11-22)
  5. AWS Free Tier (AWS) (2026-11-22)
  6. Amazon EC2 On-Demand pricing, data transfer (AWS) (2026-11-22)
Related: AWS
Enquire

Get pricing for AWS

Get a free consultation from our AI, data and security specialists. Honest advice, clear options and a reply within 6 hours.

Reply timeOur team replies within 6 hours

Request a callback

Free consultation. Our team replies within 6 hours.

Your details are safe. Our team replies within 6 hours. By sending, you agree to our privacy notice.

WhatsApp