4 Cloud Economics · AWS Cloud Practitioner

In this chapter you will learn how the economics of the cloud differ from the economics of owning a data center. You will compare capital and operational expenditure, fixed and variable costs, and the direct, indirect, and hidden parts of total cost of ownership (TCO). You will see how economies of scale, rightsizing, licensing choices, managed services, and automation reduce costs, how to build an estimate with the AWS Pricing Calculator, and how the AWS Cloud Value Framework measures value beyond cost. This chapter maps to Domain 1, Cloud Concepts, Task Statement 1.4: “Understand concepts of cloud economics.”

4.1 Two ways to pay for IT

Imagine two bakeries. The first buys its own delivery van for 40,000 dollars. It pays the full amount up front, and the van sits in the parking lot on quiet days. The second bakery uses a delivery service and pays a few dollars per delivery. On a slow Tuesday it pays almost nothing; during the holidays it pays a lot, but every dollar matches a sale. That is the core of cloud economics. Traditional IT buys the van. The cloud pays per delivery.

CapEx vs. OpEx

Capital expenditure (CapEx) is money spent up front on physical assets that the business will use for years: servers, storage arrays, network switches, data center buildings, cooling, and power systems. CapEx is paid before you get any value, and the asset is depreciated over its useful life, typically three to five years for servers. Operational expenditure (OpEx) is money spent on an ongoing basis for services and consumables. Cloud services are OpEx: you pay monthly for what you used last month, and you can stop paying when you stop using. Capacity and spending over time: on premises vs. cloud

Figure 4.1: Spending over time. On premises, money is spent in large steps ahead of demand (CapEx). In the cloud, spending follows actual usage (OpEx).

Figure 4.1 shows the practical effect. With on-premises hardware, capacity comes in large steps. You buy for the peak you

expect three years from now, so most of the time you pay for idle capacity (the gap between the step line and the demand line). If demand grows faster than predicted, you run out of capacity and lose customers while you wait weeks for new hardware. In the cloud, capacity follows the demand curve closely, so you neither waste money on idle servers nor turn customers away.

Fixed vs. variable costs

A fixed cost stays the same regardless of how much you use. A data center lease, a server you already bought, and a full-time employee who babysits that server are fixed costs. A variable cost rises and falls with usage. Cloud compute billed per second, storage billed per gigabyte-month, and data transfer billed per gigabyte are variable costs. Trading fixed costs for variable costs is one of the six advantages of cloud computing: “trade fixed expense for variable expense”. It matters most for businesses with unpredictable or seasonal demand, for start-ups that cannot afford big up-front purchases, and for experiments that may be abandoned.

Exam tip

When a question asks for a benefit of the cloud related to paying “only for what you use”, “no up-front investment”, or “no long-term commitment”, the answer is the trade of CapEx for OpEx, or fixed for variable expense.

Good to know

The cloud also offers ways to commit in exchange for discounts, such as Savings Plans and Reserved Instances, covered in the billing and pricing chapter (Chapter 15). These are still OpEx: you commit to a level of spend or usage, but you do not buy and own hardware.

4.2 Total cost of ownership (TCO)

A common mistake when comparing on-premises and cloud costs is to compare the price of a server with the price of an EC2 instance. That comparison leaves out most of the real cost. Total cost of ownership (TCO) counts everything it costs to run a workload over its lifetime.

Direct costs

Storage and network gear, software licenses,
support contracts, data center space,
power, cooling, connectivity

Indirect costs

Staff time: racking, patching, backups,
capacity planning. Facilities: security,
insurance, maintenance. Procurement,
vendor management, audits

Hidden costs

Idle capacity for rare peaks, downtime,
idle disaster recovery site, hardware
refresh every few years, opportunity
cost of slow launches

Figure 4.2: The TCO iceberg. Hardware is the visible part of on-premises cost; most costs sit below the waterline.

Direct costs

Direct costs are the ones that show up on purchase orders and are easy to attribute to IT:

Server, storage, and network hardware
Software licenses and support contracts
Data center space (rent or construction)
Power and cooling
Network connectivity
Indirect costs
Indirect costs are real but harder to attribute to a single workload:

Staff time for racking, cabling, patching, backups, capacity planning, and hardware refreshes Facilities overhead: security guards, fire suppression, insurance, and building maintenance Management overhead: procurement, vendor management, and audits

Hidden costs
Hidden costs often go uncounted entirely:
Idle capacity bought for peaks that rarely happen

Opportunity cost: engineers maintaining infrastructure instead of building products, and slow procurement delaying launches

Downtime caused by hardware failures and the revenue lost during it

Over-provisioning for disaster recovery: a second data center that sits mostly idle Technology refresh: replacing hardware every few years, plus the migration work each time

Real-world example

Real-world scenario: A logistics company compares a 10,000-dollar server against the yearly cost of an equivalent EC2 instance and concludes the server is cheaper. A fuller TCO analysis adds the share of data center rent, power, cooling, a backup appliance, a spare server for disaster recovery, a hardware support contract, and about 20 percent of a systems administrator’s time. Over four years the on-premises total is several times the hardware price, and the cloud option is cheaper, before even counting faster launches.

Tools for TCO and estimates

Migration Evaluator (introduced in Chapter 3) analyzes your actual on-premises usage and produces a business case that projects AWS costs for a migration.

AWS Pricing Calculator estimates the cost of AWS services for a planned architecture. It is free, needs no AWS account, and lets you share estimates with a link or export them.

Common trap

Older material mentions the “AWS TCO Calculator” and “Simple Monthly Calculator”. Both have been retired. For estimating the cost of new AWS resources, the answer is AWS Pricing Calculator. For a migration business case based on your current environment, it is Migration Evaluator.

4.3 Why the cloud can cost less

Economies of scale

AWS buys servers, storage, networking gear, and electricity in quantities no single company could match. It designs its own hardware, including custom chips such as AWS Graviton processors, and runs its data centers at very high utilization. Because hundreds of thousands of customers share this infrastructure, the cost per unit falls, and AWS passes savings on through lower prices. AWS has reduced prices many times since it launched. This is the cloud advantage called “benefit from massive economies of scale”.

Rightsizing

Rightsizing means matching the type and size of a resource to what the workload actually needs. On premises, servers are often bought with generous headroom “just in case”, and in the cloud teams sometimes copy those sizes. Monitoring usually shows that many instances run at 10 to 20 percent CPU.

Figure 4.3: A rightsizing example. An instance that peaks at 18 percent CPU is moved to a smaller size, cutting its cost in half while staying comfortably below capacity.

Because you can change instance size in minutes, rightsizing in the cloud is a routine task rather than a hardware project. AWS tools help:

AWS Compute Optimizer uses machine learning on CloudWatch metrics to recommend better sizes for EC2 instances, Auto Scaling groups, EBS volumes, Lambda functions, and more.

AWS Cost Explorer includes rightsizing recommendations for EC2.

AWS Trusted Advisor flags low-utilization EC2 instances and idle resources.

Exam tip

Rightsizing is usually the first and cheapest cost optimization to do, and it should happen before you buy Savings Plans or Reserved Instances. Committing to an oversized instance locks in waste.

Licensing strategies

Software licensing is often a large part of TCO, especially for databases and Windows workloads. AWS offers two broad models:

License included. The license cost is built into the hourly price of the service. For example, Amazon RDS for SQL Server or an EC2 instance running Windows Server includes the license. You need no license management, and you pay only while the resource runs.

Bring your own license (BYOL). You use licenses you already own. This avoids paying twice for software you have already bought.

Some licenses are tied to physical hardware, counted by sockets or cores. For these, Amazon EC2 Dedicated Hosts give you a physical server dedicated to your use, with visibility into the sockets and cores so you can stay compliant with server-bound licenses. AWS License Manager helps track license usage and enforce limits across AWS and on premises.

Common trap

Dedicated Hosts and Dedicated Instances both run on hardware dedicated to one customer. Only Dedicated Hosts give you visibility and control over the physical server’s sockets and cores, which is what per-socket or per-core BYOL licensing needs.

Managed services

Every server you run yourself needs patching, backups, monitoring, high availability, and scaling. Managed services such as Amazon RDS, Amazon DynamoDB, AWS Lambda, and AWS Fargate shift much of that undifferentiated work to AWS. The hourly price of a managed database may look higher than an EC2 instance running the same engine, but the TCO is often lower because:

Staff no longer spend hours on patching, backups, and failover setup.

High availability (for example, RDS Multi-AZ) is a setting rather than a project.

Serverless services scale to zero, so you pay nothing when idle.

This is the cloud advantage “stop spending money running and maintaining data centers”: focus on customers, not infrastructure.

Automation

Automation reduces cost in two ways. It cuts labor, because tasks that once needed a person run by themselves, and it cuts waste, because resources are created and removed exactly when needed. Examples:

Auto Scaling adds capacity during peaks and removes it afterward.

Scheduling stops development and test environments at night and on weekends, which can cut their compute cost by more than 60 percent.

Infrastructure as code with AWS CloudFormation creates identical environments on demand and deletes them completely when a project ends, so nothing is forgotten and left running.

Storage lifecycle rules move old S3 objects to cheaper storage classes automatically.

When the cloud does not save money

The cloud is not automatically cheaper. Organizations that rehost every server at its on-premises size, leave instances running around the clock, and never look at their bill are often surprised by the first invoices. Common causes include: Copying on-premises sizes. Servers sized for a five-year peak become oversized instances that run all month. Forgotten resources. Test instances, unattached EBS volumes, old snapshots, and idle load balancers keep billing until someone deletes them.

Ignoring pricing models. Steady workloads left on On-Demand pricing miss the large discounts that Savings Plans and Reserved Instances offer.

Data transfer surprises. Moving large amounts of data out to the internet or between Regions has a cost that is easy to overlook in an estimate.

No ownership. When nobody is accountable for a team’s spending, nobody optimizes it.

The fix is a habit, not a one-time project. Successful organizations tag resources by team and project, review costs regularly in Cost Explorer, set alerts with AWS Budgets, and treat cost as a design requirement alongside performance and security. The Well-Architected Framework captures this habit in its Cost Optimization pillar.

Real-world example

Real-world scenario: A media company rehosts 120 servers to EC2 and its first monthly bill is 30 percent higher than its data center budget. Over the next quarter it rightsizes 70 instances, schedules its development environments to stop at night, deletes 15 TB of orphaned snapshots, and covers its steady baseline with a Compute Savings Plan. Its monthly bill falls to well below the old data center cost, and the team can now launch new environments in minutes instead of weeks. Cost optimization levers and supporting AWS tools

Lower total

cost

1. Rightsize

Match size to real usage

Compute Optimizer

2. Scale with demand

Add and remove capacity

Auto Scaling

3. Right pricing model

Commit for steady use

Savings Plans, RIs, Spot

4. Licensing choices

BYOL or license included

Dedicated Hosts

5. Managed services

Less patching and admin work

RDS, Lambda, Fargate

6. Automate

Stop idle dev/test, rebuild

CloudFormation, schedules

7. Tier storage

Cold data to cheaper classes

S3 lifecycle rules

8. Measure and alert

See where money goes

Cost Explorer, Budgets

Rightsize before you commit: discounts on oversized resources lock in waste.

Figure 4.4: Key cost optimization levers in the cloud and the AWS tools that support each one.

4.4 Estimating with the AWS Pricing Calculator

The AWS Pricing Calculator is the tool for answering “what will this cost on AWS?” before you build. You add services, describe your configuration and expected usage, and the calculator shows monthly and up-front costs. You can group services, compare purchase options such as On-Demand versus Savings Plans, and save the estimate as a shareable link or export it.

Hands-on lab

Hands-on mini lab: build a quick estimate

1. Open the AWS Pricing Calculator in your browser (search for “AWS Pricing Calculator”). No sign-in is required.
2. Choose Create estimate.
3. Search for Amazon EC2 and choose Configure.
4. Pick a Region close to your users, a Linux operating system, and an instance with 2 vCPUs and 8 GiB of memory.
5. Set usage to 2 instances running all month. Note the On-Demand monthly cost.
6. Change the pricing model to a 1-year Compute Savings Plan and compare.
7. Choose Save and add service, then add Amazon S3 with 500 GB of S3 Standard storage.
8. Review the summary, then choose Share to get a link you could send to a manager.

Good to know

The Pricing Calculator estimates future costs. To analyze costs you have already incurred, use AWS Cost Explorer; to alert on spending, use AWS Budgets. The billing and pricing chapter covers these tools in depth.

4.5 The AWS Cloud Value Framework

Cost is only one part of the cloud’s value. The AWS Cloud Value Framework helps organizations build a business case across four pillars:

Pillar What it measures Example metric

Cost savings (TCO) Lower infrastructureand operating cost Infrastructure spend per customer Staff productivity People doing more valuable work Hours per week spent on patching Operational resilience Fewerand shorter outages, better security Unplanned downtime per year Businessagility Faster delivery of new ideas Time from ideato production

Some customers find the cloud saves modest amounts on infrastructure but delivers large gains in agility: a feature that took three months now ships in two weeks. Others find their biggest gain is resilience, because Multi-AZ designs reduce outages. A good business case measures all four.

Exam tip

If a question describes developers releasing features faster or experimenting cheaply, the pillar is business agility. If it describes administrators no longer racking servers and moving to higher-value work, it is staff productivity.

Chapter 4 summary

On-premises IT relies on capital expenditure (CapEx) and fixed costs paid ahead of demand; the cloud uses operational expenditure (OpEx) and variable costs that follow usage.

Capacity bought in steps leads to idle waste or shortages; cloud capacity scales with demand.

TCO includes direct costs (hardware, licenses, facilities), indirect costs (staff, overhead), and hidden costs (idle capacity, downtime, opportunity cost, refreshes).

AWS economies of scale lower unit costs, and AWS passes savings to customers through lower prices. Rightsizing matches resources to real usage and should precede commitment discounts; Compute Optimizer and Cost Explorer help.

License-included pricing bundles software cost; BYOL reuses existing licenses; Dedicated Hosts support socket- and core-based licenses.

Managed services and automation reduce labor and waste, often lowering TCO even when unit prices look higher. AWS Pricing Calculator estimates costs for planned architectures; Migration Evaluator builds a migration business case.

The Cloud Value Framework measures cost savings, staff productivity, operational resilience, and business agility.

Key terms
Term Definition
CapEx Up-front spending on physical assets thatare depreciated over time
OpEx Ongoing spending on services, paid as consumed
Fixed cost Cost that does not change with usage
Variable cost Cost that risesand falls with usage

TCO Total cost of ownership: all direct, indirect, and hidden costs overalifetime Economies of scale Lower unit costs from operating at very large volume

Rightsizing Matching resource typeand size to actual workload needs
BYOL Bring your own license: reuse licenses you already own
License included Software license cost built into the service price

Dedicated Host A physical EC2 server dedicated to you, with socketand core visibility AWS Pricing Calculator Free web tool to estimate the cost of AWS services

Cloud Value Framework Four pillars of cloud value: cost, productivity, resilience, agility Practice questions

1. A start-up wants to launch a new application without buying servers up front, and it expects usage to be unpredictable. Which cloud economic concept best describes the benefit?

A. Trading variable expense for fixed expense B. Trading capital expense for operational expense
C. Increasing depreciation of IT assets D. Locking in long-term hardware contracts

2. Which of the following are examples of capital expenditure in an on-premises data center? (Choose two.)

A. Buying storage arrays B. Paying a monthly AWS bill
C. Building a new server room D. Paying per gigabyte for data transfer
E. Paying per request for an API service

3. A company bought servers sized for the peak demand it expects in three years. Most of the time the servers run at 15 percent utilization. What does this illustrate?

A. Economies of scale B. Paying for idle capacity under a fixed-cost model
C. Variable cost pricing D. Rightsizing

4. A finance team is building a TCO comparison. Which item is an indirect cost of running on premises?

A. Purchase price of servers B. Software license fees
C. Staff time spent patching and replacing hardware D. EC2 instance hourly charges

5. Which costs are commonly hidden or overlooked in an on-premises TCO analysis? (Choose two.)

A. Opportunity cost of engineers maintaining

infrastructure

B. The invoice price of network switches
C. Idle capacity bought for rare peaks D. Monthly electricity bill for the data center
E. Annual software support contract

6. How do AWS economies of scale benefit customers?

A. Customers must commit to multi-year contracts B. AWS’s large purchasing volume and efficiency lead to lower prices for customers
C. Customers receive dedicated hardware at no extra cost D. Customers no longer pay for data transfer

7. A company has migrated to Amazon EC2 and plans to buy Savings Plans. Monitoring shows most instances run below 20 percent CPU. What should the company do first?

A. Buy 3-year Savings Plans immediately to maximize the

discount

B. Rightsize the instances, then purchase commitments
C. Move all instances to Dedicated Hosts D. Increase instance sizes to improve performance

8. Which AWS service uses machine learning to recommend optimal EC2 instance types and sizes based on utilization?

A. AWS Compute Optimizer B. AWS Pricing Calculator
C. AWS Budgets D. AWS License Manager

9. A company owns Microsoft SQL Server licenses that are counted per physical core and wants to use them on AWS while staying compliant. Which option should it use?

A. EC2 Spot Instances B. Amazon EC2 Dedicated Hosts
C. License-included Amazon RDS for SQL Server D. AWS Lambda

10. A company does not want to manage software licenses at all and wants license costs to stop when resources stop. Which licensing model fits?

A. Bring your own license (BYOL) B. License included
C. Dedicated Hosts D. Perpetual licenses purchased up front

11. A company moves its self-managed MySQL databases from EC2 to Amazon RDS. The RDS hourly price is slightly higher, yet the company’s TCO falls. What is the most likely reason?

A. RDS does not charge for storage B. Database administrators spend far less time on patching, backups, and failover
C. RDS includes free Reserved Instances D. RDS eliminates data transfer charges

12. A company wants to estimate the monthly cost of a new architecture with EC2, S3, and RDS before building anything. Which tool should it use?

A. AWS Cost Explorer B. AWS Budgets
C. AWS Pricing Calculator D. AWS Cost and Usage Report

13. Which action is an example of using automation to reduce cloud costs?

A. Buying hardware three years ahead of demand B. Scheduling development environments to stop outside business hours
C. Moving all workloads to the largest instance type D. Disabling Auto Scaling to keep capacity constant

14. After moving to AWS, a company’s developers release new features in days instead of months. Which AWS Cloud Value Framework pillar does this demonstrate?

A. Cost savings B. Staff productivity
C. Operational resilience D. Business agility

15. After migration, a company’s system administrators no longer rack servers or replace failed disks and instead work on automation projects. Which Cloud Value Framework pillar does this represent?

A. Staff productivity B. Business agility
C. Cost savings D. Operational resilience

16. A company wants a data-driven business case for moving its on-premises data center to AWS, based on its current server utilization. Which AWS offering is designed for this?

A. AWS Pricing Calculator B. Migration Evaluator
C. AWS Trusted Advisor D. AWS Cost Explorer

Answers and explanations

1. B. Avoiding up-front purchases and paying as you go is the trade of CapEx for OpEx. Option A reverses the cloud benefit (the cloud trades fixed for variable); C and D describe on-premises traits.
2. A, C. Storage arrays and building a server room are up-front purchases of long-lived assets. Monthly bills and per-use charges are operational expenses.
3. B. Capacity bought ahead of demand is a fixed cost that sits mostly idle. Rightsizing is the fix for this situation, not the description of it.
4. C. Staff time is real but harder to attribute to a workload, so it is an indirect cost. Server and license purchases are direct costs; EC2 charges are not an on-premises cost.
5. A, C. Opportunity cost and idle capacity are classic hidden costs. Invoices, electricity bills, and support contracts are visible direct costs.
6. B. AWS’s scale lowers its unit costs, and it passes savings on through lower prices. AWS does not require multi-year contracts, give free dedicated hardware, or waive data transfer charges.
7. B. Rightsize first so you do not commit to paying for oversized capacity. Buying commitments for underused instances locks in waste; bigger instances make it worse.
8. A. Compute Optimizer analyzes utilization metrics with machine learning to recommend sizes. The Pricing Calculator estimates costs; Budgets alerts on spend; License Manager tracks licenses.
9. B. Dedicated Hosts provide a physical server with visibility into sockets and cores, which per-core BYOL licensing requires. License-included RDS would mean paying for licenses again; Spot and Lambda do not provide that visibility.
10. B. License included bundles the license into the hourly price, so there is nothing to manage and no cost when stopped. BYOL and perpetual licenses require you to own and manage licenses.
11. B. Managed services reduce the labor for patching, backups, and high availability, lowering TCO even when unit prices are higher. RDS still charges for storage and data transfer and does not include free reservations.
12. C. The Pricing Calculator estimates costs for planned architectures. Cost Explorer, Budgets, and the Cost and Usage Report work with costs you have already incurred or are incurring.
13. B. Scheduling idle environments to stop is automation that eliminates waste. Disabling Auto Scaling and oversizing increase costs.
14. D. Faster delivery of new features is business agility. Staff productivity is about how people’s time is used; resilience is about uptime.
15. A. Freeing administrators from low-value tasks for higher-value work is staff productivity. Agility concerns speed of delivering new capabilities to the business.
16. B. Migration Evaluator collects on-premises usage data and projects AWS costs to build a migration business case. The Pricing Calculator estimates costs from configurations you enter by hand.