Infrastructure & DevOps8 July 2025·7 min read

What Is Serverless? A Plain Guide for Business Owners

What is serverless, and is it cheaper than a rented server? See how AWS Lambda bills you, the free monthly allowance, and when a server is the better pick.

ServerlessAWS LambdaCloud HostingHosting CostsGoogle Cloud Run

Your developer says the new booking app will be "serverless", and the quote has a line for cloud costs that nobody can pin down. You want to know what is serverless in plain words, who pays for what, and whether the monthly bill will surprise you.

This guide answers that for a shop, clinic, cafe or small factory owner. It uses only AWS and Google's own pricing pages and docs, so you can check every number yourself.

Quick answer

Serverless means the cloud company runs and looks after the servers. Your developer only uploads the code.
You pay when the code runs, not for a computer sitting idle all month.
AWS Lambda has a free allowance of 1 million requests and 400,000 GB-seconds each month, as of October 2026.
It suits apps with uneven traffic: forms, payment alerts, bookings, reports.
A small rented server can be simpler and steadier for apps that are busy all day or keep a live connection open.

What is serverless, in plain words?

Serverless is a way of running your app where the cloud company owns, patches and scales the servers for you. Servers still exist. You just never rent one by the month, and you pay only for the time your code actually runs.

AWS describes it as "an abstraction of virtualized computing" where "the cloud provider manages the server resources your application runs on". In normal words: someone else worries about the machine.

What a "server" means here

A server is a computer in a data centre that stays on all day to answer your customers. With normal hosting (the rented computer your website lives on), you pay for it every hour, even at 3 a.m. when nobody visits.

What changes with serverless

With serverless, your code sleeps until a request comes in. A request is one action by a user, such as a form being sent or a payment alert arriving. The cloud wakes the code, runs it, and charges you for those few milliseconds.

Serverless vs a rented server: which fits your business?

A rented server is a fixed monthly cost for a computer that is always on. Serverless is a variable cost that follows how much people use your app. For low or uneven traffic, serverless is often cheaper; for steady all-day load, a server can be simpler to plan.

PointServerless (for example AWS Lambda)Rented server (for example AWS Lightsail)
How you payPer request and per millisecond of run timeFixed price per month
Idle timeFreeStill billed
Busy festival dayScales up by itselfCan slow down if the plan is too small
Who patches the systemThe cloud companyYour developer
Time limit per task15 minutes on LambdaNo such limit
Smallest AWS price, Oct 2026Free up to the monthly allowanceUS$5 a month for the smallest Lightsail plan

When serverless makes sense

Enquiry and contact forms that get a few hundred sends a day.
Payment alerts from a gateway that arrive at random times.
A nightly job that builds a sales report or sends reminder messages.
Making small copies of product photos when staff upload them.

When a normal server is the better pick

Work that runs longer than 15 minutes in one go, such as a big video or a large data import.
An app that is busy every minute of the day, so you never get the benefit of idle time being free.
An app that keeps a live connection open all the time, like a kitchen screen. Ask your developer how they will handle that before you pick serverless.

How does AWS Lambda pricing work?

AWS Lambda bills you two things: the number of requests, and the run time multiplied by the memory you give the code. In the Mumbai region, as of October 2026, that is US$0.20 per million requests and about US$0.0000167 per GB-second. A monthly free allowance comes off first.

The two numbers on your bill

  1. Requests: every time the code is started, it counts as one request. AWS charges US$0.20 per million.
  2. Run time: AWS measures how long the code runs, rounded up to the nearest millisecond. That time is multiplied by the memory you chose, from 128 MB up to 10,240 MB.
  3. The result is in GB-seconds. One GB-second is code using 1 GB of memory for one second.
  4. The free allowance of 1 million requests and 400,000 GB-seconds a month comes off first.

Two worked examples (Mumbai region, October 2026)

These use AWS's own published rates. They cover the Lambda part only.

ExampleRequests a monthRun time and memoryLambda bill
Clinic booking app3 lakh0.2 seconds, 512 MBUS$0, inside the free allowance
Busy shop app50 lakh0.3 seconds, 1 GBAbout US$19 a month

For the second example: 50 lakh requests at 0.3 seconds and 1 GB is 15 lakh GB-seconds. Take off the free 4 lakh, and 11 lakh GB-seconds cost about US$18.33. The 40 lakh paid requests add US$0.80.

The costs that sit outside Lambda

AWS's pricing page warns that using other AWS services, or sending data out, can add charges. So the database, file storage and data sent out to users are billed on their own. Ask your developer to list these as separate lines in the quote.

If you are planning a booking, billing or order app, ask for the hosting cost line by line before you sign. To have that worked out for your own app, see The Beyond Horizon's custom software development service for Indian businesses.

What about the new AWS Free Tier?

AWS changed its Free Tier for new accounts in July 2025. New customers now get US$100 in credits at sign-up and can earn up to US$100 more. The free account plan ends after six months or when the credits run out, whichever comes first.

Over 30 services, Lambda among them, still have an always-free monthly allowance. A practical tip: open the AWS account in your company's name and with your own email ID, then add the developer as a user. If the developer leaves, the account and the bills stay with you.

What is serverless outside AWS?

The same idea exists on other clouds. Google Cloud has Cloud Run, which Google calls "a fully managed application platform" for running code. Google's docs say that if there are no incoming requests, "even the last remaining instance will be removed", and you are not charged while it sits unused.

Cloudflare Workers and AWS Amplify are other common choices. The idea is the same everywhere: no server to look after, and the bill follows use.

What we have built on serverless platforms

Here are a few real projects from our own work.

Aekshea Foods, a food wholesaler in Salzburg, runs its trade ordering site on Cloudflare Workers. Its hosting cost is close to zero a month on a free tier for commercial use.
HRDYAM, a platform for pilgrimage tours, astrology consultations and a products store, is deployed on AWS Amplify.
Prodeazy, a production and stock system for Bang Diamond Tools, runs on Google Cloud Run and Vercel.
Warm White Entertainment's music drop for 200 numbered NFC objects runs on Google Cloud Run.

What can go wrong with serverless?

The two common problems are a slow first request and a bill nobody is watching. Both are easy to manage once you know about them, and neither is a reason to avoid serverless for a small business app.

The slow first request

When code has not run for a while, the cloud has to start it fresh. AWS calls this a "cold start". AWS says cold starts typically happen in under 1% of calls and take from under 100 ms to over 1 second.

The bill nobody watches

Pay-per-use also means a bug that calls the code in a loop can run up a bill. Ask your developer to set a monthly budget alert in AWS Budgets on day one.

It can email you when spending passes, or is forecast to pass, the amount you pick. AWS notes there can be a delay between a charge and the alert.

Questions to ask your developer before you agree

  1. Which parts of the app will be serverless, and which will run on a normal server?
  2. What will the monthly cloud bill be at today's traffic, and at five times today's traffic?
  3. Which costs sit outside the compute bill: database, storage, data transfer?
  4. Is the cloud account in my company's name, with my email as the owner?
  5. Is a budget alert set, and who gets the email?

Getting the setup right

Once you know what is serverless and what a server costs, the choice comes down to your traffic. Serverless is a good fit for many small business apps, as long as someone has done the maths for your numbers. If you want a second opinion on a cloud quote or a plan for a new app, call or WhatsApp The Beyond Horizon on +91 75973 92744.

Frequently Asked Questions

Why is it called serverless?

Servers still run your code, but you never rent, set up or patch one. The cloud company manages them, so from your side there is no server to look after.

What is serverless on AWS?

On AWS, serverless usually means AWS Lambda, which runs your code only when a request comes in. You pay per request and per millisecond of run time, with 1 million requests and 400,000 GB-seconds free each month as of October 2026.

Is serverless SaaS or PaaS?

Neither, exactly. AWS describes serverless as an application development model, and services like Lambda are called Function as a Service, where you upload small pieces of code and the cloud runs them on demand.

What is serverless vs cloud?

Serverless is one way of using the cloud. With other cloud options you rent a virtual server by the month and manage it yourself, while with serverless the provider manages the servers and bills you only for actual use.

The Beyond Horizon Team

A software studio based in India. We build websites, mobile apps and custom software for businesses, and write plain guides on what we learn.

// SEE IT IN PRODUCTION

Related Case Studies

Related Services

Tags

ServerlessAWS LambdaCloud HostingHosting CostsGoogle Cloud Run

Build something like this?

We turn engineering depth into working products. Let's talk about yours.

Hire our engineering team

Have a Project in Mind?

We build fast, SEO-ready web and mobile applications.