Definition
Serverless is a way of running software in the cloud where you write functions, small units of code that respond to an event, and the provider runs them on demand, scales them automatically and charges you only for the milliseconds they actually execute. There are still servers, of course; the point is that you never see them, provision them or pay for them while they sit idle. Services such as AWS Lambda, Google Cloud Functions and Cloudflare Workers popularised the model.
In a company, serverless shows up in the glue between systems: a function that resizes an uploaded image, processes a webhook from the payment provider, sends a notification or runs a nightly report. It is also used for whole applications with spiky or unpredictable traffic, because scaling from zero to thousands of requests and back costs nothing when idle. The trade-offs are limits on how long a function can run, a small start-up delay called a cold start, and code that is tied to one provider's way of doing things.
The misconception is that serverless is always cheaper. For workloads that run constantly at high volume, paying per execution can cost more than a steadily running server, and several companies have moved heavy workloads back to containers for that reason. The model works best for intermittent tasks and for teams who want to spend no time on infrastructure. In 2026, edge serverless platforms that run code close to the user are the fastest-growing part of the category.
In practice
A company receives a spike of 50,000 form submissions during a TV campaign and none the following week. A serverless function handles the spike automatically and the monthly bill is a few euros, where a permanently provisioned server would have cost far more and might have fallen over.
Why it matters
Serverless removes an entire class of infrastructure work from small teams, and for irregular workloads the cost is close to zero. Ask your team where it fits, and also where a constantly busy workload is quietly costing more than a plain server would.
Frequently asked questions
- What does serverless actually mean if there are still servers?
- It means the servers are invisible to you. The cloud provider owns, provisions, patches and scales them; you only supply code that runs when an event triggers it and pay for execution time. The name describes your experience as a customer, not the physical reality.
- When should you not use serverless?
- Avoid it for long-running processes, for workloads that run at high constant volume where per-execution pricing exceeds a fixed server, for applications that need very low and consistent latency, and where portability between cloud providers matters. Containers or a plain PaaS are usually better fits there.