Tech Stack

An app or a website never consists of a single piece. Behind it lies a whole series of building blocks: the language in which the code is written, storage for users' data, a computer in a data center on which everything runs. The sum of these building blocks is called a tech stack, or technology stack. The word stack fits well, because the parts build on top of each other. At the bottom sits the hardware, above it the software that controls it, and at the very top what you see on the screen. When a company says it’s switching its tech stack, it means a restructuring within this stack.

Why the choice of building blocks locks you in for years

A tech stack is a decision with long-lasting consequences. Once a team has written hundreds of thousands of lines of code in a particular language, it doesn’t simply throw them away. A switch costs months of work during which no new features get built. Experts therefore speak of technical debt: legacy burdens that must later be laboriously worked off.

The stack also determines who a company can hire. For widespread languages like Python or JavaScript, developers are easy to find. For exotic building blocks, the search takes longer and costs more. Then there are the costs: some building blocks are free and openly accessible, others are billed per user or per month.

For investors, the term is interesting for another reason. When a large corporation switches its stack to a provider’s cloud services, that means years of revenue for that provider. Such decisions therefore regularly appear in quarterly reports and stock market announcements.

From the screen to the data center

The stack is usually divided into two halves. The frontend is everything the user directly sees and clicks on: buttons, menus, forms. The backend runs invisibly on remote servers and does the actual work, such as checking passwords or saving orders. Both halves communicate with each other via fixed interfaces, meaning agreed-upon formats for requests and responses.

Beneath the backend lie further layers. A database manages the stored information and retrieves it quickly. Below that lies the infrastructure: rented computers in a data center, often from Amazon, Microsoft, or Google. Added to this are tools that test the code and automatically deploy it to these computers.

With AI products, an additional layer comes into play. At the very bottom are special graphics chips that handle the many computational steps of a model. Above that sits the model itself, often purchased from an external provider. And above that, the programs that send requests to the model and process the responses. This is precisely why the term AI stack has been heard so frequently in recent years.

The term in job postings and company announcements

The word is most commonly encountered in job postings. There you’ll find a list like React, Python, PostgreSQL, AWS. These aren’t tasks, but the building blocks you would work with. Applicants can use this to gauge whether their own skills are a good fit. Job titles like full-stack developer mean that someone masters the entire stack, from the screen to the database.

In business news, the term comes up during restructurings and acquisitions. When a corporation buys a smaller software company, a common question is how well the two stacks fit together. If they fit poorly, the merger takes years. Security incidents are also often explained this way: a vulnerability was located in a single building block that many companies used in their stack.

A common misconception is that a tech stack is something purely technical. In practice, the term is also used for tools beyond programming, such as software for marketing or accounting. What’s always meant is the same principle: a grown collection of tools that must work together.