Return on Investment
Return on Investment, or ROI for short, measures how much profit an investment generates relative to the money put in. It is expressed as a percentage and makes very different types of expenditure comparable with one another.
Anyone who spends money in order to earn more money later wants to know whether it was worth it. This is exactly the question that Return on Investment, usually abbreviated as ROI, answers. You calculate the profit from an expenditure against the expenditure itself and get a percentage value. An example: You buy goods for 1000 euros and sell them for 1200 euros. The profit is 200 euros, so the ROI is 20 percent. The term comes from business administration, but by now it appears in almost every report on technology and Artificial Intelligence.
Why the AI industry is currently arguing about ROI
ROI is so popular because it makes very different things comparable. A new machine, an advertising campaign, and a software project can otherwise hardly be placed side by side. As a percentage figure, they suddenly appear in the same table. Companies use this as a basis for deciding which project gets funding and which does not.
Around Artificial Intelligence, ROI has for some time been the central point of contention. Corporations are pouring double-digit billions into data centers and chips. Revenue from AI products has so far been considerably smaller. Analysts are therefore openly asking when, or whether, these expenditures will ever pay off.
The question also plays a role on a smaller scale. If a company pays 30 euros a month per employee for an AI assistant, that assistant needs to demonstrably save time. If it doesn’t, the ROI is negative. It is precisely such pilot projects that are currently being halted in droves because their benefit could not be proven.
The formula and its pitfalls
The calculation is simple: profit divided by capital invested, times 100. Profit here means the surplus, i.e. revenue minus all costs. With an investment of 50,000 euros and a surplus of 5,000 euros, this results in an ROI of 10 percent. If the value turns negative, the investment has destroyed money.
The difficulty lies not in the formula but in plugging in the numbers. For an AI project, the investment includes not just the license fees. On top of that come training, adapting the system to one’s own data, and the working time of the employees who look after the system. Anyone who leaves out these hidden items makes the result look better than it is.
A second pitfall is time. ROI by itself does not account for a time period. 20 percent in one year is something completely different from 20 percent over ten years. Serious figures therefore always state the period alongside them. A third point: some effects are hard to express in euros, such as more satisfied customers or fewer errors. They fall out of the calculation even though they are valuable.
Where you encounter this metric in the news
In the quarterly reports of technology corporations, ROI is a recurring topic. When Microsoft, Alphabet, or Meta increase their spending on data centers, analysts immediately ask about the return. If the answer turns out to be vague, share prices often react sensitively. The phrase “no clear ROI visible yet” is considered a warning sign on the stock market.
The term also appears in studies on AI in everyday work. A widely cited MIT study in 2025 found that the vast majority of AI pilot projects in companies had no measurable financial effect. Such figures are disputed, but they shape the debate.
You should distinguish ROI from related metrics. Revenue only tells you how much money comes in, not what remains. Profit states the surplus in euros but does not relate it to the amount invested. Only ROI shows whether 100,000 euros of profit on an investment of one million is good or meager. And in everyday private life the idea works the same way: an expensive course is worth it if it later brings in more than it cost.