
Commoditization
Commoditization describes how a product turns from a distinctive offering into an interchangeable mass-market good. When many providers deliver nearly the same thing, price alone decides — in the AI industry, this now applies to many language models.
Some products are so similar that it barely matters who you buy them from. Gasoline is one such case, so is sugar, or electricity from the outlet. The buyer then looks only at the price, not the brand. Commoditization is the technical term for this development: a product loses its distinctiveness and becomes a mass-market good. The word comes from “commodity,” meaning a trade good or raw material. Importantly, the product does not get worse in the process. It simply gets produced so similarly by so many providers that no one can charge a premium anymore.
What happens when the margin disappears
For companies, commoditization is usually bad news. As long as a product is unique, high prices can be enforced. Once it becomes an interchangeable good, competitors drive the price down so far that hardly any profit remains. Economically, this profit share per unit sold is called the margin.
For customers, however, this is exactly good news. Televisions, memory cards, and solar panels have become dramatically cheaper over the years because so many companies can build them. So anyone who just wants to use a technology benefits from commoditization.
For investors, the term is therefore a warning sign. When analysts say a business is “commoditizing,” they mean: today’s high profits are not secure. Companies then try to differentiate themselves in other ways — through service, speed, data protection, or particularly convenient usability.
The path from invention to mass-market good
At the beginning, there is usually a single provider with a head start. It has invented something no one else can do and charges a lot of money for it. These high profits attract imitators. Over time, the knowledge spreads — through trade publications, through employees who switch jobs, through reverse engineering.
Then two accelerators come into play. First, standards: when all providers use the same interfaces, the customer can switch without effort. Second, open source, meaning software whose blueprint is publicly and freely available. When a usable free version exists, the price for everyone else automatically drops.
A common misconception is to confuse commoditization with cheap goods. Wheat is a commodity, but not inferior. What matters is solely interchangeability. And commoditization is not a law of nature: it can be slowed down, for example through patents, hard-to-obtain raw materials, or huge factories that almost no one can afford.
Language models as interchangeable goods
In the AI news of recent years, commoditization has been one of the main topics. Language models like ChatGPT were still a sensation in 2022. Today there are dozens of models of similar quality, many of them freely downloadable. Prices per amount of text processed have fallen by more than a hundredfold since launch.
That’s why experts argue about where money will be made in the AI industry going forward. At the model level itself, things are getting tighter. The picture looks different for the specialized chips used for AI computations: they are hard to manufacture, and only a few companies master this. There, commoditization is still far off.
Anyone reading stock market news encounters the term far beyond AI as well. With smartphones, cloud storage, or electric-car batteries, the question is constantly asked whether the product is becoming interchangeable. The answer often determines whether a company is regarded on the stock market as a growth hope or as a mere mass producer.