
Commoditization
Commoditization describes how a once-special product turns into interchangeable mass merchandise because many providers deliver almost the same thing. Price then becomes the most important distinguishing factor – in the AI industry, this currently affects language models and computing power.
Some products are special at first. Only one company can produce them, and it charges a lot for that. After some time, other firms can do the same, often almost as well. Then prices fall, and customers no longer care whom they buy from. This exact process is called commoditization: a special offering turns into interchangeable mass merchandise. The English word “commodity” roughly means “standard goods,” something like wheat, gasoline, or sugar, which is the same everywhere.
Why profit margins disappear in the process
As long as a provider dominates something alone, it can largely set the price itself. It earns a great deal per unit sold. As soon as ten competitors offer the same thing, that no longer works. Customers compare, and the cheapest one wins. The profit margin, meaning the gap between production costs and sales price, shrinks.
For investors, this is one of the most important questions there is. A company with a unique product is worth significantly more on the stock market than a company selling interchangeable goods. That’s why share prices react sharply when a competitor suddenly offers the same thing more cheaply. Revenue doesn’t even need to collapse immediately for this to happen. The mere expectation that high prices can’t be maintained is enough.
An important distinction is needed here: commoditization does not mean a product gets worse. Usually it even gets better and more reliable. It just loses its special status. Electricity is a good example: extremely useful, technically demanding – and yet nobody buys it because of the brand.
The path from special case to standard goods
At the start, there is usually a technical edge. A company knows or can do something others cannot. This edge rarely lasts forever, though. Experts change companies, methods get published, and copying is almost always cheaper than inventing. Bit by bit, competitors catch up.
A second driver is standards. When everyone agrees on the same interface, meaning the same way programs talk to each other, switching providers becomes easy. A company can then swap out its AI service provider without having to rewrite its own software. This very interchangeability is the core of commoditization.
The third driver is falling costs. Larger volumes, better manufacturing, and open alternatives push down the price. Open-source programs, whose blueprint anyone may use for free, are especially effective in this regard. If a free variant delivers 90 percent of the performance, it becomes hard to charge a high price for the remaining 10 percent.
Language models as an example from the AI industry
In 2022, a powerful language model, meaning a program that understands and writes text, was considered a rare cutting-edge technology. Only a few companies possessed such a thing. Today there are dozens of them, many freely available. The price for a single query has fallen by more than a hundredfold in just a few years. Providers now openly advertise being cheaper than the competition.
That’s why the term keeps popping up in business news. Analysts ask whether AI models are becoming interchangeable commodities and where the money will then be made. In response, companies often point to things that are hard to copy: proprietary data, established customer relationships, or chips that only a few factories can manufacture.
A common misconception is that commoditization is bad for everyone. For providers it’s unpleasant, but for users it’s mostly a blessing. Falling prices are precisely what brought AI features into everyday apps, educational software, and search engines in the first place. Historically, this is nothing special: hard drives, mobile networks, and cloud storage all took the same path.