Developing AI could be turning into a zero-sum game with some major losers

It seems increasingly apparent that the companies developing generative AI models such as ChatGPT, Claude, Gemini and the like face a long and costly process, one subject to intense competition in a market where it will be difficult for them to differentiate themselves, and where operating costs will make it difficult to turn a profit.

The main AI companies have entered into dynamic of incremental rather than disruptive innovation whereby they must keep scaling up computing capacity, which means investing more and more in the microprocessors that Nvidia sells them, as well as in very expensive, highly specialized personnel.

For example, OpenAI is invoicing more and more as time goes by and its algorithms grow in popularity, but its costs are spiraling out of control. The outlook is not promising to say the least.

In response, the main AI players are looking to offer smaller, cheaper models, as well as trying to spread costs around the rest of the company. At the same time, they are spending money on buying smaller AI companies that can no longer afford to stay in the game.

It’s also clear that the future of AI doesn’t lie in developing ever-more sophisticated and expensive models when they have to compete with better, more transparent and…