Artificial intelligence companies are attracting enormous investments and commanding valuations that rival some of the world’s most successful businesses. Investors are betting hundreds of billions of dollars that AI will transform the economy and generate extraordinary profits.
But there’s a fundamental problem receiving surprisingly little attention: Many of these companies are selling products that are very similar to one another.
OpenAI, Google, Anthropic, Microsoft, Meta, and many others offer AI systems that answer questions, write documents, summarize information, analyze data, and generate computer code. Many are now adding similar bots to perform similar tasks. While each has particular strengths, the differences are becoming difficult to distinguish.
I’ve used ChatGPT, Claude, and Gemini extensively. All three are impressive, and one occasionally performs better than another. But for most of my everyday tasks, they’re largely interchangeable. If one disappeared tomorrow, I could accomplish nearly everything I do with one of the others.
This seems to me to be a serious problem for companies expecting to generate large profits. In business, when multiple companies offer products with essentially the same functions, competition inevitably puts pressure on prices. Customers have little incentive to pay a premium when there are comparable alternatives that cost less.
We’ve all experienced this with consumer electronics, computers, televisions, and many other industries. Products that initially commanded premium prices eventually became commodities, with manufacturers competing on price. I remember the CD player that came out at $1000 and the many models that quickly followed for $200. We once paid $2000 and more for a large screen TV and now pay a couple of hundred dollars. AI seems to be heading down the same path, only much faster.
Consider what’s already happening. Google’s Gemini is free to consumers, Microsoft incorporates AI into its products at no cost, OpenAI and Anthropic offer both free and paid services. Many times when I run beyond my limit of one I’ll switch to another to continue my work – it requires little more than going to a different website or app.
And I’m already experiencing AI overload. Gemini wants to correct my Gmail, Co-Pilot wants to correct my Word documents, and ChatGPT and Gmail both want to send me summaries of my day each morning.
This makes it difficult to establish the customer loyalty and pricing power that support high profit margins. The underlying costs of operating AI systems are enormous. In a conventional business, high operating costs are offset by the ability to charge premium prices. But what happens when those costs keep rising while competition forces prices down?
This would seem to make a case for further risk to the AI companies and to our economy that is depending on the success of these companies. The question is how many of today’s highly valued AI companies will survive when their products become commodities and customers are unwilling to pay a premium for something they can get elsewhere for free.