You're Not Selling What You Think You're Selling
Everyone's writing about the chatbot.
Last month it came out that OpenAI is now pulling in something like two billion dollars a month, on pace for twenty five billion a year, and everyone's writing about the chatbot. Nobody's talking about the part that actually matters, which is that most of that money isn't coming from the chatbot anymore. And twelve years into doing this, here's what that tells me: OpenAI didn't get rich by building a better version of the thing they set out to build. They got rich by finally noticing what people were actually hiring it to do.
Facebook is the cleanest example there is.
I talk about this exact blind spot in my book, Strategy is not Planning, using Facebook as the example, because it's the cleanest one. Facebook set out to be a way for college kids to check each other out. What it actually became valuable for was data, and Zuckerberg didn't plan that, he just wasn't dumb enough to ignore it once it showed up.
Your intended product and your valuable product are not always the same product, and pretending otherwise is how companies plateau.
The side feature became the business.
OpenAI set out to be a nonprofit research lab. That was the entire pitch. Somewhere along the way they noticed businesses weren't just curious about the chatbot, they were building entire workflows on top of the API, integrating it into products, paying real enterprise money for something OpenAI almost treated as a side feature.
So they leaned into it, and now enterprise integration and API access are quietly doing more financial heavy lifting than the consumer app most people think of when they hear the name.
People hire products to do a job.
Clayton Christensen, the Harvard professor who came up with Jobs to Be Done theory, used to say people don't buy products, they hire them to do a job. Once you understand what job your customers are actually hiring you for, sometimes it's nothing like the job you thought you were selling. OpenAI thought they were selling a chatbot. Turns out businesses were hiring them for infrastructure.
Two businesses wearing the same name.
So stop asking what your product is supposed to do, and start asking what people are actually using it for. Half the time those are two completely different businesses wearing the same name.
Sources: OpenAI revenue figures (~$2B/month, ~$25B annualized run rate), reported early to mid 2026; OpenAI Blog, "A business that scales with the value of intelligence," Jan. 2026.