I use ChatGPT, Gemini, and a bunch of other AI tools all the time as a developer, and I like them because they are useful, not because I enjoy hype.
That is why I find the “AI is dying” takes so strange. You can argue that the hype cycle was overheated. You can argue that some products were rushed, overpriced, or badly presented. You can argue that a lot of nonsense has been written about artificial general intelligence arriving next Tuesday. All of that is fair.
But “AI is a bubble and it is going away” does not match what I see in actual work.
The market is changing. That is not the same as collapsing.
One reason people get confused is that they mistake product churn for failure.
Take Sora. Headlines around OpenAI shutting it down are exactly the kind of thing people point to when they want to claim AI is collapsing.
That matters because it is the kind of thing mature platforms do all the time. Products get consolidated. Old versions get dropped. Features move into other surfaces. Plans change. None of that means the underlying capability is fake or that the category is dead.
If anything, it usually means the opposite. It means the company is trying to simplify a messy product line now that real users exist.
AI is already part of the workflow
For developers, this is the important bit.
AI tools are already in the workflow now. People use them to explore APIs, sketch code, review ideas, explain legacy code, generate tests, rewrite ugly SQL, summarize docs, and get unstuck faster. Not perfectly. Not always correctly. But often enough that it changes how people work.
That is the part some critics still seem unwilling to admit.
They talk about AI as if it is still a novelty demo, while millions of people are already using it as a daily utility. Once a tool becomes part of the default workflow, it does not need to be magical to stick around. It just needs to be useful enough, often enough.
That threshold has already been crossed.
The bigger problem is denial
I think denial is a bigger problem than the so-called bubble.
There are still people acting as if refusing to touch AI tools is a sign of seriousness or technical purity. I do not buy that. You do not have to love every model, every company, or every pitch deck. You do not have to pretend hallucinations are solved. You definitely should not hand over your brain and stop thinking.
But refusing to learn the tools because you dislike the discourse around them is not a serious strategy. It is just opting out while other people get faster.
This is the same mistake people make whenever a tool changes the shape of work. They focus on the loudest hype, decide the whole thing is fake, and miss the quieter reality that the tools are steadily getting embedded into ordinary workflows.
My take
I do not think AI is going anywhere.
I think some companies will fail, some products will disappear, some feature sets will get merged, and some of the current excitement will cool off. That is normal. That is what happens when a market grows up.
But the underlying shift is real. As a developer, I already get too much value from these tools to take the “this is all a fad” line seriously anymore.
So no, I do not think AI is a bubble in the sense people mean when they say it is over.
I think denial is the real issue now.