Something strange is going on in Silicon Valley right now. Everyone has an opinion on whether AI is a bubble—whether OpenAI is worth more than Toyota and Coca-Cola put together, whether Jensen Huang’s cocky attitude hides a slow deflation, and whether Sam Altman’s careful admission that some parts of AI are “bubbly” is a warning or an improvement. That’s a good question. But under all of that debate is something that has been growing slowly for longer and with less attention: the race to make money off of people’s attention.
Take a moment to think about that phrase. Making money off of attention sounds like a phrase from a business class, not a warning sign. But what it really means is a race to make money off of people’s limited screen time. This has brought valuations, product decisions, and whole company strategies into areas that, from a distance, seem to have nothing to do with reality.
The AI spending frenzy is all over the news. Before the end of the year, the world is expected to have spent close to $1.5 trillion on AI. There is now a legitimate debate going on about whether that money will be well spent, even among those who are spending it. Jeff Bezos has said that it could be a “good” bubble. Eric Schmidt helped write an article telling Silicon Valley to calm down about the idea of superintelligence. These are not the words of an industry that is sure of itself. These words are used to avoid committing to something when someone thinks they may have gone too far.
But at the same time, the attention economy has been overextending itself with a lot less self-awareness. It makes sense: if you can keep a user’s attention, you can sell that attention. People have judged the value of social networks, streaming services, short-video apps, and now AI-powered content feeds by how many minutes a day they can keep people interested. To some investors, attention is still seen as a resource that can never be used up. They think that you can always add another notification, change the algorithm, or add an infinite scroll feature, and people will keep using it.
There is more and more evidence that this is not true. In a number of major markets, the average amount of screen time has stopped growing. Users, especially younger ones, are showing real signs of saturation.
These aren’t just worries about screen time in general; they’re changes in how they act. It’s true that subscription fatigue exists. Click-through rates on ads have been steadily going down for years. The platforms that are competing for users’ attention are not getting more users, but rather spending more on things like content, infrastructure for personalization, and data centers to power recommendation engines. If you have been reading about OpenAI’s burn rate, the unit economics there will look all too familiar.
It’s still not clear if the companies that are deeply involved in this race have a real way to get out of it. Adding AI to attention-economy products like smarter feeds, generated content, and AI companions meant to keep users on the site longer may make the problem worse instead of better. If the basic idea is that more automation means more engagement, which means more money, then that line of thinking must eventually run into the problem of how much time people actually have.
This seems different from the conversation about the AI bubble because Silicon Valley is at least having that conversation. Smart, well-known people have spoken out against it. The attention economy is looked at much less critically. This could be because it has been making real money for a long time, making the structural flaws beneath it easier to ignore. In the end, the railroads proved their worth. But attention doesn’t build on itself like freight rail does. A day only has so many hours. At some point, the rush to fill them starts to look less like an industry and more like a bet that people’s patience is infinite, which it most definitely isn’t—as anyone who has recently closed an app in frustration can tell you.
