Tailwind Won Because Robots Love It

Adam Wathan posted the numbers and they're deranged in the way that only things that benefited from AI slop pipelines can be deranged.

Tailwind is not popular because developers chose it. Tailwind is popular because AI writes it constantly and without hesitation, every model from every lab, in every context window, defaults to flex items-center justify-between like it's gravity.

There's a feedback loop here that nobody planned and everyone benefits from — except maybe the people who were right about CSS-in-JS, who are watching their thesis get obliterated not by a better argument but by the aesthetic preferences of a statistical averaging machine trained on GitHub. The models saw a lot of Tailwind. Now they write a lot of Tailwind. Now there's even more Tailwind on the internet. Rinse. The training data eats itself.

The interesting part isn't that Tailwind is winning. It's that "AI writes it fluently" has become a legitimate selection criterion — possibly the most important one — for any tool that wants to survive the next five years. Not "is it the right abstraction." Not "does it scale." Does the model know it cold.

Tailwind did. Everything else is scrambling.

That's not a knock on Tailwind, to be clear. The utility-class model is genuinely a good fit for how LLMs generate UI — stateless, composable, no setup, readable in a single line. You could have predicted this if you squinted at the architecture hard enough. But nobody did predict it. Adam Wathan built a tool that happened to be exactly what the machines wanted to write, and now the machines are writing it for millions of people, and the download numbers look like a hockey stick that stopped caring.