LLMs have burned Billions but couldn't build another Tailwind

This week brought a total saga: Tailwind laid off 75% of its workforce.

This is particularly striking because Tailwind has been around for a long time and is an extremely popular framework used by thousands (hundreds of thousands?) of websites and developers worldwide. This popularity is probably why LLMs frequently generate code using Tailwind.

Yet despite all this, Tailwind seems to be falling on hard times. It's disingenuous to claim that Tailwind is now useless or deprecated because of LLMs. First, the founder confirmed it's not. Second, it's irrelevant. Tailwind's situation involves a sudden change of course that doesn't follow the pattern of traditional decline.

It's worth noting that Tailwind is currently used by 1.5% of the web. That might sound like a small market share (and it is), but the web is massive. That's a lot of users.

The other thing that bothered me is users claiming Tailwind is bloated. No, it's not. Tailwind is actually very lean, which is probably why they were generous with their open source terms and single-purchase license. They weren't greedy. For such a small team, they produced a remarkable amount of work: high quality, thoughtfully crafted. They also created a fundamental framework that other businesses now build upon (React frameworks businesses and others). Compared to other corporate actors, they're incredibly efficient with resources.

And here's the thing: 2025 was largely the year of LLMs and agent coding. Millions of people and software developers burned billions (hundreds of billions?) of dollars worth of energy to produce tokens. We were promised much higher efficiency and output compared to the "old ways". Yet I don't see any small team with just a few developers creating something groundbreaking or sophisticated like Tailwind.

Instead, we might end up losing the one we currently have.