The recent discourse around Tailwind's future is mostly the usual "open source is dead" stuff. But there's something deeper here worth sitting with.
Tailwind did everything right by traditional dev tool standards. They built something genuinely useful. Gave the core away free. Created premium offerings with real value.
For years, that formula worked brilliantly. Tailwind was a genuine success story.
But things have changed.
Tailwind is a convenience layer. It makes CSS faster to write, easier to scan, quicker to iterate on. The real value is pure convenience. Less typing, context-switching and remembering = faster, more enjoyable coding.
It's genuinely valuable when a human is learning it, remembering it, and typing it. I've used it since v1.
But AI doesn't need convenience.
The cognitive load Tailwind eliminated is irrelevant to something that has the entire CSS spec in its training data (and infinite patience).
Any tool whose primary value is "saves you time" or "makes things easier to remember" is now competing against something that already knows everything and has infinite time.
Component libraries, boilerplate generators and tools that make the human coding experience better are pretty worthless in a world where we don't type the code anymore.
But here's where it gets interesting for anyone building software right now.
The question isn't "how do I build something useful?" anymore.
It's "how do I build something that remains valuable when mediated through AI?"
If you're a founder, this is the defensibility question you need to be asking:
Does my value come from convenience or from something AI can't replicate?
Because if an AI can generate your core offering from scratch every time a user needs it, you don't have a product. You have a prompt.
So where does defensibility actually live now?
Data that accumulates. AI can generate code, but it can't generate your users' history, state, or relationships. Products that get more valuable with usage still have a moat.
Being in the flow, not before it. Tools AI has to call still have a place. Stripe doesn't care if AI writes the integration code because the Stripe API still gets hit and they get their service fee.
Layers AI can't touch. Compliance, legal, finance, that sort of thing. We're not buying tools from regulated industries, we're buying safety.
Network effects that compound. Marketplaces. Communities. Human relationships. AI can generate content but it can't generate trust or build connections between two people.
The entire value hierarchy of software is being rewritten. There's still tonnes of opportunity to make amazing and profitable products—it's just going to look different.
If you're a founder working through these questions for your own product, I help Series A startups navigate exactly this kind of strategic technical decision-making.